Medical Device Epidemiology and Surveillance

Medical Device Epidemiology and Surveillance Editors S. Lori Brown, PhD, MPH Roselie A. Bright, ScD and Dale R. Tavr...

Author: S. Lori Brown | Roselie A. Bright | Dale R. Tavris

251 downloads 3104 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Medical Device Epidemiology and Surveillance

Editors

S. Lori Brown, PhD, MPH Roselie A. Bright, ScD and

Dale R. Tavris, MD, MPH Division of Postmarket Surveillance, Center for Devices and Radiological Health, Rockville, MD, USA

Medical Device Epidemiology and Surveillance

Medical Device Epidemiology and Surveillance

Editors

S. Lori Brown, PhD, MPH Roselie A. Bright, ScD and

Dale R. Tavris, MD, MPH Division of Postmarket Surveillance, Center for Devices and Radiological Health, Rockville, MD, USA

Copyright # 2007

John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England Telephone

(+44) 1243 779777

The following chapters were written by United States Federal Government employees: 1, 2, 3, 4, 5, 6, 7, 8, 9, 15, 16, 17, 18, 19, 20, 21, 22, 24, 26, 27, 28, 29, 31. Email (for orders and customer service enquiries): [email protected] Visit our Home Page on www.wiley.com All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed to the Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England, or emailed to [email protected], or faxed to (+44) 1243 770620. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The Publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the Publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. Other Wiley Editorial Offices John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany John Wiley & Sons Australia Ltd, 42 McDougall Street, Milton, Queensland 4064, Australia John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop # 02-01, Jin Xing Distripark, Singapore 129809 John Wiley & Sons Canada Ltd, 6045 Freemont Blvd, Mississauga, ONT, L5R 4J3, Canada Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Anniversary Logo Design: Richard J. Pacifico Library of Congress Cataloging-in-Publication Data Medical device epidemiology and surveillance / edited by S. Lori Brown, Roselie A. Bright, and Dale R. Tavris. p. ; cm. Includes bibliographical references and index. ISBN-13: 978-0-470-01595-7 (cloth : alk. paper) 1. Medical instruments and apparatus–Standards–United States. 2. Medical instruments and apparatus–Safety measures–United States. I. Brown, S. Lori. II. Bright, Roselie A. III. Tavris, Dale R. [DNLM: 1. Equipment Safety–standards. 2. Equipment Failure Analysis–methods. 3. Product Surveillance, Postmarketing–methods. 4. Safety Management–standards. W 26 M48903 2007] R856.6.M42 2007 610.28–dc22

2006038741

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN-13 978-0-470-01595-7 Typeset in 10.5/12.5 pt. Times by Thomson Digital Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire This book is printed on acid-free paper responsibly manufactured from sustainable forestry in which at least two trees are planted for each one used for paper production. Cover image is reproduced from the Science Museum/Science & Society Picture Library

Contents Foreword

xiii

Preface

xvii

Contributors

xix

Acknowledgments

xxv

1

Introduction Thomas P. Gross

1

2

Medical device regulation in the USA Thomas P. Gross, Celia M. Witten, and Casper Uldriks

5

3

4

Introduction Premarket review Marketing applications Postmarket oversight Conclusion

5 6 8 10 18

Medical device epidemiology Roselie A. Bright and S. Lori Brown

21

Introduction Features of medical devices that are relevant to epidemiology study design Study designs for medical device epidemiology Summary and recommendations

21 26 35 37

Surveillance of adverse medical device events Roselie A. Bright

43

Introduction Rationale for surveillance Surveillance based on adverse event reports Surveillance based on registries Active surveillance Necessary conditions for effective surveillance Ideal AMDE surveillance program Summary

43 44 47 52 53 54 55 58

vi

5

6

CONTENTS

The Medical Product Surveillance Network (MedSun) Roselie A. Bright, Marilyn N. Flack, and Susan N. Gardner

63

Historical motivation Initial considerations for the design of DeviceNet MedSun basic design Current status Is MedSun successful in promoting the safe use of medical devices? Epidemiologic considerations Summary

63 64 69 71 74 75 76

The National Electronic Injury Surveillance System (NEISS) and medical devices Brockton J. Hefflin, Thomas P. Gross, and Thomas J. Schroeder

79

Description and history of NEISS Potential uses and limitations of NEISS Utilization of NEISS to produce national medical device-associated adverse event estimates Potential for long-term utilization of NEISS for medical device surveillance

7

8

81 83

Medical device nomenclature Brockton J. Hefflin, Thomas P. Gross, Elizabeth A. Richardson, and Vivian H. Coates

87

Technical elements Current terminologies Applications of nomenclature Future developments

87 91 93 95

Data sources for medical device epidemiology studies and data mining Danica Marinac-Dabic, Baoguang Wang, Brockton J. Hefflin, Hesha J. Duggirala, Tripthi M. Mathew, and Cara J. Krulewitch Introduction Data sources Surveillance databases Registries Automated large administrative databases National surveys Data mining Future use databases for medical device epidemiology

9

79 80

Ethical requirements and guidelines for epidemiological studies of medical devices Danica Marinac-Dabic and Suzanne C. Fitzpatrick Introduction Bioethics foundations

99

99 100 101 105 112 115 119 120

127 127 128

CONTENTS

US government human subjects protection regulations HHS human subjects protection regulatory requirements FDA human subjects protection regulations Other professional ethical guidelines Ethical requirements for medical devices epidemiologic studies Future ethical requirements for epidemiologic studies

10 An industry perspective: medical device epidemiology and surveillance Martha A. Feldman Introduction The product’s life cycle: premarket (preclinical and clinical) and postmarket (PM) studies Postmarket studies Summary

11 Perspective from an academic on postmarket surveillance Lazar J. Greenfield Introduction Adverse event reporting Postmarket condition of approval studies and Section 522 studies Industry use of information from adverse event reports Academic opportunities Summary

12 Perspective from a pharmacoepidemiologist Thomas K. Hazlet Introduction Review of literature Contrasts with pharmacopeidemiology Conclusions

13 Medical device regulation and surveillance: perspective from the EU Lennart Philipson Introduction Medical devices: the European directives and definitions New approach Classification of medical devices in Europe Market surveillance Traceability of medical devices in Europe The vigilance system

14 A consumer advocate’s perspective on medical device epidemiology and surveillance Diana M. Zuckerman Examples of widely publicized problems with selected medical devices Consumer concerns

vii 128 130 131 133 136 142

145 145 148 153 156

159 159 160 163 165 167 168

171 171 171 173 175

177 177 178 179 182 183 184 184

187 189 193

viii

CONTENTS

Regulatory mechanism recommendations Consumer group accomplishments: mixed results

15 Pediatric Medical Device Use Judith U. Cope and Thomas P. Gross

197 199

203

Introduction Medical device use in children Special device risks and safety concerns for children Regulatory framework The future for pediatric medical device surveillance and epidemiology

203 204 211 212 214

16 Selected medical devices used to manage diabetes mellitus Shewit Bezabeh, Joy H. Samuels-Reid, and Dale R. Tavris

219

Introduction Insulin delivery devices Continuous glucose monitoring devices Future medical devices for the management of diabetes

219 221 225 228

17 Medical device-related outbreaks S. Lori Brown, Hesha J. Duggirala, and Dale R. Tavris

237

Introduction Endoscopes: bronchoscopes and gastrointestinal (GI) endoscopes Hemodialysis-related outbreaks Neonatal and pediatric intensive care units Miscellaneous device-related outbreaks Summary

237 239 244 247 250 251

18 Risk of transmission of prions with medical devices S. Lori Brown and Azadeh Shoaibi Iatrogenic transmission of prion disease via neurological or surgical instruments Surgical instruments used to transplant tissues known to have transmitted prion disease (corneal transplant) Surgical instruments in contact with lower risk peripheral tissues Decontaminating surgical instruments to reduce risk of TSE transmission FDA measures to minimize risk of transmission of BSE by medical products Summary

19 Surveillance and epidemiology as tools for evaluating the materials used in medical devices Roselie A. Bright Introduction General considerations that affect the design of epidemiologic studies of materials used in medical devices Example: natural rubber and latex allergy Summary

259 260 261 262 263 267 268

273 273 273 279 286

CONTENTS

20 Exploring methods for analyzing surveillance reports on electromagnetic interference with medical devices S. Lori Brown, Nilsa Loyo-Berrı´os, Miche`le G. Bonhomme, Donald M. Witters, Nancy A. Pressly, and Jeffrey L. Silberberg Electromagnetic interference with medical devices The MAUDE database Adverse event reports to FDA (December 1984–October 1995): EMI with cardiovascular devices Adverse events reported to FDA (January 1994–March 2005): a recent epidemiological analysis Discussion

21 Alternative and complementary medical devices S. Lori Brown and Joannie C. Shen

ix

291

292 292 293 297 311

319

Introduction Acupuncture needles Ear candles Magnets Adverse events associated with alternative or complementary devices

319 322 326 328 331

22 Drug-eluting coronary stents Hesha J. Duggirala, Thomas P. Gross, and David E. Kandzari

335

History and evolution of coronary stents: clinical perspective Novel drug-eluting stent programs Comparative trials of drug-eluting stents Application of drug-eluting stents in complex coronary lesion subsets Drug-eluting stents: regulatory perspective

335 340 342 345 346

23 The treatment of abdominal aortic aneurysms Dale R. Tavris and Louis M. Messina Natural history and indications for treatment FDA experience with the postmarket assessment of an endovascular graft Advances in stent-graft design and performance Conclusion

24 Cardiovascular devices: aortic valves Ronald G. Kaczmarek and Chih-Hsin K. Liu Introduction Operative mortality The FDA guidance for replacement valves and the objective performance criteria (OPC) Adverse event reports The need for epidemiological studies Conclusions

25 Hemostasis Devices Dale R. Tavris, Beverly Gallauresi, and Ralph G. Brindis The ideal vascular hemostasis device Origin of FDA concern with the postmarket performance of hemostasis devices

355 355 358 361 363

367 367 370 372 373 373 376

379 380 382

x

CONTENTS

FDA study to evaluate the risk associated with hemostasis device use Possible reasons for FDA findings of apparent protective effects of hemostasis devices Follow-up FDA study to assess the safety of hemostasis devices Conclusion

26 ENT devices: cochlear implants James K. Kane and Eric A. Mann Cochlear implant description Epidemiological investigations involving meningitis associated with cochlear implants Conclusions

27 Silicone gel-filled breast implants: surveillance and epidemiology S. Lori Brown and Joan Ferlo Todd Breast implant types A brief regulatory history of breast implants FDA surveillance studies on breast implants Summary

28 Ophthalmic devices and clinical epidemiology Malvina B. Eydelman, Gene Hilmantel, James Saviola, and Don Calogero Introduction Epidemiological contributions to IOL evaluation Epidemiology of contact lens ulcers and public policy Conclusion

29 Orthopedic devices: epidemiologic considerations Ronald G. Kaczmarek, Michele G. Bonhomme, Stanley A. Brown, Judith U. Cope, and Daniel S. McGunagle Introduction Selected orthopedic materials Artificial hips Intervertebral disc replacement Data sources and selected methodological issues to consider in epidemiologic studies of orthopedic medical devices Summary

30 Clinical epidemiology of intrapartum fetal monitoring devices Danica Marinac-Dabic, Barry S. Schifrin, Cara J. Krulewitch, and Roscoe M. Moore Introduction Electronic fetal monitoring Scalp and umbilical cord blood gases analyses Fetal pulse oximetry Fetal ECG waveform monitoring Challenges of intrapartum fetal monitoring modalities Conclusion

387 389 389 390

395 395 396 405

407 407 409 411 421

427

427 427 430 437

441

441 441 444 452 456 458

463

463 464 470 471 473 476 478

CONTENTS

31 The postmarket surveillance of medical devices: meeting the challenge Susan N. Gardner and Daniel Schultz Index

xi

483

487

Foreword The problem of patient safety first leapt into the public conscience in 1999 with the publication of the Institute of Medicine’s landmark ‘‘To Err Is Human’’ report [1]. The report summarized a variety of studies which had been done in hospitalized patients and suggested that large numbers of patients were dying annually in the US as the result of care they were receiving, and many more patients were being injured. Two of the leading causes of injuries were surgical care and medications. Since that time, medication safety has received a great deal of attention, and much more is known today about the epidemiology of medication-safety related issues, and how to prevent adverse drug events than in 1999. Similarly, monitoring of medication safety after drugs are released is relatively advanced. Surgical injuries are more diverse, and may be harder to address; in any event, they have received less attention. Medical devices represent another extremely important type of medical intervention. While they are clearly beneficial in the aggregate, they also carry important risks. They represent an essential part of surgical and interventional care in particular, although they are used in all health care, and their importance is growing. The market capitalization for the device field was recently estimated to be $75 billion [2]. While this is not nearly as big as for drugs, it is still very large. Further complicating things, there appear to be approximately 8000 different companies in the medical device field and over 80% of these have fewer than 50 employees. These smaller enterprises in particular may find monitoring the safety of their devices challenging, especially as this monitoring competes for scarce resources with other parts of the company. Monitoring of device safety is primarily the responsibility of the Food and Drug Administration, which works closely with industry to ensure public safety. Perhaps not surprisingly, many of the leading experts in the field of medical device epidemiology and surveillance work at the FDA. Thus, this authoritative book, edited by Drs Brown, Bright and Tavris from the FDA, is especially welcome. Many of the chapters are written by authors from the FDA, but there are also contributors from industry, academia, a consumer group, a consulting group, and a foreign government. The book attempts to describe the issue of device safety, and to begin to develop a theoretical framework for the study of problems with devices, much as pharmacoepidemiology did for pharmaceuticals. Such a framework is badly needed, since many of the issues with devices are different from thosewith medications. For example, many devices are permanently inserted inside patients, and may have long-term consequences. Many other devices are not inserted into

xiv

FOREWORD

patients at all. Others require operation by people, making human factors issues especially prominent. Many devices are life-saving, yet poor performance can be catastrophic. Implantable defibrillators are the poster child for this class of device. In the right patient population, they substantially improve lifespan, yet are extremely expensive, and sometimes fail. Both failure to deliver a shock when needed and delivery of excessive shocks are potential disasters. Intravenous pumps represent another important class of devices. Intravenous medications are especially high-risk compared to oral drugs, and if for example ten times too much medication is delivered, the consequences can be life-threatening, especially for some medications such as heparin or insulin. Yet errors like this can occur with a single extra key-press. Another critically important class of devices are those used in orthopedics. Many are inserted and then subjected to mechanical stresses over a period of many years. The complete natural history of these devices may not be known at the time they are inserted, which makes building the ability to track their performance and failure rates over extremely long periods very valuable. Tracking the performance and failure rates of devices has thus long represented an important challenge. A particular issue is that it has been relatively hard to identify adverse medical device events (or AMDEs). The principal mechanism for finding these events (as for adverse drug events) has been to rely on spontaneous reporting. This results in substantial underestimates of the true rates; a study from the University of Utah found 83.7 AMDEs per 1000 admissions using the combination of computerized flags, discharge codes, and spontaneous reporting, a rate which was 52 times as high as that identified through spontaneous reporting alone [3]! This landmark study clearly demonstrates that more work needs to be done to more efficiently identify these events. If these figures are accurate, there are probably over a million AMDEs annually in US hospitals. Drs Bright and Brown discuss these issues in much more detail in Chapter 3 [4]. Another key issue in this area is that increasingly it is possible for implantable devices to set them up in ways that allow them to communicate back about how they are performing and whether problems have occurred. As electronic records become increasingly ubiquitous, it may be possible to do a great deal of device monitoring by having the devices send back information to the electronic record. This will demand addressing many issues around standards, confidentiality, privacy, and ethics, but this has the promise to revolutionize safety monitoring, and the potential benefits are enormous. Devices that are not implanted can also frequently communicate now wirelessly with networks, which may have important safety benefits. All of these and many other areas are covered in this book, which represents the most important effort to date to describe the myriad issues around medical device surveillance, including many of the practical problems involved, and the complex epidemiological issues associated with assessing device safety. Clearly, devices will play an increasingly important and beneficial role in healthcare, but this will at the same time demand major advances in measuring and monitoring their safety. The obvious implication is that more research is needed in this area, which should be federally supported, as it represents a

FOREWORD

xv

public good. Achieving the potential benefits of advances in the device field will require the development and growth of the field of device safety which is just in its infancy, as well as close collaboration between the private and public sectors. David W. Bates, MD, MSc Division of General Internal Medicine, and Department of Medicine, Brigham and Women’s Hospital, Partners Healthcare, Harvard Medical School, and Harvard School of Public Health, Boston, MA, 02115

References 1. Institute of Medicine. To Err is Human: Building a Safer Health System. Washington, D.C.: National Academy Press, 1999. 2. U.S. Department of Commerce. FY2004 industry assessment: medical equipment. 2004. 3. Samore MH, Evans RS, Lassen A, Gould P, Lloyd J, Gardner RM et al. Surveillance of medical device-related hazards and adverse events in hospitalized patients. JAMA 2004; 291(3):325–334. 4. Bright RA, Brown SL. Medical device epidemiology. In: Brown, Bright, Tavris, editors. Medical Device Epidemiology and Surveillance. London: John Wiley & Sons, 2006.

Preface Medical devices are instruments, apparatuses, implements, machines, contrivances, implants, or in vitro diagnostics and are intended for the cure, mitigation, treatment, diagnosis, or prevention of disease. They are intended to affect the structure or function of the body through means other than chemical action and independently of being metabolized. In the past few decades there has been greater recognition of how critical they are to delivering state-of-the-art medical care, as their use in the delivery of medical care has always been ubiquitous. In some cases, medical devices will provide the principal therapy for patients. Medical devices are more than scalpels and gravity-fed infusion bags and dental drills. Devices such as contact lenses, insulin pumps, amalgams for filling caries, and tympanostomy tubes for otitis media are so commonplace that we don’t give them much thought. But it is important to consider such issues as whether long-term wear of contact lenses may be associated with corneal ulcers; or whether patients with insulin pumps fare better over the long term in keeping their disease under control and reducing or delaying the sequelae of diabetes; or whether mercury-containing dental amalgams cause or contribute to poorly defined syndromes; or whether tympanostomy tubes mitigate a delay in speech and language acquisition in children with recurrent otitis media. In many cases, ascertaining the answers to questions such as these, and developing a thorough understanding of the effects of medical devices on exposed populations, may only be possible with epidemiologic studies. With the dizzying pace of new technology in medical devices, surveillance for adverse events and a systematic approach to their epidemiology are critical for detecting and reacting to unforeseen problems with medical devices in a timely manner. With this book, we hope to stimulate interest in medical device epidemiology and surveillance as an important sub-discipline in epidemiology – like pharmacoepidemiology. It is for these reasons that we have undertaken the challenge of putting together a book on medical device epidemiology and surveillance. We see this discipline as one rich with possibility for medical providers looking for the best available treatment for their patients, or medical device manufacturers looking to improve their device for an edge over their competitors, or academics looking for a dissertation project for a student. This book has chapters devoted to describing broad themes in medical device epidemiology and surveillance, as well as chapters that describe epidemiological issues related to

xviii

PREFACE

specific medical devices. Understanding medical device utilization and its consequences (good and bad) is crucial to improving medical device-related patient care and to advancing patient safety. S. Lori Brown, PhD, MPH Roselie A. Bright, ScD Dale R. Tavris, MD, MPH

Contributors Shewit Bezabeh, MD, MPH Medical Officer, Division of Anesthesiology, General Hospital, Infection Control, and Dental Devices, Office of Device Evaluation, Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd, HFZ-480, Rockville, MD 20850, USA. Email: [email protected] Michele G. Bonhomme, PhD Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA Roselie A. Bright, ScD Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Ralph G. Brindis, MD, MPH, FACC Regional Senior Advisor for Cardiovascular Diseases, Northern California Kaiser Permanente, 280 West MacArthur Boulevard, Oakland, CA 94611, USA. Email: [email protected] S. Lori Brown, PhD, MPH Epidemiologist, Investigations Branch, Seattle District Office, Food and Drug Administration, 22201 23rd Drive SE, Bothell, WA 98021, USA. Email: [email protected] Stanley A. Brown, PhD Research Biomedical Engineer, Division of Solid and Fluid Mechanics, Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, Food and Drug Administration, 12720 Twinbrook Parkway, HFZ-150, Rockville, MD 20857, USA. Email: [email protected] Don Calogero, MS Division of Ophthalmic and ENT Devices, Office of Device Evaluation, Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd, HFZ-460, Rockville, MD 20850, USA. Email: [email protected] Vivian A. Coates, PhD ECRI, 5200 Butler Pike, Plymouth Meeting, PA 19462, USA. Email: [email protected] Judith U. Cope, MD, MPH Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected]

xx

CONTRIBUTORS

Hesha J. Duggirala, PhD, MPH Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Malvina B. Eydelman, MD Senior Medical Advisor, Division of Ophthalmic and ENT Devices, Office of Device Evaluation Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd, HFZ-460, Rockville, MD 20850, USA. Email: [email protected] Martha A. Feldman, RAC President and CEO, Drug & Device Development Co, Inc., PO Box 3515, Redmond, WA 98073-3515, USA. Email: [email protected] Suzanne C. Fitzpatrick, PhD, DABT Senior Science Policy Analyst, Office of the Commissioner, Food and Drug Administration, Parklawn Bldg, HF-32, 5600 Fishers Lane, Rockville, MD 20850, USA. Email: [email protected] Marilyn N. Flack Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-500, Rockville, MD 20850, USA. Email: [email protected] Beverly Gallauresi, RN Nurse Analyst, Product Evaluation Branch I, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-520, USA. Email: [email protected] Susan N. Gardner, PhD Director, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-500, Rockville, MD 20850, USA. Email: [email protected] Lazar J. Greenfield, MD Professor of Surgery and Chair Emeritus, University of Michigan, 2101 TC/Box 0346, Ann Arbor, MI 48109-0346, USA. Email: [email protected] Thomas P. Gross, MD, MPH Director, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-520, Rockville, MD 20850, USA. Email: [email protected] Thomas K. Hazlet, PharmD, DrPH Associate Professor, Pharmaceutical Outcomes Research and Policy Program, University of Washington, Department of Pharmacy, Room H-375, Health Sciences Center Box 357630, Seattle, WA 98195-7630, USA. Email: [email protected] Brockton J. Hefflin, MD, PhD, DPS Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Gene Hilmantel, OD, MS Optometrist/Statistician, Division of Ophthalmic and ENT Devices, Office of Device Evaluation, Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd., HFZ-460, Rockville, MD 20850, USA. Email: [email protected]

CONTRIBUTORS

xxi

Ronald G. Kaczmarek, MD, MPH Medical Officer, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] David E. Kandzari, MD Assistant Professor of Medicine, Department of Medicine, Duke Clinical Research Institute, 2400 Pratt Street, Durham, NC 27705, USA. Email: [email protected] James K. Kane Reviewer, Ear, Nose, and Throat Devices Branch, Division of Ophthalmic and ENT Devices, Office of Device Evaluation, Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd, HFZ-460, Rockville, MD 20850, USA. Email: [email protected] Cara J. Krulewitch, CNM PhD University of Maryland School of Nursing, Department of Child, Women’s and Family Health, 655 W. Lombard St. Suite 575A, Baltimore, MD 21201, USA. Email: [email protected] Chih-Hsin K. Liu, RN Nurse Analyst, Product Evaluation Branch I, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-520, Rockville, MD 20850, USA. Email: [email protected] Nilsa Loyo-Berrı´os, PhD, MS Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Eric A. Mann, MD Branch Chief, Ear, Nose, and Throat Devices Branch, Division of Ophthalmic and ENT Devices, Office of Device Evaluation, Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd, HFZ-460, Rockville, MD 20850, USA. Email: [email protected] Danica Marinac-Dabic, MD, PhD, MMSc Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Tripthi M. Mathew, MD, MPH Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Daniel S. McGunagle Chemist, Product Evaluation Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-520, Rockville, MD 20850, USA. Email: [email protected] Louis M. Messina, MD Professor and Chief, Division of Vascular Surgery, University of California, San Francisco, E.J. Wylie Endowed Chair in Surgery, Vice Chair Department of Surgery, Director UCSF Heart and Vascular Center, 400 Parnassus Avenue A-581, San Francisco, CA 94143-0222, USA. Email: [email protected]

xxii

CONTRIBUTORS

Roscoe M. Moore, Jr, DVM, PhD, DSc Rear Admiral, US Public Health Service (Retired), 14315 Arctic Ave, Rockville, MD 20853, USA. E-mail: [email protected] Lennart Philipson, PhD Associate Professor, Director, Medical Products Agency, Medical Devices, PO Box 26, SE-751 03 Uppsala, Sweden. Email: [email protected] Nancy A. Pressly, BSEng Policy Analyst, Issues Management Staff, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-510, Rockville, MD 20850, USA. Email: [email protected] Elizabeth A. Richardson, MSN, MPH, RN PA 19462, USA. Email: [email protected]

ECRI, 5200 Butler Pike, Plymouth Meeting,

Joy H. Samuels-Reid, MD, FAAP Medical Officer, Division of Anesthesiology, General Hospital, Infection Control, and Dental Devices, Office of Device Evaluation, 9200 Corporate Blvd, HFZ-480, Rockville, MD 20850, USA. Email: [email protected] James Saviola, OD Senior Supervisory Regulatory Officer, Division of Ophthalmic and ENT Devices, Office of Device Evaluation, Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd., HFZ-460, Rockville, MD 20850, USA. Email: [email protected] Barry S. Schifrin, MD 6345 Balboa Blvd, Bldg II Suite 245, Encino, CA 91316, USA. E-mail: [email protected] Thomas J. Schroeder, MS US Consumer Product Safety Commission, 4330 East West Highway, Bethesda, MD 20814, USA. Email: [email protected] Daniel Schultz, MD Director, Center for Devices and Radiological Health, Food and Drug Administration, 9200 Corporate Blvd, HFZ-1, Rockville, MD 20850, USA. Email: [email protected] Joannie C. Shen, MD, MPH, PhD Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Azadeh Shoaibi, MS, MHS Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Jeffrey L. Silberberg, MSEE Electronics Engineer, Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, Food and Drug Administration, Twinbrook Parkway, HFZ-160, Rockville, MD 20850, USA. Email: jeffrey. [email protected] Dale R. Tavris, MD, MPH Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected]

CONTRIBUTORS

xxiii

Joan Ferlo Todd, RN Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-520, Rockville, MD 20850, USA. Email: [email protected] Casper Uldriks Consumer Safety Officer, Office of Compliance, Center for Devices and Radiological Health, Food and Drug Administration, OAK4 Building, HFZ-300, 2094 Gaither Road, Rockville, MD 20850, USA. Email: [email protected] Baoguang Wang, MD, DrPH Epidemiologist, Epidemiology Branch, Division of Postmarket Surveillance, Office of Surveillance and Biometrics, Center for Devices and Radiological Health, Food and Drug Administration, 1350 Piccard Drive, HFZ-541, Rockville, MD 20850, USA. Email: [email protected] Celia M. Witten, MD, PhD Director, Office of Cellular Tissues and Gene Therapy, Center for Biologics Evaluation and Research, Food and Drug Administration, WOC-1 Building, HFM-700, 1401 Rockville Pike, Rockville, MD 20852, USA. Email: Celia.witten@fda. hhs.gov Donald M. Witters, Jr, MS Regulatory Review Scientist, Office of Science and Engineering Laboratories, Center for Devices and Radiological Health, Twinbrook Parkway, HFZ-130, Rockville, MD 20852, USA. Email: [email protected] Diana Zuckerman, PhD President, National Research Center for Women and Families, 1901 Pennsylvania Ave NW, Suite 901, Washington, DC 20006, USA. Email: dz@center4 research.org

Acknowledgments We owe a debt of gratitude to the many authors who contributed to this book. Each of the manuscripts for this book went through a review process. The reviewers are so important because their discussions with the authors helped to clarify and strengthen the chapters. The reviewers were from both inside and outside the Food and Drug Administration. FDA reviewers include: Samie N. Allen, Karen Baker, R.N., Helen J. Barr, M.D., Patricia A. Bernhardt, Ashley B. Boam, M.S.B.E., Barbara D. Buch, M.D., Robert J. Cangelosi, P.E., Paul L. Chandeysson, M.D., Sahar M. Dawisha, M.D., Patricia C. Delaney, Charles Durfor, Ph.D., Patricia A. Fox, Phil Frappaolo, John W. Gardner, M.D., Dr.P.H., Steven I. Gutman, M.D., M.B.A., Lorraine M. Harkavy, R.N., M.S., Joseph C. Hutter, Ph.D., John J. Langone, Ph.D., Larry G. Kessler, Sc.D., Steven B. Kurtzman M.D., Sheila A. Murphey, M.D., Manette T. Niu, M.D., William F. Regnault, Ph.D., Kimber C. Richter, M.D., Harvey Rudolph, Harry R. Sauberman, Joy Samuels-Reid, M.D., Wolf Sapirstein, M.D., Paul Seligman, M.D., M.P.H., David S. Shindell, Ph.D., Judy A. Staffa, Ph.D., R.Ph., Emil P. Wang, Mary Ann Wollerton, Norman Wong, Aron S. Yustein, M.D. Reviewers from outside FDA included: Chunliu Zhan, M.D., Ph.D. (Agency for Healthcare Research and Quality, Rockville, Maryland), Thomas K. Hazlet, Pharm.D., Dr. P.H., (Pharmaceutical Outcomes Research and Policy Program, University of Washington, Seattle, Washington); Martha A. Feldman, RAC, President and CEO, (Drug & Device Development Co., Inc., Redmond, Washington). I (S.L.B.) would also like to extend my thanks to the FDA Seattle District Director, Charles M. Breen; the Director of Investigations Branch, Celeste M. Corcoran; and my supervisor, Janelle K. Martin, for their patience and encouragement while I worked on this book – while trying to balance my workload in the district with my responsibilities as an author and editor. And special thanks to Dr Larry G. Kessler, the Director of the Office of Science and Engineering at the Center for Devices and Radiological Health, for sharing tips from a seasoned senior manager’s perspective on how to finish this book.

1 Introduction Thomas P. Gross US Food and Drug Administration, Rockville, MD, USA

Medical Device Epidemiology and Surveillance, as the name implies, is a book about the discipline of epidemiology as applied to medical devices. Epidemiology, known as the basic science of public health, aims to study the distribution and determinants of diseases in populations [1]. Epidemiologic methods have been successfully applied in a wide variety of public health arenas, including those involving both acute (infection, injury) and chronic (cancer, atherosclerosis) conditions. Surveillance, a key practice within epidemiology, involves not only the systematic collection, analysis, and interpretation of health data essential to public health, but also its dissemination and application [2]. Thus, for example, when applied to medical devices, epidemiology may describe patterns of use or factors associated with use or characterize the risk for certain outcomes in defined subgroups. Likewise, surveillance may identify unanticipated adverse events, related to medical device exposure, which may ultimately lead to product removal. Although both epidemiology and surveillance are essential to assuring the safe and effective use of marketed medical devices, as amply demonstrated throughout this book, their optimal application has yet to be fully realized. The ‘gap’ between current practice and future potential is magnified in importance when one considers the burgeoning field of medical device development and use, and the millions exposed daily to these products. Medical devices, in short, are instruments, apparatus, implements, machines, contrivances, or in vitro reagents intended to diagnose, cure, mitigate, treat, or prevent disease, or intended to affect the structure or function of the body through means other than chemical action [3]. Given this definition, medical devices encompass a wide variety of products, from single-use disposable to short- and long-term implantable to multiple-use durable capital equipment, and are utilized throughout medical and surgical practice. Their very nature presents a degree of complexity not seen with other regulated Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

2

CH1

INTRODUCTION

consumer products, such as drugs. Depending upon the device, a variety of factors, such as those related to biocompatibility, hardware or software design, or maintenance, may need to be taken into account in evaluating its initial and ongoing safety and effectiveness. In addition, accounting for both the user and the patient (at times the same) and venues for use (e.g. home vs. hospital care) are critical to a complete assessment of safe and effective use. Although testing and evaluation of medical devices prior to marketing may be extensive, assurance of their continued safety and/or effectiveness is heavily dependent upon robust postmarket oversight, most importantly provided through the practice of epidemiology and surveillance. Their significant role assumes greater importance when considering well-known limitations of premarket testing and clinical trials (even those randomized) and the iterative nature of device development. Epidemiology, through the conduct of postmarket observational studies (significant examples of which are cited throughout this book), may assess several important device-related issues, including: (a) patterns of use (both on- and off-label); (b) long-term performance (including effects of re-treatment and product change); (c) ‘real-world’ performance (including effectiveness of training); (d) subgroup risk characterization; (e) general public health burden of adverse events; and (f) specific safety signals of significant potential public health impact. The latter are initially identified via device-related adverse event passive reporting systems, both US-based and international. These surveillance systems are key components for continuous postmarket monitoring and assessment of medical device safety. Noting the importance of epidemiology and surveillance for assuring postmarket device safety and effectiveness, the US Congress called for a study to be conducted by the Institute of Medicine (IOM) to assess whether ‘the system under the Federal Food, Drug, and Cosmetic Act for the postmarket surveillance of medical devices provides adequate safeguards regarding the use of devices in pediatric populations’ [4]. ‘Postmarket surveillance’ was broadly defined to include observational studies. Although the focus of the study was pediatric, the IOM’s report, Safe Medical Devices for Children, made recommendations for improvements to ‘postmarket surveillance’ conduct and infrastructure generally applicable to all age groups and devices [5]. Among infrastructure improvement recommendations were development of device identification standards and enhanced outcome documentation in patient health records as a means to optimize epidemiologic use of healthcare-related databases. Others have noted similar database and related limitations [6,7]. Some have called for a more comprehensive and integrated approach, along with further database development and linking [8–12]. Partnering among relevant stakeholders, particularly the major government agencies overseeing device development and use, as well as professional medical societies, healthcare institutions, and product developers, has been another recent theme [7,11–14]. Major efforts to enhance postmarket epidemiology and surveillance via use of national registries have been recently initiated by the US Center for Medicare and Medicaid Services and the US National Institutes of Health, in conjunction with professional medical societies and the US Food and Drug Administration [15,16].

REFERENCES

3

These recent developments are a testimony to the importance of epidemiology and surveillance in the postmarket oversight of medical devices. Much has already been achieved, as documented throughout this book. The tools and databases currently utilized are thoroughly described, in both their applications and their strengths and limitations, and the unique aspects of safe and effective device use and optimal postmarket assessment are highlighted via select medical specialty chapters. Closing the ‘gap’ between current practice and future potential is what this book is all about. It provides a starting place to formally recognize medical device epidemiology and surveillance as a coherent field with enormous potential for growth, recognizing the vast and rapidly growing impact of medical devices in patient care and public health. To fully realize its potential, this field will require the concerted effort of both public and private parties.

References 1. Last JM (ed.). A Dictionary of Epidemiology, 2nd edn. New York, NY: Oxford University Press, 1988. 2. Comprehensive Plan for Epidemiologic Surveillance. Atlanta, GA: Centers for Disease Control, 1986. 3. Section 201(h) of the US Food, Drug, and Cosmetic Act, Title 21, US Code No. 321. 4. Section 212 of the Medical Device User Fee and Modernization Act of 2002. 5. Committee on Postmarket Surveillance of Pediatric Medical Devices, Board on Health Sciences Policy. Safe Medical Devices for Children, Field MJ and Tilson H (eds). Washington, D.C. Institute of Medicine of the National Academies, July 2005. 6. Bright RA. Special methodological issues in the pharmacoepidemiology studies of devices. In Pharmacoepidemiology, 3rd edn, Strom BL (ed.). New York: Wiley; 733–747. 7. Califf RM. The need for a national infrastructure to improve the rational use of therapeutics. Pharmacoepidemiol Drug Safety 2002; 11: 319–327. 8. Kereiakes DJ, Willerson JT. Medical technology development and approval: the future is now. Circulation 2004; 109: 3078–3080. 9. Mehran R, Leon MB, Feigal DA, Jefferys D et al. Post-market approval surveillance: a call for a more integrated and comprehensive approach. Circulation 2004; 109: 3073–3077. 10. Baim DS, Mehran R, Kereiakes DJ, Gross TP et al. Post-market surveillance for drug-eluting coronary stents: a comprehensive approach. Circulation 2006; 113: 891–897. 11. Gottlieb S. Opening Pandora’s pillbox: using modern information tools to improve drug safety. Health Affairs 2005; 24(4): 938–948. 12. Maisel W. Safety issues involving medical devices: implications of recent implantable cardioverter-defibrillator malfunctions. J Am Med Assoc 2005; 294(8): 955–958. 13. Kessler LG, Ramsey SD, Tunis S, Sullivan SD. Clinical use of medical devices in the ‘Bermuda Triangle’. Health Affairs 2004; 23(1): 200–207. 14. Califf RM. Evaluation of diagnostic imaging technologies and therapeutic devices: better information for better decisions: proceedings of a multidisciplinary workshop (submitted). 15. Center for Medicare and Medicaid Services. Decision memo for implantable defibrillators (CAG-00157R3). http://www.cms.hhs.gov/mcd/viewdecisionmemo [accessed 27 January 2005]. 16. National Institutes of Health. Interagency registry of mechanically assisted circulatory support (Contract No. HHSN268200548198C). February 2005.

2 Medical device regulation in the USA Thomas P. Gross, Celia M. Witten, and Caspar Uldriks US Food and Drug Administration, Rockville, MD, USA

Introduction The US Food and Drug Administration (FDA) is first and foremost a public health and consumer protection agency. It regulates products worth over $1 trillion, accounting for one-fourth of all dollars spent annually by American consumers [1]. The agency is responsible, through enforcement of the Federal Food, Drug, and Cosmetic Act (‘the Act’) and several related public health laws, for ensuring that: (a) foods are safe, wholesome, sanitary, and properly labeled; human and veterinary drugs are safe and effective; there is reasonable assurance of the safety and effectiveness of devices intended for human use; cosmetics are safe and properly labeled; and public health and safety are protected from electronic product radiation; (b) regulated products are honestly, accurately, and informatively represented; and (c) these products are in compliance with the law and FDA regulations; non-compliance is identified and corrected; and any unsafe or unlawful products are removed from the marketplace [2]. The Center for Devices and Radiological Health (CDRH) is that part of the agency that helps ensure that medical devices are safe and effective and helps reduce unnecessary exposure to radiation from medical, occupational, and consumer products. The industry that CDRH regulates has a US market valued at more than $75 billion as of 2002 and consists of approximately 8000 medical device firms, more than 80% of whom have fewer than 50 employees [3,4]. For regulatory purposes, a medical device is defined as an instrument, apparatus, implement, machine, contrivance, implant, in vitro reagent, or other similar or related Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

6

CH2 MEDICAL DEVICE REGULATION IN THE USA

article, including any component, part, or accessory, which is: (a) recognized in the official National Formulary, or the US Pharmacopoeia, or any supplement to them; (b) intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease in man or other animals; or (c) intended to affect the structure or any function of the body of man or other animals, and which does not achieve its primary intended purposes through chemical action within or on the body of man or other animals and which is not dependent upon being metabolized for the achievement of any of its primary intended purposes (Section 201 of the Act, Title 21 US Code §321). The agency’s public health and consumer protection mandate is carried out through both premarket product evaluation and postmarket oversight that continues over the lifetime of the product, from early design to widespread use and, ultimately, to obsolescence. At major junctures of a product’s life cycle, FDA must weigh the product’s benefits and risks. Central to this risk management function is FDA’s decision for marketing, one that must ensure that beneficial medical products are available (and labeled with adequate information on their benefits and risks) while protecting the public from unsafe products or false claims [5]. Once marketed, a product’s continued safety and effectiveness must be ensured not only by oversight on the part of industry and FDA, but most importantly by healthcare providers’ and patients’ appropriate product selection and use, based on the product’s labeling. This chapter will highlight the agency’s premarket product evaluation process, regulatory tools used in its postmarket surveillance and risk assessment functions, and touch on the agency’s postmarket enforcement activities.

Premarket review The FDA is responsible for ensuring that there is reasonable assurance that a medical device will be useful while not posing unacceptable risks to patients once product marketing begins. Operationally, this goal is accomplished through the FDA’s use of regulatory controls and the classification process.

Controls General controls include provisions relating to labeling, registration and listing, premarket notification, good manufacturing practices (now called Quality Systems regulation), and records and reports. Each of these will be briefly touched on below: Labeling: the label for any prescription device needs to contain information conveying the ‘. . . intended uses of the device and relevant warnings, precautions, side effects, and contraindications . . .’ (Section 502, the Act). Registration and listing: all medical device manufacturers are required, before device manufacturing begins, to register their facility with FDA and list each generic type of medical device manufactured at that facility.

PREMARKET REVIEW

7

Premarket notification: unless specifically exempt, any manufacturer or specification developer intending to market a medical device must submit an application at least 90 days before beginning commercial distribution. The agency then determines whether the device is substantially equivalent to a predicate device. A predicate device is defined as an unclassified device that was legally marketed prior to May 28 1976, or a classified device in Class I or II, or a classified device in Class III for which premarket approval has not yet been required. The concepts of a predicate device and substantial equivalence, and how these are evaluated, will be explained later in the chapter. Manufacturing practices: manufacturing requirements ‘govern the methods used in, and the facilities and controls used for, the design, manufacture, packaging, labeling, storage, installation, and servicing of all finished devices intended for human use’ [Section 520 (f), the Act]. The corresponding comprehensive Quality Systems regulation is designed to help ensure that finished devices will be safe and effective. Records and reports: under Section 519 of the Act, the FDA is authorized to require manufacturers to maintain records and submit reports and such information as is necessary to ensure, among other things, a device’s safety and effectiveness. For some devices, general controls alone may not be adequate to ensure safety and effectiveness. For a subset of these devices, there may be sufficient information to establish special controls in addition to general controls to provide this assurance. Although this concept originally meant that performance standards could be established, Congress broadened the term ‘special controls’ to include such controls as patient registries, guidances, and standards. FDA guidance documents play an important role in the premarket review process, frequently serving as special controls. They are available on the Internet at http:// www.fda.gov/cdrh/guidance.html. Although guidance documents are non-binding on industry and the FDA, they assist industry in preparing regulatory submissions and also assist FDA staff in the review process. They cover a range of topics, including: (a) interpretation of regulatory requirements; (b) contents of an application for a specific device type, or group of devices; and (c) clinical study design for devices to treat certain indications. Special controls also include national and international consensus standards which are developed through accredited standards development organizations (SDOs), such as the American Society of Testing and Materials (ASTM) or the Association for Advancement of Medical Instrumentation (AAMI), with the full participation of the government, industry, and academia. The agency has recognized more than 500 standards. Most of these standards pertain to test methods that can be used to evaluate a device, or material specifications that give the type and quality of the materials used in the manufacture of the devices. If a manufacturer chooses to declare conformity to one of the FDArecognized standards in a new device application, FDA accepts that declaration. It

8

CH2 MEDICAL DEVICE REGULATION IN THE USA

then becomes incumbent on the manufacturer to maintain records that demonstrate conformity to the standard. By declaring conformity to a recognized standard, a manufacturer may be able to eliminate the need to submit some of the detailed information to the agency during the application for approval to market.

Classification Medical devices are classified into three classes, based on the ability of the controls noted above to provide reasonable assurance of safety and effectiveness. The basis for regulatory decisions on medical device clearance takes into account the device together with its intended use. Thus, a scalpel that was labeled to cut tissue could be placed in one regulatory class, while the same scalpel that was labeled to ‘cure cancer’ would be classified differently. The definition of each of the three regulatory classes is provided below: Class I devices: devices for which general controls alone are sufficient to provide reasonable assurance of safety and effectiveness [Sec. 513 (a)(1)(A), the Act]. Examples of Class I devices include manual surgical instruments for general use, hot or cold disposable packs, and limb orthoses. Class II devices: products for which ‘general controls alone are insufficient to provide reasonable assurance of safety and effectiveness, but for which there is sufficient information to establish special controls’ [Sec.513 (a)(1)(B), the Act]. Examples of Class II devices include: large volume infusion pumps, inferior vena cava filters, and hospital beds. Class III devices: devices for which there may be insufficient information to determine that the application of general controls alone, or general and special controls, are sufficient to provide reasonable assurance of safety and effectiveness. If such a device is also life-sustaining, life-supporting, or for use of a substantial importance in preventing impairment of human health, or presents a potential unreasonable risk of illness or injury, then that device is placed into Class III [Sec 513 (a)(1)(C), the Act]. Examples of Class III devices include: spinal fusion cages, silicone gel-filled breast implants, and drug-eluting coronary stents.

Marketing applications To market a device, manufacturers are required to obtain premarket clearance or approval unless the device is specifically exempt. The marketing application will vary, depending on the device classification and regulatory history. What follows is a description of these and the types of devices eligible.

MARKETING APPLICATIONS

9

510(k) premarket notification For Class II products, for the few Class I products that are not exempt from the need for a marketing application, and for the few Class III pre-amendments products (existed prior to the 1976 Medical Device Amendments to the Act) for which FDA has not called for safety and effectiveness data, a 510(k) marketing application is the appropriate mechanism for market entry. This type of application, named after Section 510(k) of the Act, requires a sponsor to provide evidence that the device proposed for market is ‘substantially equivalent’ to a predicate device. (In general, Class I and Class II devices can serve as predicate devices; in some cases Class III devices can also serve as predicate devices if they were on the market before the1976 Medical Device Amendments.) For a device to be deemed substantially equivalent to a previously marketed device, it must be considered to be as safe and effective as its predecessor. In particular, the proposed device must have the same intended use as the predicate device, and also either the same technological characteristics or, if the characteristics are different, the device must be as safe and effective as the predicate device and not raise different questions of safety and effectiveness. Thus, a manufacturer needs to provide a description of the device proposed for market, the predicate device, and a comparison of the technological differences between the products in a marketing application. For technological differences, a discussion and possibly bench testing and/or clinical data regarding the effect of these differences on device performance may be needed. There are a number of regulatory paths a sponsor may follow to market a Class III product that is subject to a premarket approval. If the sponsor has completed data collection on the product, this information would be provided in a premarket application (PMA). A manufacturer who has not completed testing but wants approval of the plan for development of the data may submit a product development protocol (PDP). The humanitarian device exemption path is an alternative for devices that are indicated for limited patient populations (fewer than 4000 individuals). For all three of these types of marketing applications, a summary of the preclinical and clinical testing that form the basis of the regulatory decision is available on the FDA website (http://www.fda.gov/ cdrh/programs.html). Because it is predominantly used, the PMA will be discussed in more detail.

Premarket approval application A manufacturer of a Class III product subject to a premarket approval needs to demonstrate reasonable assurance of safety and effectiveness in order to market the product. The safety and effectiveness of a device are determined with respect to the target population and conditions of use. The effectiveness of the device is to be determined based on ‘well-controlled investigations, including one or more clinical investigations where appropriate, by experts qualified by training and experience to evaluate the effectiveness of the device, from

10

CH2 MEDICAL DEVICE REGULATION IN THE USA

which investigations it can fairly and responsibly be concluded by qualified experts that the device will have the effect it purports or is represented to have under the conditions of use prescribed, recommended, or suggested in the labeling of the device’ or based on other ‘valid scientific evidence’ (Section 513, the Act). FDA has published regulations defining ‘valid scientific evidence’. This hierarchy of evidence is described in the Code of Federal Regulations (21 CFR 860.7) as ‘evidence from well-controlled investigations, partially controlled studies, studies and objective trials without matched controls, welldocumented case histories conducted by qualified experts, and reports of significant human experience with a marketed device . . .’. FDA has used all of these forms of valid scientific evidence in making decisions on PMA products. Clinical data are often required in support of a marketing application when preclinical testing alone may not answer all the questions about device performance. Not all data reviewed in support of medical device approval are from randomized trials. Clinical trials, are, however, frequently carried out to investigate the safety and effectiveness of medical devices.

Postmarket oversight As noted previously, the law requires that medical devices be reasonably safe and effective prior to market entrance. ‘Reasonably safe’, however, is not synonymous with ‘risk-free’. For example, the agency approves a device when it deems that the product’s benefits outweigh its risks for the intended population and use. Thus, when device marketing begins, there is reasonable assurance that the product will be useful while not posing unacceptable risks to patients. For the majority of marketed products, however, no, or very limited, clinical data are required. Of 783 devices classified by regulation into Class I, 720 (92%) exempt the device from premarket notification. Similarly, of the 898 devices classified by regulation into Class II, 75 (8%) exempt the device. For the Class I and II products requiring premarket notification, many applications do not include clinical data. Even when clinical trial information is provided (for Class III devices), these data have inherent limitations. Device clinical trials are typically conducted with limited numbers of patients (< 1000); therefore, detection of rare adverse events or product problems may be unlikely. For instance, if an adverse event occurs in 1 in 10 000 exposed patients, there is only a 9.5% probability of observing that event at least once in a trial of 1000 patients [6]. Trials are typically of short duration (< 3 years); thus, long-term effects may go undetected, whether of long latency or late term (e.g. related to durability). Inclusion criteria for many trials tend to describe a more homogeneous patient population than the populations for which the device is indicated and may be restricted in age (no children, elderly), gender (no pregnant women), comorbid conditions, and disease complexity. In addition, investigators in premarket clinical trials tend to be those physicians at the ‘cutting edge’ of product development and who are most familiar with the device characteristics and applications. Thus, limited information may be generated on human factor concerns, such as optimal design for ease of use, optimal use environment (e.g. free

POSTMARKET OVERSIGHT

11

of electromagnetic interference), labeling that anticipates less sophisticated use or that minimizes maintenance error, or the consequences of re-use on device performance and safety. Once in the marketplace, the devices may be used by a wide array of physicians and other clinicians of varying skill levels, training, and experience. In addition, less stringent diagnostic and other criteria are typically applied reflecting either non-optimal product choice or off-label use, the latter a hallmark of the evolving practice of medicine. Since no device is free of adverse events and product problems, and since premarket clinical data are limited, postmarket oversight is needed as a ‘safety net’ to assure the continued safety and effectiveness of marketed products. ‘Postmarket oversight’ refers to postmarket surveillance and risk assessment as well as postmarket enforcement. The former refers to the systematic process of adverse event/product problem reporting, monitoring, and evaluation, as well as the further, more formal, assessments of identified potential patient risks. The latter refers to investigations of a device firm’s compliance with statutory and regulatory requirements. The purposes of both these processes (surveillance and enforcement) are to: (a) disseminate information regarding newly emerging device problems to appropriate stakeholders (particularly healthcare professionals and the public); (b) incorporate the information into the device testing and approval process; and (c) provide findings to the device industry to aid in product corrections and improvements. Both processes are integral to product development and evolution. The remainder of this section will focus on the programs constituting postmarket oversight, beginning with postmarket surveillance.

Postmarket surveillance The goals of postmarket surveillance and risk assessment are identification of previously unknown or not well-characterized adverse events/product problems (‘signals’), identification and characterization of subgroups at risk, collection and evaluation of information on issues not directly addressed in premarket submissions (e.g. long-term effectiveness), and development of a public health context to interpret these data. FDA’s postmarket ‘tools’ to achieve these goals primarily are: (a) adverse event/product problem reporting, through the Medical Device Reporting system, MedWatch, PMA conditions of approval, and the pilot Medical Device Safety Network (MedSun); (b) mandated postmarket studies, including conditions of approval and Section 522 studies; and (c) applied epidemiologic research. MedSun and applied epidemiologic research will be addressed in separate chapters.

Adverse event/product problem reporting FDA monitors postmarket device-related adverse events/product problems (AEs: this is shorthand for adverse events/product problems and will be used throughout this chapter), through both voluntary and mandatory reporting, to detect ‘signals’ of potential public

12

CH2 MEDICAL DEVICE REGULATION IN THE USA

health safety issues. Voluntary reporting to FDA began in 1973 and presently continues under MedWatch (7), a program created in 1993 to encourage voluntary reporting by all interested parties, but principally among healthcare professionals as a critical professional and public health responsibility. It was not until 1984 that the FDA implemented mandatory reporting per the Medical Device Reporting (MDR) regulation. This regulation required device manufacturers and importers to report device-related deaths, serious injuries, and malfunctions to the FDA. Additional legislative initiatives in the 1990s resulted in significant changes to mandatory reporting. Under the Safe Medical Devices Act (SMDA) of 1990, universal reporting of adverse events by user facilities (hospitals, nursing homes, ambulatory surgical facilities, and outpatient diagnostic and treatment facilities) and distributors was required. Under the FDA Modernization Act (FDAMA, Section 213), and in response to experience with distributor and user facility reporting, distributor reporting was no longer required and universal user facility reporting was limited to a ‘. . . subset of user facilities that constitutes a representative profile of user reports . . .’. The conceptual framework for these sentinel sites, collectively referred to as the Medical Product Safety Network (MedSun), is discussed in its own chapter (Chapter 5). To better understand reporting of AEs under the current MDR regulations (21 CFR 803, last revised in 1998), requirements should be noted and terms defined. Manufacturers and importers are currently required to submit reports of device-related deaths, serious injuries, and malfunctions. User facilities are required to report deaths to the FDA and deaths and serious injuries to the manufacturer. Serious injuries are defined as lifethreatening events, events that result in permanent impairment of a body function or permanent damage to a body structure, and events that require medical or surgical intervention to preclude permanent impairment or damage. Malfunctions are defined as the failure of a device to meet its performance specifications or otherwise perform as intended. The term ‘device-related’ means that the event was or may have been attributable to a medical device, or that a device was or may have been a factor in an event, including those occurring as a result of device failure, malfunction, improper or inadequate design, poor manufacture, inadequate labeling, or use error. Guidance is issued to reporting entities as needed to more clearly define the reporting of specific events (e.g. implant failures). Since its inception in 1973, FDA’s database of voluntary and mandatory reports of device AEs has received slightly more than 1.25 million reports and currently averages approximately 200 000 per year, with mandatory reports accounting for about 98% of the total. The reports are submitted on standardized forms which capture information on device specifics (e.g. brand name, model number), event description, pertinent dates (e.g. event date), and patient characteristics. The FDA has devised methods for report triage to enhance signal detection of previously unforeseen or not-well-characterized adverse events; thus, only about one-fourth of reports received require individual review. Wellcharacterized and understood AEs are either automatically computer-screened or submitted as periodic tabular summaries by the manufacturer (for detection of trends). Clinical staff individually review the reports from a variety of perspectives, including the potential for device failure (e.g. poor design, manufacturing defect), use error

POSTMARKET OVERSIGHT

13

(e.g. device misassembly, incorrect clinical use, misreading instructions), packaging error, support system failure, adverse environmental factors, underlying patient disease or comorbid conditions, idiosyncratic patient reactions (e.g. allergy), maintenance error, and adverse device interaction (e.g. electromagnetic interference) [8]. Aside from routine requests for follow-up information, several actions may be taken and include inspections of manufacturers (which may ultimately lead to label changes or product recall and, rarely, product seizure or injunction), internal expert meetings (which may lead to public health notifications or requirements for additional postmarket study), and alerting regulatory authorities outside the USA. Public health notifications can be issued by the firm (‘Dear Doctor’ or ‘Dear Patient’ letters) or by the FDA, depending upon the specific device issues. Other uses of the AE data are widespread and include input into classification and monitoring of recalls; product classifications and 510 k exemptions; standards efforts; premarket review (by providing human factors insights and information on product experience in the general population); educating the clinical community through newsletters, literature articles (peer-reviewed and professional and trade journals), and the agency’s Patient Safety News program (http://www.fda.gov/ cdrh/psn); and as a general information resource for healthcare providers and the general public (http://www.fda.gov/cdrh/MAUDE). An example of reports of AEs typifies the system in action. In June 2002, the agency received reports of bacterial meningitis in patients with cochlear implants for treatment of hearing loss. Early speculation by manufacturers and implanting surgeons implicated the implant positioner (a silastic wedge that is inserted next to the implanted electrode to facilitate transmission of the electrical signal by pushing the electrode against the medial wall of the cochlea). The one manufacturer that made implants with a positioner voluntarily withdrew their product in both Europe and the USA in July 2002. Other manufacturers, however, notified the agency of additional cases of meningitis, principally in children. A nationwide collaborative investigation was begun by the agency and the Centers for Disease Control and Prevention (CDC) that involved several thousand implanted children. These children were found to have far greater risk of developing pneumococcal meningitis compared to children in the general population, and those with positioners had over four times the risk of developing meningitis compared to recipients of other cochlear implant types [9]. Throughout this process, the agency posted periodic updated public health notifications on its website to keep the public informed [10]. In addition, the CDC Advisory Committee on Immunization Practices added cochlear implant recipients to the list of high-risk patients needing routine immunizations [11]. For additional information on the cochlear implant issue, please see Chapter 26. As is typical of passive surveillance systems, FDA’s system has notable weaknesses, including: (a) data may be incomplete or inaccurate and are typically not independently verified; (b) data may reflect reporting biases driven by event severity or uniqueness or publicity and litigation; (c) events are generally underreported and this, in combination with lack of denominator data, precludes determination of event incidence or prevalence; and (d) causality cannot be inferred from any individual report. With regard to causality assessment, device retrieval and analysis data are often inadequate or lacking. Although there are no current reporting authorities that systematically require such data, FDA

14

CH2 MEDICAL DEVICE REGULATION IN THE USA

Quality Systems regulations do require investigation of any failure of a device to meet its performance specifications. The system also has many strengths, including: (a) provides nationwide safety surveillance from a variety of sources, thus providing insight into AEs related to ‘real-world’ use; (b) is one of only a few means to detect rare AEs; (c) is relatively inexpensive, considering the scope of surveillance; (d) data collected are uniform in terms of a standardized form with prespecified data elements; (e) is accessible and the information is open to the public. This data is posted on the FDA website and updated quarterly. Supplementing this reporting system are PMA conditions of approval. All products with approved PMAs have conditions of approval [21 CFR 814.82 (a)(9)]. One of these conditions is the submission of information on AEs outside the MDR regulatory requirements. Examples of this include labeled AEs occurring with unexpected severity or frequency. This requirement helps the agency cast a wider ‘safety net’ in its surveillance of AEs.

Mandated postmarket studies Another ‘tool’ that FDA uses to achieve its surveillance and risk assessment goals is the mandated postmarket study, conducted under either PMA conditions of approval or FDAMA (Section 522) authorities. A sponsor may be required to perform a postapproval study as a condition of approval for a PMA [21 CFR 814.82(a)(2)]. The study questions may relate to longer-term performance of an implant, or focus on specific safety issues that may have been identified during review of the product and for which additional postmarket information is needed. Results from these studies may be included as revisions to the product’s labeling (including patient- and clinician-related material). In addition to the PMA authority, the agency may, under Section 522 of FDAMA, impose postmarket study requirements on certain devices. This provision, originally added to the Act in 1990 under SMDA, allows the agency to order a manufacturer of a Class II or Class III device to conduct a postmarket study if the device: (a) is intended to be implanted in the human body for more than 1 year; (b) is life-sustaining or lifesupporting (and used outside a device user facility); or (c) failure would reasonably be likely to have serious adverse health consequences. Although this discretionary authority overlaps the PMA postapproval authority for some products (e.g. PMA implants), it effectively extends FDA authority to cover 510(k) products as well. Unless there are unusual circumstances, the Section 522 authority is typically reserved for 510(k) products and can be ordered at any time, starting from the time of clearance. Prior to issuing an order, FDAwill discuss the public health concern with the firm. The concern may arise from questions about a product’s long-term safety, about performance of a device in general use or in a new user setting (e.g. a move from professional use only to also including home use), or notable AEs. Upon receiving an order, the firm has up to 30 days in which to submit their study plan and, by statute, studies are limited to 3 year patient follow-up (or longer if agreed to by the firm). The possible study approaches vary widely (designed to capture the most practical, least burdensome approach to produce a

POSTMARKET OVERSIGHT

15

scientifically sound answer) and include: detailed review of complaint history and the literature; non-clinical testing of the device; telephone or mail follow-up of a patient sample; use of registries; other observational study designs; and, rarely, randomized controlled trials. FDA issued a regulation (21 CFR 822) in mid-2002 that clearly specifies the requirements for the study plan, conduct, and follow-up, and the potential consequences of failing to meet these requirements. Generally speaking, these mandated postmarket studies (under both PMA conditions of approval and Section 522) require the participation of both the involved firms and the clinical community. However, problems may arise in the conduct of these studies if, for instance, it is difficult to recruit physician investigators or accrue patients, or if industry lacks incentive. These issues particularly resonate with rapidly evolving technologies, where rapid device evolution may make prior models obsolete by the time their studies are completed (although patients with prior generation permanent implants have to live with the risks/benefits of their older devices). Although there may be difficulties in study conduct, an example of a Section 522 study reveals the authority’s public health importance and its risk assessment role. In 1991, FDA scientists demonstrated that it was possible for polyurethane to break down under laboratory conditions to form 2,4-toluenediamine (TDA). TDA had been shown to be an animal carcinogen. Prior to this, it was thought that breakdown could only occur at very high temperatures and pH extremes. The firm which manufactured polyurethane foamcoated breast implants ceased sales in 1991 and agreed to a clinical study under Section 522. The study involved comparing TDA levels in urine and serum samples from women with and without the implants. Although minute amounts of TDA were found in the majority of women with the implants, the increase in cancer risk was determined to be vanishingly small (1 in 1 million) [12,13]. FDA issued a public health correspondence (FDA Talk Paper) on the results and their reassuring implications [14].

Postmarket enforcement Enforcement activities for FDA’s premarket and postmarket medical device programs focus on developing and implementing legal and administrative remedies to effect compliance with applicable federal statutory and regulatory requirements. Regulatory and enforcement policies address specific issues, such as device labeling, manufacturing practices, AE reporting, recalls, advertising and promotion, and performance standards for medical and non-medical electronic products that emit radiation. Enforcement actions usually are directed at a manufacturer or importer, but may also include a sponsor of a PMA, an institutional review board or, on rare occasions, a particular individual. Enforcement activities, however, are not the agency’s preferred venue for effecting compliance. The agency speaks with the device industry through formal and grassroots meetings to explain policies, to resolve issues in dispute, to review industry complaints, and to provide training and guidance to new and existing firms. The agency remains upto-date on industry practices, trade developments, scientific innovations, the scientific literature, and the activities of professional associations. When appropriate, the agency

16

CH2 MEDICAL DEVICE REGULATION IN THE USA

amends regulatory requirements to reflect changes in trade practices, technology, and professional practices. FDA’s medical device tracking program (see below) is an example of a postmarket program that is periodically amended due to technological advancements and changes in professional practices.

Tracking FDA’s medical device tracking program serves as a regulatory safety net to ensure that manufacturers of certain devices establish a tracking system. The system will enable them, on their own initiative, to promptly locate and remove devices in commercial distribution and provide patient/physician notification (regarding the safety concerns that may require special clinical management) if that is the recall strategy. Tracking also augments FDA’s recall authority to order a mandatory recall under Section 518(e) of the Act. By law, the agency may require tracking for a Class II or Class III device: (a) the failure of which would be reasonably likely to have serious adverse health consequences; or (b) which is intended to be implanted in the human body for more than one year; or (c) which is life-sustaining or life-supporting and used outside a user facility [Section 519(e) of the Act]. When exercising its discretion to require tracking, the agency considers other factors similar to those identified in the MDR regulation that comprise serious injuries, i.e. likelihood of sudden catastrophic failure; likelihood of significant adverse clinical outcome; and need for prompt professional intervention. The agency may add or remove devices from the list of tracked devices as a result of its review of premarket applications, postmarket surveillance (including reports of AEs), recalls, or other information coming to its attention. Manufacturers who receive tracking orders must implement tracking procedures and collect information required by the tracking regulation (21 CFR 821, as amended). Permanently implanted devices and life-sustaining or life-supporting devices that are intended for use by a single patient over the life of the device must be tracked to the patient using the device. Manufacturers are required to audit their tracking system, which requires effective communication through the chain of distribution. Manufacturers need to ensure that distributors and hospitals comply with their informationreporting obligations. Final distributors of a tracked device, which includes doctors and hospitals, must report to the manufacturer, among other items, the name, address, and telephone number of the patient to whom it distributed the device, as well as the prescribing physician and physician who regularly follows the patient [21 CFR 821.30(b)]. Manufacturers have used their tracking system to conduct a voluntary recall. FDA, however, has not ordered a manufacturer to produce tracking information in conjunction with a FDA-ordered recall (FDA guidance on medical device tracking is available on the Internet at www.fda.gov/cdrh/modact/tracking.pdf). A published case study of medical device tracking highlights its important public health function [15]. Ventritex, a manufacturer of implantable cardioverter defibrillators

POSTMARKET OVERSIGHT

17

(ICDs), used its tracking database to alert patients and their physicians of an immediate need to reprogram model V-110 and V-112 ICDs. This action was taken in response to reports received by the firm of a fatal tachycardia, and analysis of the device indicating the potential for a failure mode that could affect device performance in a variety of ways. Of the approximately 5600 patients implanted with the device and alive at the time of the alert, 98.7% were successfully located and their devices reprogrammed within the first 60 days of the notification. Ultimately, > 99.8% of devices in patients were reprogrammed. This case study demonstrated that most tracked device recipients could be located and receive medical intervention.

Recall authority/field inspections Recall activities involving medical devices are termed ‘corrections’ and ‘removals’ by FDA. A ‘correction’ covers a number of activities, including the repair, modification, adjustment, relabeling, destruction, inspection, or patient monitoring of a device, even without physical removal from its point of use. A ‘removal’ also covers a number of activities, including the physical removal of a device from its point of use to some other location for repair, modification, adjustment, relabeling, destruction, or inspection. Under Section 519(f)(1) of the Act and 21 CFR 806, manufacturers and importers must report to FDA any action undertaken: (a) to reduce a risk to health posed by the device; or (b) to remedy a violation of the Act caused by a device which may present a risk to health. No report to FDA is required if the correction or removal does not present a risk to health; however, a firm may voluntarily report a correction or removal as part of FDA’s voluntary recall policy (21 CFR 7) and it must keep a record of the correction or removal. The definition of ‘risk to health’ plays an important role in whether a correction or removal must be reported to the agency. The phrase ‘risk to health’ means: (a) a reasonable probability that use of, or exposure to, the product will cause serious adverse health consequences or death; or (b) that use of, or exposure to, the product may cause temporary or medically reversible adverse health consequences, or an outcome where the probability of serious adverse health consequences is remote [21 CFR 806.2(j)]. Reporting requirements of corrections and removals may appear to overlap AE reporting requirements. This is particularly germane for AE reports that involve death or serious injury, that require remedial action to prevent an unreasonable risk of substantial harm to the public health. Typically, however, individual AE reports will not lead to remedial action; rather, remedial action is generally based on an AE case series, pattern, or trend. Manufacturers must implement procedures to collect and analyze information about their medical devices to identify potential product and quality problems that may need preventive action or that require recall consideration (21 CFR 820, Quality System regulation) [16]. A manufacturer’s assessment may be based on analysis of information provided through AE reporting, quality control records, or internal quality audit reports. During an on-site inspection of a manufacturer, a FDA field

18

CH2 MEDICAL DEVICE REGULATION IN THE USA

investigator will determine whether the firm analyzes data concerning non-conforming product (i.e. one that does not meet specifications). The FDA investigator will also review the firm’s statistical control techniques and other data relevant to failure investigations including non-statistical methods such as ad hoc committee reviews. Where a firm’s failure investigation identifies a root cause, the agency will assess the adequacy of the firm’s corrective action. Corrections and removals are considered a form of corrective action. Recalls usually are conducted on a voluntary basis, in order to protect the public health or correct a violation of the Act. In 1990, Section 519(f) of the Act was amended by the SMDA, requiring FDA to issue reporting and record-keeping requirements concerning the recall activities. Congress established this statutory requirement because it believed that device manufacturers and importers would conduct recalls without notifying FDA in a timely fashion, if at all. When a firm fails to correct or remove dangerous devices from the market promptly, and FDA finds that there is a reasonable probability that a device would cause serious adverse health consequences or death, FDA will issue an order, under Section 518(e) of the Act, to require the appropriate person to: (a) immediately stop distribution of the device; and (b) immediately notify health professionals and device user facilities of the order and instruct them to stop using the device [17]. When a device is the subject of a mandatory recall order and the device has been subject to medical device tracking requirements, the agency expects the responsible firm to use the relevant tracking information required by the tracking regulation (21 CFR 821). If necessary, the agency can order the firm to produce product distribution records, obtained under the tracking authority, and use that information to ensure that a mandatory recall is effective and prompt. Firms are expected to conduct effectiveness checks to make sure the users or consignees have received notice of the correction or removal and that they have taken the appropriate action. FDA field staff will audit the firm’s effectiveness checks. When a correction or removal involves a device that poses substantial risk of serious injury or death, the agency will contact each user or consignee to ensure that the appropriate action has been taken.

Conclusion The seamless integration of premarket review with postmarket oversight is the basis for an optimal system, designed to assure that the benefits of a medical device outweighs its risks throughout the product’s total life cycle up to obsolescence. Once marketed, a product’s continued safety and effectiveness must be ensured, not only by oversight on the part of industry and FDA, but also, most importantly, by healthcare providers’ and patients’ appropriate product selection and use, based on the product’s labeling. This chapter briefly highlighted the agency’s premarket product evaluation process, regulatory tools used in its postmarket surveillance and risk assessment functions, and touched on the agency’s postmarket enforcement activities.

REFERENCES

19

References 1. Food and Drug Administration. Frequently asked questions: FDA (general). Rockville, MD: US Food and Drug Administration, March 1999. Food and Drug Administration internet site: www.fda.gov 2. Food and Drug Administration. Index: FDA’s mission. Rockville, MD: US Food and Drug Administration; March 1996, October 1998. Food and Drug Administration internet site: www.fda.gov 3. US Department of Commerce. FY2004 industry assessment: medical equipment. 2004. 4. Gallivan M. The 1997 Global Medical Technology Update: The Challenges Facing US Industry and Policy Makers. Washington, DC: Health Industry Manufacturers’ Association, 1997; 37–51. 5. Food and Drug Administration. Managing the risks from medical product use: creating a risk management framework. Report to the FDA Commissioner from the Task Force on Risk Management. Rockville, MD: US Food and Drug Administration; May 1999: http://www.fda. gov/oc/tfrm/Tableofcontents.htm 6. O’Neil RT. Assessment of safety. In Biopharmaceutical Statistics for Drug Development, Peace KE (ed.). New York: Marcel Dekker, 1988; 543. 7. Kessler DA. Introducing MeDWatch: a new approach to reporting medication and device adverse effects and product problems. J Am Med Assoc 1993; 269: 2765–2768. 8. ECRI. Medical device problem reporting for the betterment of healthcare. Health Devices 1998; 27(8): 277–292. 9. Reefhuis J, Honein MA, Whitney CG et al. Risk of bacterial meningitis in children with cochlear implants, USA 1997–2002. N Engl J Med 2004; 349: 435–439. 10. Food and Drug Administration. Risk of Bacterial Meningitis in Children with Cochlear Implants. FDA Public Health Web Notification. Rockville, MD: US Food and Drug Administration, 2002, 2003. 11. Center for Disease Control and Prevention. Recommended childhood and adolescent immunization schedule: United States, 2003. Morb Mortal Wkly Rep 2003; 52: Q1–Q4 [Erratum, Morb Mortal Wkly Rep 2003; 52: 191]. 12. Hester TR Jr, Ford NF, Gale PJ et al. Measurement of 2,4-toluenediamine in urine and serum samples from women with Meme or Replicon breast implants. Plast Reconstr Surg 1997; 100(5): 1291–1298. 13. DoLuu HM, Hutter JC, Bushar HF. A physiologically based pharmacokinetic model 2,4-toluenediamine leached from polyurethane foam-covered breast implants. Environ Health Perspectives 1998; 106(7): 393–400. 14. Food and Drug Administration. TDA and Polyurethane Breast Implants. Food and Drug Administration talk paper. Rockville, MD: US Food and Drug Administration, June 1995. 15. Kaczmarek RG, Beaulieu MD, Kessler LG. Medical device tracking: results of a case study of the implantable cardioverter defibrillator. Am J Cardiol 2000; 85: 588–592. 16. Food and Drug Administration. Guide to Inspections of Quality Systems. Rockville, MD: US Food and Drug Administration, August 1999. 17. Food and Drug Administration. Regulatory Procedures Manual. Rockville, MD: US Food and Drug Administration, August 1997.

3 Medical device epidemiology Roselie A. Bright US Food and Drug Administration, Rockville, MD, USA

S. Lori Brown US Food and Drug Administration, Bothell, WA, USA

Introduction In this overview chapter, medical devices are defined, examples of devices given, and the ubiquity and importance of medical devices for patient safety and public health presented. The features of medical devices that must be accounted for when designing epidemiologic studies are discussed, and we compare and contrast them with the analogous features of drugs. Descriptive and hypothesis-testing study designs are discussed and illustrated by examples of studies that have dealt successfully with the challenges of studying medical devices.

Medical Devices Medical devices have been used for thousands of years. They are ubiquitous in health care settings, including the home, school, outdoors, and places of business, as well as hospitals, nursing homes, clinics, and health care offices. In the USA they are regulated by the Food and Drug Administration (FDA), which lists over 2000 types. They serve a wide and expanding variety of uses (Table 3.1).

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

22

CH3 MEDICAL DEVICE EPIDEMIOLOGY

Table 3.1 Purposes of medical devices with examples. The organization of purposes is similar to the organization of the legal definition of devices [1] Device purpose

Examples

Diagnose, prevent, monitor, treat, or alleviate disease

In vitro diagnostic test kits . Pregnancy test kits . Drug test kits . Occult blood test kits MRI Endoscopes Latex gloves Infant apnea monitors Multiphysiologic monitors Blood glucose monitors Cardioverter defibrillators Sutures Surgical staplers Needles and syringes Infusion pumps Hot water bottles Bandages Intracranial pressure monitor Wheelchairs Contact lenses Eyeglasses Hearing aids Dialysis devices Cardiac pacemakers Artificial heart valves Joint prostheses Respirators Acupuncture needles Condoms Contraceptive diaphragms

Monitor, treat, or compensate for injury/handicap

Replace or modify anatomy/physiological process

Regulatory definition of medical devices A medical device is defined by US law as: ‘. . . an instrument, apparatus, implement, machine, contrivance, implant, in vitro reagent, or other similar or related article, including any component, part, or accessory, which is recognized in the official National Formulary, or the United States Pharmacopeia, or any supplement to them, intended for the use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease, in man or other animals, or intended to affect the structure or function of the body of man or other animals, and which does not achieve its primary intended purposes through chemical action within or on the body of man or other animals and which is not dependent upon being metabolized for the achievement of its primary intended purposes’ [1].

INTRODUCTION

Figure 3.1

23

Prostheses that are generally implanted long-term

In other words, virtually every medical product that isn’t a drug, vaccine, or therapeutic biologic, is a medical device. Examples of prostheses that are generally implanted long-term are illustrated in Figure 3.1. Many medical specialties use implanted devices, including orthopedic, cardiovascular, ophthalmic, dental, and cosmetic. Some of these devices, such as some cardiovascular devices, may be life-preserving or life-saving, while others serve a purely cosmetic function, such as breast or chin implants.

Ubiquity and importance of medical devices Medical devices are used in the home, school, outdoors, and places of business, as well as hospitals, nursing homes, clinics, and health care offices. Users include physicians,

24

CH3 MEDICAL DEVICE EPIDEMIOLOGY

dentists, nurses, health care technicians, laboratory technicians, physician assistants, aides, patients, and patient’s family members. Our society relies on medical devices for every level of disease prevention, health maintenance, diagnosis and treatment, as suggested by the functions of devices listed in the definition. Many common diseases are prevented and treated with surgery, medication, devices, or a combination (Table 3.2). For example, unwanted fertility can be prevented permanently with surgery or temporarily with spermicides, hormones, condoms, or Fallopian tube devices. Three types of combinations are also used: independently effective treatments used together (e.g. spermicides with either condom, diaphragm, or cervical cap); drug delivered with a device (e.g. hormone delivered via injection, patch, intravaginal ring, or subcutaneous or intra-uterine capsule); and products that are drug–device integrated combinations of products that each contribute therapeutic effect (e.g. coppereluting intra-uterine device). Another example is coronary artery blockage, which can be treated with surgery, a tissue plasminogen activator, a stent, or a drug-eluting coronary stent. Atrial fibrillation treatments include drugs or an automatic implantable cardioverter defibrillator (ICD). Gastro-esophageal reflux disease is commonly treated with proton pump inhibitor drugs or injectable esophageal bulking agents.

Importance of medical device epidemiology Reducing medical adverse events (AEs) and improving patient safety is firmly established as an international priority [2]. Insights gained in recent years include calculations of the pervasiveness and seriousness of AE [2–5], an increasing recognition that the culture of blame must be replaced with an emphasis on systems and work practice solutions [2], and a receptiveness to what can be learned from industries outside of healthcare [6–11]. The major studies of AEs [4,12,13] did not subdivide types of events in ways that would specifically indicate those related to medical devices; one showed evidence that important adverse medical device event (AMDE) information was not specifically categorized for the report [14], and another showed that it was sometimes not noted in the medical charts [15]. Several authors have opined that problems specifically associated with medical devices have been relatively understudied [16–19]. Substantial literature on specific devices exists in the cardiology [20–23], orthopedics [24–26], and anesthesia specialties [16,27]. The few studies specifically done to estimate the overall frequency of medical device problems have been funded by FDA: A 1 year study of visits to emergency departments measured the frequency of serious AMDEs that occur outside of hospitals [28]. The results estimated that in a 1 year period, 452 000 visits to emergency departments were for injuries associated with medical devices. Of these, 58 000 patients died in the emergency department or were hospitalized. More details about this study are included in Chapter 6. A tertiary care hospital study [29] evaluated different types of surveillance systems, including an online incident reporting system, computer flags, and a retrospective

Coronary artery bypass graft Balloon angioplasty

Coronary artery blockage ACE inhibitor Calcium channel blocker Proton pump inhibitor

Tissue plasminogen activator

Spermicide Hormone (estrogen and/or progestin) delivered orally

Medication

Automatic implantable cardioverter defibrillator Injectable esophageal bulking agents

Coronary stent

Condom Fallopian tube clips Fallopian tube insert

Form of therapy Device

Spermicide with: . Condom . Diaghragm . Cervical cap Hormone delivered via: . Intramuscular injection . Transdermal patch . Intravaginal ring . Subcutaneous implanted capsule . Intra-uterine capsule Copper-eluting intra-uterine device Drug-eluting coronary stent

Combination

*Note that surgery involves the use of devices. This column is restricted to procedures that do not leave devices behind in the patient.

Gastro-esophageal reflux disease

Atrial fibrillation

Fallopian tube sterilization Hysterectomy Oopherectomy

Surgery*

Various prevention and treatment options for some common conditions

Unwanted fertility

Condition to be treated

Table 3.2

26

CH3 MEDICAL DEVICE EPIDEMIOLOGY

method using discharge claim diagnosis codes. The detection rates for each system were significantly different: 1.6/1000 discharges for online reports, 27.7/1000 discharges for computer flags, and 64.6/1000 discharges for diagnoses listed in the discharge claims. More AMDEs were recorded in computerized patient records reflecting real-time in-hospital AMDEs (17-fold) and discharge codes (40-fold) reflecting either reason for admission (approximately 55%) or in-hospital AMDEs (approximately 45%) than were in the hospital’s AE database. The list of diagnosis codes used by Samore et al. [29] to examine discharge claims was examined in the publicly available database HCUPnet, which is available on the website of the Agency for Healthcare Quality and Research [30]. The analysis resulted in the estimate that over a million such AMDEs occur annually in US hospitals, at a rate of 6.3 AMDEs/1000 patient days [31]. The study [32] that was done to further investigate the rates of AMDEs made the highest overall estimate so far of the rate of adverse medical device events (240/1000 patient days), did so by directly observing patients in intensive care units (ICUs). This rate was higher than the rate found in their earlier records-based study [29] (15/1000 patient days hospital-wide) and the more recent estimate [29] from hospital discharge claim diagnosis codes of 6.3/1000 patient days. Probably because the method is more rigorous, the rate found by direct observation is also much higher than previously reported device-related rates (expressed per 1000 patient days), for a neonatal ICU (> 16 [33]) and a pediatric ICU (> 19 [34]). These focused studies were compared to published studies that estimated rates of adverse events per patient day, shown in Table 3.3. For those that allowed a direct internal comparison of the rates of device vs. drug adverse events (again, per 1000 patient days), the rates for devices were in similar range (> 16 related to devices and 20 related to drugs in the Frey et al. study [33]) or greater (> 19 device-related and three drug-related in the Stambouly et al. study [34]). Given the emphasis placed by the Institute of Medicine report To Err is Human [2] on the importance of adverse drug events, it is interesting to note that the rates reported for medical devices have been comparable to those for drugs (expressed per 1000 admissions): for adverse drug events in hospitals (10 [35], 2.3 [36], 12 [37]), hospital general medical service (3 [38]), geriatric and internal medicine clinical centers (4.8 [39]), a medical ICU (19 [37]), a surgical ICU (11 [37]), and a medical ICU and coronary care unit (38 [40]).

Features of medical devices that are relevant to epidemiology study design Three types of features are important to the design of epidemiologic studies of medical devices: regulatory, intrinsic, and cultural.

27

FEATURES OF MEDICAL DEVICES THAT ARE RELEVANT TO EPIDEMIOLOGY STUDY DESIGN

Table 3.3 Summary of calculated rates of medical adverse events, defined in the studies as meeting at least one of the following criteria: prolonged the hospitalization, resulted in permanent severe harm, or required medical or surgical intervention Adverse event rate (n/1000 patient days) Population Hospital Adult tertiary hospital All US hospitals Hospital general medicine service Urban tertiary hospital French hospitals Hospital 41 Italian geriatrics and internal medicine clinical centers Intensive care unit (ICU) Adult shock/trauma ICU and adult medical/ surgical ICU Neonatal ICU Pediatric ICU Medical ICU Surgical ICU Medical ICU and coronary care unit

Reference

Device-related

Samore et al. [29] Bright and Shen [31] Chaudry et al. [38]

Drug-related

All*

3

9.3

15 6.3

Bates et al. [35] Hanesse et al. [36] Bates et al. [37] Carbonin et al. [39]

10 2.3 12 4.8

Samore et al. [32]

240

Frey et al. [33] Stambouly et al. [34] Bates et al. [37] Bates et al. [37] Rothschild et al. [40]

>16 >19

20 3 19 11 38

67 27

81

*All adverse events of any type.

Regulatory features Many aspects of devices are similar to drugs but are simply expressed in different terms. Most of these differences in terminology derive from laws and regulations. Table 3.4 shows the terms that have similar meanings for devices and drugs.

Table 3.4 Terms with similar meanings regarding medical device and drug regulation Medical device term

Drug term

Biocompatibility Failure Malfunction Quality Systems (ISO 9000) Postapproval studies Device–device or device–drug interactions Product code, for type of device

Toxicology Ineffectiveness (doesn’t perform at all) Ineffectiveness (doesn’t perform as intended) Good manufacturing practices (cGMP) Phase IV studies Drug interactions National Drug Code, for specific drug

28

CH3 MEDICAL DEVICE EPIDEMIOLOGY

One of the major regulatory differences between drugs and devices is the product identification code system. For drugs, the FDA uses the National Drug Code (NDC), which specifies the active ingredient, manufacturer, form, strength, and package. US manufacturers, pharmacists, and health care providers also use NDC codes, which are clearly marked on drug packages [41–43] (similar systems are used in other countries [44–46]). By contrast, for devices, FDA has been using the Product Code system, which places devices into general categories [47]. Product Codes are not indicated on product labels and are not used by the health care delivery system. An alternate system, the Universal Medical Device Nomenclature SystemTM, was developed by a private consulting firm and is used by many countries [48]. A new system, the Global Medical Device Nomenclature, has recently been initiated and is discussed in Chapter 7 of this book.

Intrinsic features Many intrinsic features of devices are different from drugs, as shown in Table 3.5. Each of these features is an important consideration when designing epidemiologic studies of devices. Devices can have short to long product cycles (compared to the long product cycle for drugs), from conception to obsolescence. In addition, device manufacturers may incrementally modify the design of the device so that, over a period of time, the device being marketed is quite different from the device that was originally cleared. This type of incremental change would never be seen with a drug because of the different regulatory environment. Changes in device design over time can result in a mix of device Table 3.5

Features of devices compared to drugs

Device feature

Drug feature

Short to long product cycles, from conception to obsolescence Incremental changes or modifications over time that renew product cycles Single or multiple active components

Long product cycle from conception to obsolescence

Single or multiple uses of same device Single or multiple users of same device Some medical devices have expiration dates Short to long life of individual devices Cessation of use simple for external devices and temporary implants; surgical and sometimes complex cessation of use for long-term implants Long-lived individual equipment may undergo replacement of components

Active chemical rarely changes over product cycles Single or multiple active ingredients stay fixed after approval Single or multiple uses of drug Single user of same drug Drug expiration dates required Short half-life within body Cessation of use simple

[No drug counterpart]

FEATURES OF MEDICAL DEVICES THAT ARE RELEVANT TO EPIDEMIOLOGY STUDY DESIGN

29

features at any point in time, which in turn can complicate interpretation of the impact of particular device designs. Although some devices, such as sutures, are simple, many devices are quite complex and have multiple components. Examples of complex devices include magnetic resonance imaging (MRI) machines and ventilators. When complex equipment undergoes routine maintenance or replacement of failed components, it may be adjusted to new specifications or acquire new parts that are updated or from a different manufacturer. In some cases, an old piece of equipment may be made of parts from several manufacturers. An example of complexities that can make epidemiologic assessment very difficult is the potential for multiple users of the same device, such as infusion pumps, endoscopes, ventilators, and MRI units, which can result in difficulties when assessing device exposure rates. Re-use is unique to medical devices. Endoscopes, for example, are designed to be used for many years on many patients, and to be cleaned and sterilized between each use. Insufficient cleaning and sterilization can increase the risk of infection transmission between patients (see Chapter 17 on ‘Medical device-related outbreaks’). However, overly harsh cleaning processes may degrade the integrity of the endoscope material, resulting in outright breakage or unsuccessful cleaning. Dialysis membranes are designed to be used many times by the same patient, with cleaning and sterilization between uses; because of economic factors, the number of uses is often higher than specified by the manufacturer, even though the membrane integrity degrades with each use. Angioplasty catheters, for example, are so expensive that even though they are designed to be used once, some facilities reprocess them for re-use on other patients, with attendant loss of functionality and sterility. The FDA now regulates such facilities that reprocess single-use medical devices as manufacturers [49]. Individual devices may have a short (disposable needle and syringe) or long (X-ray machine or orthopedic implant) life, but drugs generally have a short half-life. Compared to the simple cessation of drug use, cessation of device use can be complicated if it is an implant. More of the features of devices and issues that are relevant to designing epidemiologic studies for medical devices are shown in Table 3.6. Devices may be used in complex Table 3.6

Some issues relevant to designing epidemiologic studies of medical devices

Issue Numerous brands or models may be used on same patient Other devices or drugs could interfere with performance The environment can impact device performance Human factors are important: Attributing the cause of an adverse event can be difficult The user interface can be complex, compounded by more than one device used at the same time Device material could be critical to performance or safety Exposure to devices is not routinely recorded in medical records AMDEs are not routinely recorded in medical records

30

CH3 MEDICAL DEVICE EPIDEMIOLOGY

multi-device situations, such as for anesthesia or in the surgery suite, complicating attribution of an adverse event to one or a combination of devices (one device, or drug, could interfere with another). The presence of other devices, drugs, or comorbidities may also increase the probability of a device-related adverse event, as in the following examples: The proximal environment of an MRI machine can cause magnetic devices, such as wheelchairs, i.v. poles, and aneurism clips, to move and injure patients or other people [50] (see also Chapter 20). Diathermy in patients with deep brain stimulators, or any implanted lead, is contraindicated by the possibility of serious thermal injury [51]. ACE inhibitors can increase the likelihood of bradykinin-mediated anaphylactoid reactions related to certain dialysis membranes [52], LDL columns, or blood filters. Connections between devices may be confused (such as Luer connections [53]) or electromagnetic waves from one medical device or other electronic products may inadvertently influence the operation of another (see Chapter 20). Other environmental factors that can be associated with device problems include extreme temperatures that cause material breakdown (such as the latex used for gloves or condoms), increased humidity (such as inaccurate readings resulting from storing home blood glucose test kits in bathrooms), and radio waves (such as cell phone waves disturbing pacemaker function; see Chapter 20). Device-related adverse events may also occur because of a packaging problem, such as a tear or puncture that compromises sterility. Human factors issues are inherent in the use of virtually every device [54]. A piece of equipment, such as an infusion pump, may have a different set of controls from another one that has the same function. Conversely, two infusion pumps may look exactly the same but be set up with different internal software, including different default settings and procedures for entering the infusion orders. Circumstances such as these can lead to confusion and mistakes. Furthermore, since devices are designed, maintained, and used by people, human factors interact with device integrity in multiple ways [55]. Figure 3.2

Use-related error

Figure 3.2

?

Device failure

Difficult etiology of problems related to medical devices [74]

FEATURES OF MEDICAL DEVICES THAT ARE RELEVANT TO EPIDEMIOLOGY STUDY DESIGN

31

is a simple diagram illustrating the difficulty in ascribing ‘fault’ for a device problem to the user or the device. Human factors are clearly related to the successful use of complex devices. Figure 3.3 shows photos of ventilators that are currently in use in hospitals today. Note the profusion of controls and displays on each. FDA encourages device manufacturers to use good human factors design for new devices [56] so as to avoid errors. The importance of human factors is highlighted by the increasing use of devices in the home, where the users tend to be the least familiar with them.

Figure 3.3 Ventilators currently on the market. (A) The EspritNico2 ventilator. Reproduced by permission of Respironics, Inc. (B) The Newport HT50 Ventilator in the open position, for homecare, transport, emergency, and general purpose. Reproduced by permission of Newport Medical Instruments, Inc. (C) The Engstrom Carestation ventilator. Reproduced by permission of DatexOhmeda, Inc. (D) The SERVO-I Universal ventilator. Reproduced by permission of MAQUET, Inc.

32

CH3 MEDICAL DEVICE EPIDEMIOLOGY

In addition to the complexity of medical devices, some environments where devices are used are also complex from a human factors standpoint, because there are so many devices performing so many functions with displays and alarms that are not consistent with each other, as often happens in operating rooms (Figure 3.4) and critical care areas (Figure 3.5). The materials used in a device may be key to patients’ reactions (see Chapter 19 for a full discussion). Some materials are used in many types of devices, while one may have a choice of material when selecting a particular type of device.

Cultural features The feature of medical device use that is probably the most important to the design of epidemiologic studies is the generally poor quality of documentation of device use. Exposure to some devices, generally disposable single-use devices such as gloves, gauze, and syringes, is virtually never recorded. Exposure to equipment is generally presumed, such as infusion pumps for delivery of intravenous drugs, but the type of or specific pump is generally not noted in records. Use of short-term implants, such as catheters, is also usually documented in a general manner [57]. Permanent implants are generally well noted in the surgical chart, although the recent profusion of coded stickers provided with the equipment has resulted in confusion regarding which stickers indicate insertion and sizing tools and which indicate the permanently placed items; there are no publicly available comprehensive catalogs one can use to look up a code on a sticker to find out the device type and size. Procedure codes have been used to indicate device use [58]. For procedures that can be performed on either one or both sides of the body (such as on the eyes, hips, or knees), codes can be difficult to use because they do not indicate which side was involved. Many procedure codes are not specific to the type of device used; for example, many codes related to cardiovascular prostheses refer to animal tissue as well as artificial devices. Furthermore, one must be aware that devices may have a number of uses, some of which may be off label. An associated problem is the general lack of documentation regarding problems related to device use. Recent studies have shown that actual and potential harm related to devices is under-charted by at least an order of magnitude [29,32,57]. Furthermore, a chart review of central venous catheter use and non-infection problems found that use was more extensively documented when there was a problem, thus increasing the likelihood of study bias [57]. The lack of documentation can be especially troublesome in the common situation with medical device treatment, where a patient may see a specialist for advanced therapy and be followed by the less specialized practitioner. The specialist may not see the full range of responses to the therapy and the following practitioner may be less aware of the full implications of the choice of specialty device in relation to the patient’s progress.

Figure 3.4 Typical operating rooms in the USA. Note the high number of devices, displays, and controls. (A) Laparoscopic surgery of the head. Note the table of tools, preoperative images, laparoscopic display, and anesthesia workstation. Photograph copyright # 2006, Roselie A. Bright, reproduced with permission. (B1) Spinal surgery. Note two tables of tools and implants, preoperative images, and laptop station for monitoring the status of nerves in the back. Photograph copyright # 2006, Roselie A. Bright, reproduced with permission. (B2) The same spinal surgery. Note the cell saver, special operating table, and anesthesia workstation. Photograph copyright # 2006, Roselie A. Bright, reproduced with permission

34

CH3 MEDICAL DEVICE EPIDEMIOLOGY

Figure 3.5 Typical critical care areas in a hospital. Note the high number of devices, displays, and controls. (A) Adult critical care. Note the dialysis machine, pump pole with two intravascular infusion pumps and a feeding tube pump, array of emergency equipment arranged over the patient’s head, multiphysiologic monitor, ventilator, urinary catheter, and venous compression devices. Photograph copyright # 2006, Roselie A. Bright, reproduced with permission. (B) Neonatal critical care. All of these devices serve one neonate. Four of these stations are in one bay. Photograph copyright # 2006, Roselie A. Bright, reproduced with permission

STUDY DESIGNS FOR MEDICAL DEVICE EPIDEMIOLOGY

35

Study designs for medical device epidemiology Epidemiology is ‘the study of the occurrence of illness’ and causes that can explain that occurrence [59]. In this chapter, ‘epidemiology’ is used to refer to stand-alone descriptive studies or investigations of hypotheses. The usual types of study designs are discussed in great depth in major textbooks. This chapter will give examples of the major types, with references to medical device examples in the literature and other chapters in this book. The choice of study type depends on the study question, available resources, expense, and the existence of available data. As noted above, lack of good documentation of device use and associated problems is generally a barrier to conducting studies. Data sources that have been typically used for device epidemiology studies are: National surveys. Hospital discharge databases. Registries. Modification of a large study by adding medical device data to it. Medical device studies featuring entirely new data collection are infrequent. Data sources for medical device epidemiology are discussed further in Chapter 8.

Descriptive epidemiology studies These studies usually aim to describe the extent, character, and/or nature of use of a particular type of device. They may also include some general information about devicerelated problems. National surveys, for example, have been used to describe the prevalence of: Implants, such as heart valves [60], hips [61], pacemakers [62], intraocular lenses [63], breast implants [64], orthopedic fixation devices [65], and tympanostomy tubes [66]. Home use of devices, such as oxygen therapy [67]. Registries have been used to describe the short- and long-term experience of patients with devices, such as: Short-term effects of hemostasis devices (see Chapter 25). Long-term follow-up of pacemaker leads [68,69].

36

CH3 MEDICAL DEVICE EPIDEMIOLOGY

Health care record-based studies have been used to describe the frequency of multiple types of AMDE, for example: Visits to emergency rooms for AMDE injury [28]. Frequency of AMDEs in a tertiary care hospital [29]. Frequency of AMDEs in intensive care units [32]. Frequency of AMDEs mentioned in discharge claims [70]. Original data collection has been used to study the frequency of AMDEs in ICUs, such as the direct observation of the frequency of AMDEs in intensive care units [32].

Hypothesis-testing studies The two main types of observational hypothesis-testing study designs are ‘cohort’ (sometimes known as ‘follow-up’) and ‘case-control’. Hybrid designs incorporate elements of both. All types have been used for medical device epidemiology.

Cohort studies A major advantage of cohort studies is the ease with which they are understood. For studies of medical device use, the poor state of routine documentation gives cohort studies, particularly registries, an overwhelming advantage over other study designs. Registries can form the basis for cohort studies if they include a large number of subjects and minimize the rate of, and bias associated with, study dropouts. If all or a fixed fraction of users are enrolled as a device enters and penetrates the market, a registry has a further advantage of reflecting the nature of the users, and accumulating person–time data in direct proportion to the need-to-know of any public health threats that might be posed by the device. However, the use of registries generates problems by: Increasing the study expense, compared to ideally being able to use existing databases. Creating bias, because some practitioners and patients will refuse registration. Requiring the investigators to wait for device experience to accumulate in real time. Limiting later analysis to the data that were collected, which includes only the safety problems that the investigators anticipated. In other words, if one studies cardiac catheters and then later realizes that one also wants to study the devices used to close

SUMMARY AND RECOMMENDATIONS

37

the wound through which the catheters were passed into the body, one will have to start a brand new study. This actually happened in the instance of adverse events related to hemostasis devices (see Chapter 25). The hemostasis devices registry studies involved comparisons of hemostasis device types and brands, and other hemostasis methods. The comparison of cohorts distinguishes cohort studies from simple descriptive studies. Besides the studies of hemostasis devices, other registry-based cohort studies include: Cardiovascular registry studies (see Chapter 24). Latex student studies (see Chapter 19). Cohort studies have also been formed by using manufacturer warranty lists in the case of cochlear implants, and comparing the subsequent rates of meningitis for them and for a comparison cohort [71]. Other cohort studies have involved following patients only for the duration of their hospitalization, using discharge data [58].

Case-control studies In case-control studies, patients with a particular adverse outcome are compared to other patients regarding the use of particular medical devices (e.g. extended wear of contact lenses [72]). This type of study can be harder to conduct well because a control group appropriate for comparison to the case group must be identified. The main advantage of a case-control study is lower expense.

Hybrid studies Another option that is more and more commonly used is a hybrid between the cohort and case-control designs. One option is the case-cohort, where cases of adverse outcome are identified within a larger cohort of exposed subjects (e.g. extended wear of contact lenses [73], and meningitis associated with cochlear implant [71]). This option allows a combination of the advantages of both follow-up (identifying subjects on the basis of exposure) and case-control (identifying outcome status for all cases and just a sample of controls) designs.

Summary and recommendations Each field of health care delivery relies heavily on a large range of types of medical devices. Studies have shown both the importance of medical device epidemiology and the need to develop the capability for rigorous epidemiologic studies. Many factors

38

CH3 MEDICAL DEVICE EPIDEMIOLOGY

influence the successful use and study of medical devices: the range of types of medical devices, human factors, and various intrinsic features of devices. All types of epidemiologic study designs have been successfully used to study medical devices. However, widespread study has been hampered by the lack of existing databases that include medical device use descriptors. Quality of documentation varies by medical device type and is generally much poorer for devices than it is for drugs. Proper device use documentation would enable: The creation of automated datasets for device epidemiology. Further development of device epidemiology methods. Participation of academia in device epidemiology studies. More complete reporting of medical devices. Better information on factors that may affect adverse outcomes. These, in turn, would lead to: Improved studies of the importance of medical devices to patient safety, followed by: Projects to increase the safety of medical device use. The capability to quickly and validly study medical device epidemiology will allow the patient safety community to understand and address adverse medical device events, which are a substantial proportion of all medical adverse events.

References 1. United States Congress. Federal Food, Drug, and Cosmetic Act. Section 201 (321). Definitions (h). 1998. 2. Kohn L, Corrigan J, Donaldson M. To Err is Human: Building a Safer Health System, 1st edn. Washington, DC: National Academy Press; 2000. 3. Rothschild JM, Keohane CA, Cook EF et al. A controlled trial of smart infusion pumps to improve medication safety in critically ill patients. Crit Care Med 2005; 33(3): 533– 540. 4. Thomas EJ, Studdert DM, Burstin HR et al. Incidence and types of adverse events and negligent care in Utah and Colorado. Med Care 2000; 38(3): 261–271. 5. Thomas EJ, Brennan TA. Incidence and types of preventable adverse events in elderly patients: population-based review of medical records. BMJ 2000; 320(7237): 741–744. 6. What patient safety lessons can health care leaders learn from aviation safety? Healthc Leadersh Manag Rep 2002; 10(2): 8–9.

REFERENCES

39

7. Gaba DM, Singer SJ, Sinaiko AD, Bowen JD, Ciavarelli AP. Differences in safety climate between hospital personnel and naval aviators. Hum Factors 2003; 45(2): 173–185. 8. Hudson P. Applying the lessons of high risk industries to health care. Qual Saf Health Care 2003; 12(suppl 1): i7–12. 9. Johnson CW. How will we get the data and what will we do with it then? Issues in the reporting of adverse healthcare events. Qual Saf Health Care 2003; 12 (suppl 2): ii64–67. 10. Randell R. Medicine and aviation: a review of the comparison. Methods Inf Med 2003; 42: 433–436. 11. Wilf-Miron R, Lewenhoff I, Benyamini Z, Aviram A. From aviation to medicine: applying concepts of aviation safety to risk management in ambulatory care. Qual Saf Health Care 2003; 12: 35–39. 12. Brennan TA, Leape LL, Laird NM et al. Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med 1991; 324(6): 370–376. 13. Wilson RM, Runciman WB, Gibberd RW, Harrison BT et al. The Quality in Australian Health Care Study. Med J Aust 1995; 163: 458–471. 14. Wilson RM, Harrison BT, Gibberd RW, Hamilton JD. An analysis of the causes of adverse events from the Quality in Australian Health Care Study. Med J Aust 1999; 170(9): 411–415. 15. Brennan TA, Localio AR, Leape LL et al. Identification of adverse events occurring during hospitalization: a cross-sectional study of litigation, quality assurance, and medical records at two teaching hospitals. Annals Intern Med 1990; 112: 221–226. 16. Beydon L, Conreux F, Le Gall R, Safran D, Cazalaa JB. Analysis of the French health ministry’s national register of incidents involving medical devices in anaesthesia and intensive care. Br J Anaesth 2001; 86(3): 382–387. 17. Field M, Tilson H (eds). Safe Medical Devices for Children. Washington, DC: The National Academies Press, 2005. 18. O’Shea JC, Kramer JM, Califf RM, Peterson ED. Part I: Identifying holes in the safety net. Am Heart J 2004; 147(6): 977–984. 19. Small SD. Medical device-associated safety and risk: surveillance and strategems. JAMA 2004; 291(3): 367–370. 20. Chittock DR, Dhingra VK, Ronco JJ et al. Severity of illness and risk of death associated with pulmonary artery catheter use. Crit Care Med 2004; 32(4): 911–915. 21. Chugh A, Scharf C, Hall B et al. Prevalence and management of inappropriate detection and therapies in patients with first-generation biventricular pacemaker-defibrillators. Pacing Clin Electrophysiol 2005; 28(1): 44–50. 22. Pratt TR, Pulling CC, Stanton MS. Prospective postmarket device studies versus returned product analysis as a predictor of system survival. Pacing Clin Electrophysiol 2000; 23(7): 1150–1155. 23. Rosenqvist M, Beyer T, Block M, den Dulk K et al. Adverse events with transvenous implantable cardioverter-defibrillators: a prospective multicenter study. European 7219 Jewel ICD investigators. Circulation 1998; 98(7): 663–670. 24. Gill G, Mills D, Joshi A. Mortality following primary total knee arthroplasty. J Bone Joint Surg Am 2003; 85: 432–435. 25. Parvizi J, Sullivan T, Trousdale R, Lewallen D. Thirty-day mortality after total knee arthroplasty. J Bone Joint Surg Am 2001: 1157–1161. 26. Taylor H, Dennis D, Crane H. Relationship between mortality rates and hospital patient volume for Medicare patients undergoing major orthopedic surgery of the hip, knee, spine, and femur. J Arthroplasty 1997; 12: 235–242. 27. Lin L, Vicente KJ, Doyle DJ. Patient safety, potential adverse drug events, and medical device design: a human factors engineering approach. J Biomed Inform 2001; 34(4): 274–284.

40

CH3 MEDICAL DEVICE EPIDEMIOLOGY

28. Hefflin B, Gross T, Schroeder T. Estimates of medical device-associated adverse events from emergency departments. Am J Prev Med 2004; 27(3): 246–253. 29. Samore MH, Evans RS, Lassen A et al. Surveillance of medical device-related hazards and adverse events in hospitalized patients. JAMA 2004; 291(3): 325–334. 30. Health Care and Cost Utilization Project (HCUP): http://www.ahrq.gov/data/hcup/ [accessed October 2005]. 31. Bright R, Shen J. Use of a free, publicly-accessible data source to estimate hospitalizations related to adverse medical device events. FDA 2006 Science Forum poster abstract: http://www. accessdata.fda.gov/scripts/oc/scienceforum/sf2006/search/preview.cfm?keyword¼Bright& abstract_id¼997&type¼author&backto¼search [accessed April 2006]. 32. Samore M, Anderson S, Wiessner P et al. Medical device problems in intensive care units: dangers and diversity of types (manuscript cleared for publication by CDRH, 2005). 33. Frey B, Kehrer B, Losa M et al. Comprehensive critical incident monitoring in a neonatalpediatric intensive care unit: experience with the system approach. Intensive Care Med 2000; 26(1): 69–74. 34. Stambouly JJ, McLaughlin LL, Mandel FS, Boxer RA. Complications of care in a pediatric intensive care unit: a prospective study. Intensive Care Med 1996; 22(10): 1098– 1104. 35. Bates DW, Leape LL, Petrycki S. Incidence and preventability of adverse drug events in hospitalized adults. J Gen Intern Med 1993; 8(6): 289–294. 36. Hanesse B, Legras B, Royer R, Guillemin F, Briancon S. Adverse drug reactions: comparison of two report methods. Pharmacoepidemiol Drug Safety 1994; 3: 223–229. 37. Bates DW, Cullen DJ, Laird N et al. Incidence of adverse drug events and potential adverse drug events. Implications for prevention. ADE Prevention Study Group. JAMA 1995; 274(1): 29–34. 38. Chaudhry SI, Olofinboba KA, Krumholz HM. Detection of errors by attending physicians on a general medicine service. J Gen Intern Med 2003; 18(8): 595–600. 39. Carbonin P, Pahor M, Bernabei R, Sgadari A. Is age an independent risk factor of adverse drug reactions in hospitalized medical patients? J Am Geriatr Soc 1991; 39(11): 1093–1099. 40. Rothschild JM, Landrigan CP, Cronin JW et al. The Critical Care Safety Study: the incidence and nature of adverse events and serious medical errors in intensive care. Crit Care Med 2005; 33(8): 1694–1700. 41. Hennesy S, Carson JL, Ray WA, Strom BL. Medicaid databases. In Pharmacoepidemiology, 4th edn, Strom BL (ed.). Chichester: Wiley, 2005. 42. Chan KA, Davis RL, Gunter MJ et al. The HMO research network. In Pharmacoepidemiology, 4th edn, Strom BL (ed.). Chichester: Wiley, 2005. 43. Strom BL, Melmon KL. The use of pharmacoepidemiology to study beneficial drug effects. In Pharmacoepidemiology, 4th edn, Strom BL (ed.). Chichester: Wiley, 2005. 44. Downey W, Stang M, Beck P, Osei W, Nichol JL. Health services databases in Saskatchewan. In Pharmacoepidemiology, 4th edn, Strom BL (ed.). Chichester: Wiley, 2005. 45. Leufkens HG, Urquhart J. Automated pharmacy record linkage in The Netherlands. In Pharmacoepidemiology, 4th edn, Strom BL (ed.). Chichester: Wiley, 2005. 46. Gelfand JM, Margolis DJ, Dattani H. The UK general practice research database. In Pharmacoepidemiology, 4th edn, Strom BL (ed.). Chichester: Wiley, 2005. 47. Product Code Classification Database, June 15 2005: http://www.fda.gov/cdrh/prodcode.html [accessed November 2005]. 48. Universal Medical Device Nomenclature SystemTM (UMDNS): http://www.ecri.org/Products_ and_Services/Products/UMDNS/Default.aspx [accessed November 2005]. 49. Reuse of single-use devices: http://www.fda.gov/cdrh/reuse/ [accessed October 2005].

REFERENCES

41

50. A primer on medical device interactions with magnetic resonance imaging systems. US Food and Drug Administration: http://www.fda.gov/cdrh/ode/primerf6.html 51. FDA Public Health Notification: diathermy interactions with implanted leads and implanted systems with leads. US Food and Drug Administration: http://www.fda.gov/cdrh/safety/ 121902.html 52. Verresen L, Fink E, Lemke H, Vanrenterghem Y. Bradykinin is a mediator of anaphylactoid reactions during hemodialysis with AN69 membranes. Kidney Int 1994; 45(5): 1497–1503. 53. Eakle MR, Gallauresi BA, Morrison A. Luer-lock misconnects can be deadly. Nursing 2005; 35(9): 73. 54. Sawyer D. Do it by design. An introduction to human factors in medical devices: http:// www.fda.gov/cdrh/humfac/doitpdf.pdf [accessed November 2005]. 55. Case and commentary: the wrong channel. Agency for Health Care Research and Quality (AHRQ) Morbidity and mortality rounds on the web: http://www.webmm.ahrq.gov/case. aspx?caseID¼105 [accessed December 2005]. 56. Sawyer D. CDRH’s approach to providing human factors information, December 1996: http:// www.fda.gov/cdrh/humfac/ [accessed November 2005]. 57. Bright R, Mermel L, Richards C, Eakle M, Yoder D. Mechanical and allergic adverse events related to central vascular catheters: epidemiology in the Medicare in-hospital population, 2002. Pharmacoepidemiol Drug Safety 2004; 13(s1): s293–294. 58. Astor B, Kaczmarek RG, Daley WR. Mortality after aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000; 70: 1939–1945. 59. Rothman K, Greenland S (eds). Modern Epidemiology, 2nd edn. Philadelphia, PA: LippincottRaven, 1998. 60. Garver D, Kaczmarek RG, Silverman BG, Gross TP, Hamilton PM. The epidemiology of prosthetic heart valves in the United States. Tex Heart Inst J 1995; 22(1): 86–91. 61. Sharkness CM, Hamburger S, Moore RM Jr, Kaczmarek RG. Prevalence of artificial hips in the United States. J Long Term Eff Med Implants 1992; 2(1): 1–8. 62. Silverman BG, Gross TP, Kaczmarek RG, Hamilton P, Hamburger S. The epidemiology of pacemaker implantation in the United States. Public Health Rep 1995; 110(1): 42–46. 63. Sharkness CM, Hamburger S, Kaczmarek RG, Hamilton PM et al. Racial differences in the prevalence of intraocular lens implants in the United States. Am J Ophthalmol 1992; 114(6): 667–674. 64. Bright RA, Jeng LL, Moore RM Jr. National survey of self-reported breast implants: 1988 estimates. J Long Term Eff Med Implants 1993; 3(1): 81–89. 65. Moore RM Jr, Bright RA, Jeng LL, Sharkness CM et al. The prevalence of internal orthopedic fixation devices in children in the United States, 1988. Am J Public Health 1993; 83(7): 1028– 1030. 66. Bright RA, Moore RM Jr, Jeng LL, Sharkness CM et al. The prevalence of tympanostomy tubes in children in the United States, 1988. Am J Public Health 1993; 83(7): 1026–1028. 67. Silverman BG, Gross TP, Babish JD. Home oxygen therapy in Medicare beneficiaries, 1991 and 1992. Chest 1997; 112(2): 380–386. 68. Furman S, Benedek Z. Survival of implantable pacemaker leads. The Implantable Lead Registry. Pacing Clin Electrophysiol 1990; 13: 1910–1914. 69. Moller M, Arnsbo P. Appraisal of pacing lead performance from the Danish Pacemaker Register. Pacing Clin Electrophysiol 1996; 19: 1327–1336. 70. Bright R, Shen J. Use of a free, publicly-accessible data source to estimate hospitalizations related to adverse medical device events (draft manuscript, 2005). 71. Reefhuis J, Honein MA, Whitney CG et al. Risk of bacterial meningitis in children with cochlear implants. N Engl J Med 2003; 349: 435–445.

42

CH3 MEDICAL DEVICE EPIDEMIOLOGY

72. Schein OD, Glynn RJ, Poggio EC, Seddon JM, Kenyon KR. The relative risk of ulcerative keratitis among users of daily-wear and extended-wear soft contact lenses. A case-control study. Microbial Keratitis Study Group. N Engl J Med 1989; 321: 773–778. 73. Poggio EC, Glynn RJ, Schein OD et al. The incidence of ulcerative keratitis among users of daily-wear and extended-wear soft contact lenses. N Engl J Med 1989; 321: 779–783. 74. Kaye R (personal communication, 2003).

4 Surveillance of adverse medical device events Roselie A. Bright US Food and Drug Administration, Rockville, MD, USA

Introduction In this chapter, I use ‘surveillance’* to mean ongoing ‘just looking’ or ‘monitoring’ for adverse events (AE), as opposed to ‘epidemiology’{, meaning research with a predefined objective. There is a multitude of surveillance methods. Data collection tools for surveillance include anecdotes, surveys, standard data forms, and use of existing data. Anecdotes may take the form of adverse event reports (voluntary or mandatory) or published case reports or case series. Surveys may be conducted by any organization and may be designed to question healthcare providers or patients. Standard data forms can be used for recording adverse event reports, survey responses, ongoing care of particular types (such as for registries), or abstraction of existing data. Existing data could be original paper or electronic healthcare records or healthcare financial claims. Data may come from convenience samples or random samples, or the population.

*‘

Surveillance is a continuous and systematic process of collection, analysis, interpretation, and dissemination of descriptive information for monitoring health problems . . . [B]roadly, surveillance provides a basis for shaping public health policy.’[48] { Epidemiology is ‘the study of the occurrence of illness’ and causes that can explain that occurrence (Rothman K, Greenland S. Measures of disease frequency. In Modern Epidemiology, 2nd edn, Rothman KJ, Greenland S (eds). Philadelphia, PA: Lippincott-Raven, 1998.

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

44

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

Rationale for surveillance Surveillance goals The main goal of medical device surveillance is to obtain warnings of previously unidentified adverse medical device events (AMDEs) and to detect patterns of actual or potential AMDEs. An ideal surveillance program should be based on epidemiologic principles, so that inferences can be made about the specific, overall, and relative public health burdens of different types of AMDE. This population-based knowledge would form the basis of more effective efforts to mitigate and prevent AMDEs. A program of surveillance systems should ideally meet all of the following goals simultaneously: Detect rare or unexpected AMDEs. Find problems in ‘real’ users with multiple comorbidities (including vulnerable populations) in ‘real world’ settings (including many years after initial device exposure). Premarketing studies are usually done by experts in the field, using patients who generally have only the condition to be addressed by the medical device. However, once a device is on the market, users may not be expert in use of the device and patients may have many other comorbidities or other complicating factors. Have complete capture of AMDEs and reliable information on device exposure, including the specific nature of the device, brand and model number. Allow full appreciation of the public health burden imposed by AMDEs of specific types or related to specific device types. These goals combine the purposes of surveillance with the quality of data required to fulfill the purposes. Individual surveillance systems could be measured against these goals. Their selection into a program of surveillance systems could be based on their ability to advance the program towards meeting the goals. In practice, the strategy would be to add systems that have complementary strengths and weaknesses. One should base the surveillance program on a variety of data collection methods that capture information specific to a variety of device types, use settings, and users. Table 4.1 lists the current FDA AMDE surveillance programs and whether they meet the goals.

Why monitor marketed medical devices? In spite of governments’ regulatory efforts and the best intentions of healthcare providers, things may go awry for any of the following reasons:

X X

X

X

Detect unexpected AMDEs Find problems in ‘real’ users with multiple comorbidities in ‘real-world’ settings Complete capture of AMDEs Reliable data on device exposure Allow calculation of the public health burden imposed by AMDEs of specific natures Allow calculation of the public health burden imposed by AMDEs related to specific device types

X

X

Detect rare AMDEs

Voluntary passive reporting Any healthcare setting

Mandatory passive reporting Any healthcare setting

MedWatch (see Chapter 2)

Type of system Source of AMDE data

MDR, ASR, UFR (see Chapter 2)

Surveillance system name

X

X

Sentinel passive reporting Volunteer hospitals and nursing homes

MedSun (see Chapter 5)

The existing FDA AMDE surveillance program: component surveillance systems and their properties

System characteristic

Table 4.1

X

X

X

X

Active surveillance Statistical sample of emergency departments

NEISS (see Chapter 6), potential capability

46

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

Human beings, thus human error, are involved in every stage of healthcare and regulation of medical products. Practice of medicine, including off-label device use (unregulated by FDA), is a factor [1]. The decision to use devices in ways that are not included on the official device label is considered ‘practice of medicine’ and is not, on its own, considered illegal. Standards of care change over time [1]. New discoveries and theories of care are constantly introduced into the practice of medicine. As medicine advances, the standard of care of the past becomes the substandard care of the present. Premarket studies that provided reasonable assurance that a medical device is safe and effective may not have anticipated various problems that could occur in the postmarket period [2], because the premarket study may have: *

Been too small to capture uncommon adverse events.

*

Included insufficient numbers of subjects from especially vulnerable subgroups to adequately assess safety and effectiveness in these groups.

*

Been conducted in an environment that is in some ways unlike the ‘real-world’ environment of the postmarket period (e.g. providers utilized in the premarket clinical trials may be better trained and skilled than typical providers in the ‘real world’).

*

Been of insufficient duration to capture problems related to device (e.g. implants) failure that occur over a long period of time.

These reasons are not exclusive to medical devices but are generally the same as for other medical products.

Types of device problems and established FDA tools FDA has tools established by law (see Chapter 2 for details) for addressing several kinds of medical device-related problems: Prevention of common and known problems – Labeling requirements (including indications, intended use, directions for use, and warnings), Quality System regulations (these apply to device design, the manufacturing process, monitoring, and improvement), and response to user reports of device malfunctions or AMDEs [3]. Immediate concern that arises before approval – Annual Reports or Post-Approval Studies (both can be required by CDRH upon approval of the device) [4].

SURVEILLANCE BASED ON ADVERSE EVENT REPORTS

47

Uncommon, unknown, or ‘rare’ event – AMDE reports, either submitted to FDA or collected in other locations, such as by FDA inspection of manufacturers’ files [5,6]. Long-term concern or issue that arises in the postmarket period – Postmarket Surveillance Studies (Section 522) [7] and ‘Controls’, which are applied by FDA to particular cleared devices. ‘Special controls are defined in section 513(a)(1)(B) of the Act as those controls, such as performance standards, postmarket surveillance, patient registries, development and dissemination of guidelines, recommendations and other appropriate actions that provide reasonable assurance of the device’s safety and effectiveness’ [8]. Any problem – ad hoc epidemiology studies. Recent laws have increased FDA regulatory responsibility in the postmarket period. Furthermore, most device experience is postmarket, and device problems can be catastrophic. The only tool established by law for surveillance of unknown or unanticipated problems is the collection of AMDE reports; even though section 522 is labeled ‘surveillance’, under the definitions used for this chapter, it is an authority to require specific (including epidemiology) investigations rather than an authority to require surveillance.

Surveillance based on adverse event reports Around the world, the primary basis of AMDE surveillance continues to be adverse event reports. The barriers to reporting and incompleteness of reports are discussed below, followed by evidence of underreporting and programs to enhance reporting.

Barriers to reporting There are many barriers to reporting AMDEs to FDA [9], which can be divided by reporting step: recognition, reporting within the institution, and reporting to FDA. First, there are barriers to recognizing a potential relationship between a medical device and an AMDE. A few reasons for this lack of recognition include that: the AMDE can be reasonably explained by other causes; the AMDE is a common condition; there was a time delay from device use to the AMDE; or the AMDE occurred in an organ system different from the one being treated with the device. Second, there are many barriers to reporting an AMDE once the potential relationship between the medical device and adverse event is recognized. Some reasons are related to the seeming triviality of the AMDE, such as the AMDE having resolved or already being ‘known’ (i.e. already listed in the label or otherwise publicized). Furthermore, the healthcare provider could be very busy, assume that others have already reported this event, may not see that reporting would be useful, could be concerned about being blamed for the AMDE, or may be unaware of the FDA medical device reporting program.

48

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

In addition, an individual provider may be collecting a series of cases for publication, and therefore hold off reporting. One of the most significant barriers specific to medical devices is the general lack of recognition of medical devices as products related to AEs. In studies of AEs, AMDEs were often not explicitly reported in a medical device category [10–13]. In a study by Frey et al. [14], only one category of medical care problems was specified as being device-related (‘equipment dysfunction’, 15 problems), yet many (36 of 134) of the problems in the other non-drug categories (‘management/environment’, ‘procedures’, ‘respiration’, and ‘nosocomial infections’) that were described in the text related to devices. Another significant barrier to reporting is that documentation of device use is often missing from patient records [15–18]. When documentation is present, the lack of a standard identification system for devices (addressed in Chapter 7) hampers understanding of the AMDE. There are many other barriers to recognizing and reporting AMDEs that are specifically related to the involvement of medical devices. Instructions for device use are generally written in medical jargon for healthcare providers and are difficult for lay users or patients to understand and follow. Other contributors to AMDEs that complicate reporting include adverse interactions between the device and other therapies and complex multi-device situations. For diagnostic devices, it can be difficult to recognize false-positive and false-negative results. Re-used devices can present their own problems: devices manufactured for single use may be reprocessed for further use, while devices meant for multiple uses are refurbished and may get replacement parts made by other manufacturers; it can be difficult to understand what went wrong. A major barrier to reporting is that devices are most often thought to injure as a result of device failure or ‘user error’. However, human factors analysis and patient safety research has revealed difficulties with trying to assign one or the other of these causes; for example, design flaws make error-free use difficult [19–29] and poor maintenance can lead to device failure [30] (see also Chapter 17). Infusion pumps and defibrillators have drawn particular attention by researchers in the human factors field for being difficult to use successfully [31–35]. Another area where anesthesiologists, epidemiologists, and human factors engineers (including from FDA) have worked together and where continuous improvement has occurred in deliberate cycles is anesthesia safety [36]. The complexity of determining the root causes of AMDEs has led the human factors team at FDA to advocate using ‘use-related error’ (rather than ‘user error’) as a blame-neutral term [37]. To illustrate the current ease with which users tend to be blamed for AMDEs, note the excerpt from the UK web page in Figure 4.1. All the items in the figure put the onus on the user to alter behavior, rather than on the manufacturer to improve the usability of devices. The blame inherent in the term ‘user error’ is important because AMDEs are less likely to be reported by the user if the user rather than the device is seen as being ‘at fault’ [29,38].

Incompleteness of reports Adverse medical device event reports also commonly suffer from the following deficiencies:

SURVEILLANCE BASED ON ADVERSE EVENT REPORTS

Figure 4.1

49

A ‘one-liner’ from the UK [64] which places all responsibility for problems on the user

50

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

Few or no data on the extent of device use. The nature of AMDE reports emphasizes cases of adverse outcome and ignores instances of successful use. There is no inherent mechanism for reporting the total amount of both successful and unsuccessful device use. Limited coded or free text information [39]. The report narratives submitted by manufacturers and other reporters are frequently incomplete and may not sufficiently detail underlying problems. Many significant problems are discovered and addressed by the manufacturers well before FDA recognizes the problem or takes action (e.g. the Sulzer hip discussed in Chapter 29). Conversely, manufacturers may not sufficiently address problems with their products until FDA also becomes aware of them, for example, cochlear implant hermeticity problems (see Chapter 26) and esophageal bulking agent migration problems [40]. Inadequate product group categorizationz. For some product classes (e.g. multi-device systems, etc.), difficulty inferring which product seems to be associated with the AMDE. All these deficiencies contribute to difficulties in analyzing adverse events. Data-mining tools are being developed to address some of these problems (see Chapter 8).

Underreporting of adverse medical device events Most reports are from manufacturers and are based on reports from their sales staff or product complaints they have received from their customers (who may be healthcare providers or patients). Relatively few reports come to FDA directly from healthcare providers or patients, despite renaming and publicizing the voluntary reporting program as ‘MedWatch’ [41]. Healthcare facilities may feel no tie to FDA and mistakenly believe that FDA regulates their every move, although FDA’s only regulatory functions with respect to providers regard User Facility Reporting and the re-use of single-use devices [42]. A study [43] done in the 1980s by the General Accounting Office (charged by the US Congress with investigating how well US federal agencies administer federal law) found that less than 1% of hospital AMDEs were reported to FDA. Since then, underreporting of AMDEs has remained a major problem, as documented in recent FDA-funded studies:

z‘Product Codes’ are also known as ‘procodes’. They are three-letter codes assigned to devices when they are cleared or approved by FDA. Assignments have not been systematic, leading to problems: procodes may contain more than one generic device group; devices that should be grouped under one procode are distributed among more than one; and for some procodes, the types of devices intended to be included are unknown. These difficulties are being addressed by the new nomenclature, discussed in Chapter 7.

SURVEILLANCE BASED ON ADVERSE EVENT REPORTS

51

Hefflin et al. [44] (described in further detail in Chapter 6) found that 58 000 of the patients with device-associated injuries who came to the emergency department in a 1 year period (July 1999–June 2000) died or were hospitalized; even though distinguishing site of injury is difficult in the FDA AMDE report database, it is unlikely that most of these 58 000 injuries were reported to FDA, since the total number of reports to FDA, including serious injuries, deaths and malfunctions, were just 90 000 in 1999 and 99 000 in 2000. Samore et al. [15] tried to find evidence of AMDEs in the records of a tertiary hospital. They found that more AMDEs were recorded in computerized patient records (17-fold) and discharge codes (40-fold) than were recorded in the hospital’s AE database. There was little overlap between data sources for types of events that should have been detectable by more than one method, and there was no other way to assess the completeness of capture. That research group therefore embarked on a direct observation study to define the incidence and spectrum of problems associated with the use of medical devices in intensive care units (chosen because of the high frequency of device use and the particular vulnerability of the patients) [16]. *

They reported that the directly observed rate of AMDEs during intensive care unit stays was 194 times higher than that found by checking the incident reports log.

*

The rate of reporting of AMDEs by ICU staff directly to the research staff was higher than the rate of standard incident reporting and chart documentation, but much less than the directly observed rate.

A study of a national sample of hospital discharge diagnoses related to AMDEs (using the same list of codes used by Samore et al.) estimated that there were 820 000–1 100 000 such discharges per year in the USA during 1997–2003 [45], compared to a high total in 1994 from any source of 87 000 serious injury or death reports to FDA’s reporting system during the same time period (after 1996, the majority of reporting from manufacturers was under the Alternative Summary Reporting program). Even MedSun (see below, and Chapter 5) has suffered severe underreporting. Since the beginning of MedSun, the program has received 587 reports of deaths and serious injuries and 3929 reports of malfunctions through October 2005. If 300 of the participating hospitals reported at a rate similar to the discharge claim rate found in the Samore 2004 study for one tertiary care hospital (1000 adverse medical device events/year), MedSun would receive 300 000 reports/year. Limited study outside the USA has also demonstrated underreporting [46]. Underreporting is such a significant problem that, even if steps are taken to increase reporting, independent sources of information on AMDEs are critical.

52

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

Enhanced adverse event reporting MedSun is the Medical Product Surveillance Network, set up to enhance reporting of AMDEs and potential AMDEs to FDA. It was mandated in the FDA Modernization Act of 1997, which said that the current User Facility Report system (unique to CDRH) shall eventually be replaced by one that is limited to a ‘subset of user facilities that constitutes a representative profile of user reports’. It is the subject of Chapter 5, so is described only briefly here. A few hundred hospitals participate; each has one or two MedSun liaisons, who are generally the Risk Manager or Biomedical Engineer. These liaisons file AMDE reports with a MedSun contractor, who in turn works with the liaisons to make sure the reports are as detailed as possible. In summary, it is possible for MedSun to detect unexpected AMDEs, but AMDEs in general are probably underreported within the participating facilities. Currently, there is no way to know whether the reported problems are a fair representation of those that occur.

Surveillance based on registries Registries are often mentioned in the context of surveillance for AMDEs because they provide a solution to the problem of routine records not including sufficient information about device exposure. Registries must be set up device by device. If the outcomes are a predetermined set of adverse events, the registry forms the basis of an observation study. If a comparison followed group is also available, the investigation is also a hypothesistesting study. These instances are discussed in Chapter 3. If the registry allows the capture of unanticipated AMDEs, the investigation may be thought of as a form of surveillance; a recent example is a new registry for mechanically assisted circulatory support [47]. The disadvantages of using a registry to conduct surveillance are: Expense. Bias, due to some practitioners and patients refusing registration. Having to wait for device experience to accumulate in real time. This can be a problem if the device has already been in use. However, if the device is new, having the registry to capture safety information in a timely manner is a major advantage. Limiting later analysis to the data that were collected, which include only the exposures and covariates that the investigators had thought to collect. In other words, if one studies cardiac catheters and then later realizes that one also wants to study the devices used to close the wound through which the catheters were passed into the body, one will have to start new data collection for that device. This actually happened in the instance of adverse events related to hemostasis devices (described further in Chapter 25).

ACTIVE SURVEILLANCE

53

The most substantial challenge to successful registry-based surveillance is incomplete follow-up of registrants. Proper device use documentation would enable FDA to encourage the development of better device surveillance methods. This, in turn, would lead to the creation of automated datasets for device surveillance and the participation of academia in device surveillance studies. Proper device use documentation would also ease studies of the importance of medical devices to patient safety and then lead to projects to increase the safety of medical device use. Proper documentation of device use and related problems is necessary for the development of the fields of medical device-related patient safety, as well as adequate medical device surveillance systems.

Active surveillance The surveillance goals listed at the beginning of this chapter included complete capture of AMDEs, reliable information on device exposure, and calculation of the public health burden imposed by AMDEs of specific types or related to specific device types. These goals require collecting data that was recorded during routine patient care, or ‘active surveillance’. In ‘active’ surveillance of any type of adverse event (AE), someone systematically looks for AEs to add to the AE database; this may be accomplished by systematic solicitation of care providers for reports, systematic searching of records for evidence of AEs, or systematic wholesale downloading of primary patient records [48]. This active data collection stands in contrast to ‘passively’ waiting for reports of AEs, which is defined as ‘passive surveillance’ [48]. An example is the model provided by research on methods for surveillance of adverse drug events (ADEs). The scientific literature is quite well developed on the topics of ADE descriptions, development of drug surveillance systems, and prevention of ADEs; these are a major focus of the US Institute of Medicine report, To Err is Human [49]. A number of other studies have also found that soliciting actual or potential adverse events of all types yields more reports than routine incident reporting systems [50,51] or criteria-driven record review [52,53]. Some studies found that solicitation did not yield substantially higher numbers than other methods, but did reveal significant numbers that were not otherwise found [53]. Between three major studies [54–56], four progressively more active strategies were used to find indicators of ADEs: Reports solicited mechanically via easily accessible and numerous computer workstations or logbooks set up for reporting. Reports solicited personally by study staff. Chart review for indicators. Computer monitoring of indicators in charts or logs.

54 Table 4.2

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

Results of studies comparing four methods of discovering adverse drug events (ADEs) Study Classen et al. [54] (% of events)

ADE collection strategy Mechanical solicitation Personal solicitation Chart review for indicators Computer monitoring of indicators Total number of events

1

Cullen et al. [55] (% of events) 7 15 98

99 731

55

Jha et al. [56] (% of events)

4 65 45 617

The main results of the studies, shown in Table 4.2, demonstrated that systematic review of written or computerized records is far superior to solicited reports. As already suggested in the preceding paragraph, the ‘active’ vs. ‘passive’ nature of a surveillance system is a relative, rather than an absolute, concept. For example, MedSun has features of both an active and a passive surveillance system. One active feature of the MedSun program is its intense educational efforts. A passive aspect of the program is the fact that facility representatives wait for reports from within their facilities that are then reported to FDA. If the facilities actively searched their patient records for evidence of device-related problems their program would be more active, and therefore the MedSun system would capture AMDEs more completely. A surveillance program that does actively search patient records for AMDEs is the National Electronic Injury Surveillance System (NEISS), operated by the Consumer Product Safety Commission and described further in Chapter 6. NEISS is based on all the injury visits to a statistical sample of emergency departments. Reliable information on device exposure would require documentation of device exposure on a routine basis. This has been problematic, as shown separately by Samore et al. [15], in research in progress on central vascular catheters [17], and by direct observation of device use in hospital intensive care units [16]. Active surveillance that ensures complete capture of AMDEs and reliable information on device exposure for the entire population or a statistical sample of the population also would allow the calculation of the public health burden imposed by AMDEs of specific types or related to specific device types. NEISS (Chapter 6) is under development to try to address these goals.

Necessary conditions for effective surveillance Two things are necessary for effective surveillance of medical device AMDEs: A rational system of unique codes for devices, such as the Unique Medical Device Identification Code [57] that is currently under development.

IDEAL AMDE SURVEILLANCE PROGRAM

55

Routine documentation in patient records of device use and problems associated with device use. Both surveillance and epidemiology of medical devices will be vastly improved by a good coding system and routine documentation. If the device codes are placed on devices and device users recognize that the code describes the device, using the code in documentation and reporting will greatly enhance understanding of device-related communication. Routine documentation of device use and device-related problems will allow the possibility of conducting surveillance based on routine records. It has already been well established in other areas of epidemiology, such as drug epidemiology, that using available records decreases the cost of studies tremendously and also has other methodological advantages. One of the reasons drug epidemiology is well advanced is the existence and use of the National Drug Code in the USA. Another important reason is that healthcare providers are accustomed to meticulous documentation of drug prescriptions and administration. Drug epidemiology is now a robust field, funded by manufacturers, insurers, and the US National Institutes of Health because FDA funded the development of the field of drug epidemiology in the 1960s–1980s. Patient safety related to drugs began to be funded by a US health agency (founded as the Agency for Healthcare Policy and Research in 1989 [58] and later renamed the Agency for Healthcare Research and Quality [59]) and now forms a large part of the patient safety movement. Device epidemiology, on the other hand, is still primitive because FDA began to regulate devices relatively late, after the time of large US federal budgets for development of regulatory resources. The state of the infrastructure for medical device epidemiology is even less impressive in other countries.

Ideal AMDE surveillance program The ideal national surveillance program would be a set of systems (see Table 4.3) that each offer unique advantages and complement each other to form a comprehensive program. The important features to consider are the abilities of each system to meet the surveillance goals, as well as cost. Each component system should have an epidemiologic basis and the program as a whole should cover the entire national healthcare experience. The system would be based on routine and precise documentation of device use and device-related problems. The frequency of device use and AMDEs is so high that comprehensive national data collection is not feasible; statistical sampling is a practical approach. A model for this is the program used by the Consumer Product Safety Commission, which is composed of three complementary components: systematic AMDE collection from a random sample of emergency departments (NEISS); voluntary reports; and ad hoc investigations [60]. An Institute of Medicine report on injury surveillance recommended this model for other government injury surveillance programs [61]. For devices, statistical sampling of all types of healthcare organizations should be used to obtain the full spectrum of device use and AMDEs. Since there may be

56

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

Table 4.3 An ideal national AMDE surveillance program: component surveillance systems and their properties Surveillance system name System characteristic

Ideal voluntary reporting, such as MedWatch

Type of system Source of AMDE data

Voluntary passive reporting Any healthcare setting

Detect rare AMDEs Detect unexpected AMDEs Find problems in ‘real’ users with multiple comorbidities in ‘real-world’ settings Complete capture of AMDEs Reliable data on device exposure Allow calculation of the public health burden imposed by AMDEs of specific natures Allow calculation of the public health burden imposed by AMDEs related to specific device types

X X X

Ideal programs based on statistical samples of healthcare providers Active surveillance Separate statistical samples of hospitals, nursing homes, clinics, offices, and private homes, etc. X X

X X X

X

AMDEs that are too rare to notice in the statistical sample, but severe enough to require regulatory action, voluntary reporting will still need to be part of the system. All surveillance programs would involve information flowing to the monitoring organization and education and feedback flowing from that monitoring organization to the entire healthcare community. Manufacturers, healthcare providers, and insurers would also take advantage of this system when conducting any studies of marketed products. The ideal surveillance system could be implemented in any country, and using similar designs would allow smaller countries to cooperate with other countries to achieve economies of scale.

Ideal voluntary reporting Voluntary reporting is crucial to the high quality of reports. In the USA, MedWatch is the only effective system for learning of rare or unexpected AMDEs. Improvements to MedWatch would include the following:

IDEAL AMDE SURVEILLANCE PROGRAM

57

Increasing the use of MedWatch by the health provider community by educating them about MedWatch and the criticality of their information regarding rare and unexpected medical device AMDEs. Allowing anonymity for reporters. Developing data mining and other AMDE analysis support capabilities.

Ideal programs based on statistical samples of healthcare providers For these new programs to work, active surveillance tools need to be better developed. The problems of inadequate documentation of device use and AMDEs appear to be pervasive and worldwide (see Chapter 3). While these two problems are being addressed, the initial development of active surveillance programs can proceed. When the abovenoted problems have been solved, enhancements such as automated active surveillance (using electronic patient records) may also become possible. Sampling frames would probably need to be defined separately by type of provider, such as hospital, outpatient clinic, office, nursing home, and home care agency. Details for defining the sampling frames may depend on the nature of healthcare provision and payment system in the specific country. Another potential issue for consideration is the device purchasing practices for large healthcare systems; for example, a large health organization may either make bulk device purchases for all member hospitals, or allow each hospital to have complete purchasing autonomy. In the ideal statistically-based surveillance program, member healthcare providers would submit comprehensive data on device use and AMDEs in their organizations. A central surveillance organization, such as the FDA in the USA, would compile and statistically weight the information to compute national estimates. The central datacollecting organization would also need a continuing quality control program. Once the national estimates are computed, the information could be used to inform regulatory activities; educate healthcare providers, manufacturers, and patients; and measure the impact of regulatory and education efforts. The methods used by the participating providers to collect their internal data on device use and AMDEs will be critical to the success of the program. If all device uses and AMDEs are documented, the transfer of data to the central datacollecting organization will be smoothest if the original documentation and transfer are both electronic and use standard formats. Until this infrastructure is developed, it may be worthwhile to consider using data-collection tools that are less comprehensive, but much better than reporting, such as 996 ICD9 codes on the insurance claims, text string searches of electronic medical records for key phrases that indicate AMDEs, and intensive solicitation of AMDE reports from healthcare staff. The 996 codes [15,16,45] and report solicitation [16,51, 62,63] have been demonstrated to yield much more AMDE information than passive reporting.

58

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

Summary National surveillance of medical devices is necessary to protect public health and therefore it should be based on scientific (epidemiological) methods. Developing the tools for device nomenclature and documentation is crucial to achieving a high quality AMDE discovery system that covers all healthcare settings. These tools are also critical for providing the capability of conducting retrospective and more efficient prospective epidemiologic studies of device safety.

References 1. Kessler L, Ramsey S, Tunis S, Sullivan S. Clinical use of medical devices in the ‘Bermuda Triangle’. Health Aff (Millwood) 2004; 23(1): 200–207. 2. Brown SL, Bright RA, Tavris DR. Medical device epidemiology and surveillance: patient safety is the bottom line. Expert Rev Med Devices 2004; 1: 1–2. 3. Good Manufacturing Practices (GMP)/Quality System (QS) regulation. US Food and Drug Administration: http://www.fda.gov/cdrh/devadvice/32.html [accessed March 2005]. 4. Postapproval requirements. US Food and Drug Administration, 2002. http://www.fda.gov/cdrh/ devadvice/pma/postapproval.html [accessed January 2004]. 5. Medical Device Reporting. United States Code of Federal Regulations, Title 21, Part 803, revised April 1 2004. 6. White G, Weick-Brady M, Goldman S et al. Improving patient care by reporting problems with medical devices. CRNA 1998; 9(4): 139–156. 7. Postmarket surveillance studies. US Food and Drug Administration: http://www.fda.gov/cdrh/ devadvice/352.html [accessed January 2005]. 8. How to prepare an abbreviated 510(k). US Food and Drug Administration: http://www.fda.gov/ cdrh/devadvice/3145.html [accessed December 2004]. 9. Balka E, Doyle-Waters M, Lecznarowicz D, Fitzgerald J. Technology, governance and patient safety: systems issues in technology and patient safety. Int J Med Informat 2006 Sep 22; (epublication ahead of print). 10. Brennan TA, Leape LL, Laird NM et al. Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med 1991; 324(6): 370–376. 11. Rothschild JM, Landrigan CP, Cronin JW et al. The Critical Care Safety Study: the incidence and nature of adverse events and serious medical errors in intensive care. Crit Care Med 2005; 33(8): 1694–1700. 12. Thomas EJ, Studdert DM, Burstin HR et al. Incidence and types of adverse events and negligent care in Utah and Colorado. Med Care 2000; 38(3): 261–271. 13. Wilson RM, Harrison BT, Gibberd RW, Hamilton JD. An analysis of the causes of adverse events from the Quality in Australian Healthcare Study. Med J Aust 1999; 170(9): 411–415. 14. Frey B, Kehrer B, Losa M et al. Comprehensive critical incident monitoring in a neonatalpediatric intensive care unit: experience with the system approach. Intensive Care Med 2000; 26(1): 69–74. 15. Samore MH, Evans RS, Lassen A et al. Surveillance of medical device-related hazards and adverse events in hospitalized patients. JAMA 2004; 291(3): 325–334. 16. Samore M, Anderson S, Wiessner P et al. Medical device problems in intensive care units: dangers and diversity of types (manuscript cleared for publication by CDRH, 2005).

REFERENCES

59

17. Bright R, Mermel L, Richards C, Eakle M, Yoder D. Mechanical and allergic adverse events related to central vascular catheters: epidemiology in the Medicare in-hospital population, 2002. Pharmacoepidemiol Drug Safety 2004; 13(s1): s293–294. 18. Brennan TA, Localio AR, Leape LL et al. Identification of adverse events occurring during hospitalization: a cross-sectional study of litigation, quality assurance, and medical records at two teaching hospitals. Annals Intern Med 1990; 112: 221–226. 19. Bogner M (ed.). Human Error in Medicine. Mahwah, NJ: Lawrence Erlbaum, 1994. 20. Amoore J, Ingram P. Learning from adverse incidents involving medical devices. Nurs Stand 2003; 17(29): 41–46. 21. Gosbee J. Introduction to the human factors engineering series. Jt Comm J Qual Saf 2004; 30(4): 215–219. 22. Gosbee LL. Nuts! I can’t figure out how to use my life-saving epinephrine auto-injector! Jt Comm J Qual Saf 2004; 30(4): 220–223. 23. Gosbee J. Human factors engineering and patient safety. Qual Saf Health Care 2002; 11(4): 352–354. 24. Gosbee J, Gardner-Bonneau D. The human factor. Systems work better when designed for the people who use them. Healthc Inform 1998; 15(2): 141–142, 144. 25. Lin L, Vicente KJ, Doyle DJ. Patient safety, potential adverse drug events, and medical device design: a human factors engineering approach. J Biomed Inform 2001; 34(4): 274– 284. 26. Render ML. Research and redesign are safer than warnings and rules. Crit Care Med 2004; 32(4): 1074–1075. 27. Ward JR, Clarkson PJ. An analysis of medical device-related errors: prevalence and possible solutions. J Med Eng Technol 2004; 28(1): 2–21. 28. Welch DL. Human factors in the healthcare facility. Biomed Instrum Technol 1998; 32(3): 311–316. 29. Mosenkis R. Human factors in design. In Medical Devices: International Perspectives on Health and Safety, van Gruting C (ed.). New York: Elsevier, 1994; 41–51. 30. Reid MH, Sawyer D. The human factors implications of peritoneal dialysis: cycler overfill incident reports. Int J Trauma Nurs 1999; 5(2): 68–71. 31. Fairbanks RJ, Caplan S. Poor interface design and lack of usability testing facilitate medical error. Jt Comm J Qual Saf 2004; 30(10): 579–584. 32. Gosbee J. Who left the defibrillator on? Jt Comm J Qual Safety 2004; 30(5): 282–285. 33. Maisel WH, Sweeney MO, Stevenson WG, Ellison KE, Epstein LM. Recalls and safety alerts involving pacemakers and implantable cardioverter-defibrillator generators. JAMA 2001; 286(7): 793–799. 34. Rosenqvist M, Beyer T, Block M, den Dulk K, Minten J, Lindemans F. Adverse events with transvenous implantable cardioverter-defibrillators: a prospective multicenter study. European 7219 Jewel ICD investigators. Circulation 1998; 98(7): 663–670. 35. Rothschild JM, Keohane CA, Cook EF et al. A controlled trial of smart infusion pumps to improve medication safety in critically ill patients. Crit Care Med 2005; 33(3): 533–540. 36. Pierce E Jr. Looking back on the anesthesia critical incident studies and their role in catalysing patient safety. Qual Saf Health Care 2002; 11: 282–283. 37. Kaye R, Crowley J. Medical device use-safety: incorporating human factors engineering into risk management; identifying, understanding, and addressing use-related hazards. Guidance for industry and FDA premarket and design control reviewers, 2000: http://www.fda.gov/cdrh/ humfac/1497.pdf [accessed September 2005]. 38. Cook DM. Iatrogenic illness: a primer for nurses. Dermatol Nurs 2002; 14(1): 15–20, 52.

60

CH4 SURVEILLANCE OF ADVERSE MEDICAL DEVICE EVENTS

39. Kaye R, North R, Peterson M. UPCARE: an analysis, description, and educational tool for medical device use problems: https://www.medsun.net/participants/Uploads/kaye_2491.pdf [accessed November 2005]. 40. Schultz DG. FDA Preliminary Public Health Notification: recall of Boston Scientific ENTERYX1 procedure kits and ENTERYX1 injector single packs for treatment of gastroesophageal reflux disease (GERD), October 17 2005: http://www.fda.gov/cdrh/safety/101405enteryx.html [accessed April 2006]. 41. Kessler D, Kennedy D. MedWatch: FDA’s new medical products reporting program. J Clin Eng 1993; 18(6): 489–492. 42. Field M, Tilson H (eds). Safe Medical Devices for Children. Washington, DC: National Academies Press, 2005. 43. Medical devices: early warning of problems is hampered by severe underreporting. US General Accounting Office, GAO/PEMD 87–1; 1987. 44. Hefflin B, Gross T, Schroeder T. Estimates of medical device-associated adverse events from emergency departments. Am J Prev Med 2004; 27(3): 246–253. 45. Bright R, Shen J. Use of a free, publicly-accessible data source to estimate hospitalizations related to adverse medical device events (draft manuscript 2005). 46. Sievanen H. User reporting of medical device-related incidents. Med Device Technol 2003: 14(4): 26–29. 47. Interagency Registry for Mechanically Assisted Circulatory Support: Manual of Operations: http://www.uab.edu/ctsresearch/mcsd/documentlibrary.htm [accessed April 2006]. 48. Buehler J. Surveillance. In Modern Epidemiology, 2nd edn, Rothman KJ, Greenland S (eds). Philadelphia, PA: Lippincott-Raven, 1998. 49. Kohn L, Corrigan J, Donaldson M. To Err is Human: Building a Safer Health System, 1st edn. Washington, DC: National Academy Press, 2000. 50. Weingart SN, Ship AN, Aronson MD. Confidential clinician-reported surveillance of adverse events among medical inpatients. J Gen Intern Med 2000; 15(7): 470–477. 51. Welsh CH, Pedot R, Anderson RJ. Use of morning report to enhance adverse event detection. J Gen Intern Med 1996; 11(8): 454–460. 52. Michel P, Quenon JL, de Sarasqueta AM, Scemama O. Comparison of three methods for estimating rates of adverse events and rates of preventable adverse events in acute care hospitals. BMJ 2004; 328(7433): 199. 53. O’Neil AC, Petersen LA, Cook EF, Bates DW et al. Physician reporting compared with medical-record review to identify adverse medical events. Ann Intern Med 1993; 119(5): 370–376. 54. Classen DC, Pestotnik SL, Evans RS, Burke JP. Computerized surveillance of adverse drug events in hospital patients. JAMA 1991; 266(20): 2847–2851. 55. Cullen DJ, Bates DW, Small SD, Cooper JB et al. The incident reporting system does not detect adverse drug events: a problem for quality improvement. Jt Comm J Qual Improv 1995; 21(10): 541–548. 56. Jha AK, Kuperman GJ, Teich JM et al. Identifying adverse drug events: development of a computer-based monitor and comparison with chart review and stimulated voluntary report. J Am Med Inform Assoc 1998; 5(3): 305–314. 57. Report on meeting to discuss Unique Device Identification. US Food and Drug Administration: http://www.fda.gov/cdrh/ocd/uidevices061405.pdf [accessed September 30 2005]. 58. AHRQ Profile. Advancing Excellence in Healthcare. US Agency for Healthcare Research and Quality. AHRQ Publication No. 00-P005, March 2001: http://www.ahrq.gov/about/profile.htm [accessed September 30 2005].

REFERENCES

61

59. Agency for Healthcare Research and Quality: Reauthorization fact sheet. US Agency for Healthcare Research and Quality. AHRQ Publication No. 00-P002, December 1999: http:// www.ahrq.gov/about/ahrqfact.htm [accessed September 30 2005]. 60. Performance and accountability report, fiscal year 2004. US Consumer Product Safety Commission: http://www.cpsc.gov/about/gpra/04perfrpt.pdf [accessed September 30 2005]. 61. Bonnie R, Fulco C, Liverman C (eds). Reducing the Burden of Injury: Advancing Prevention and Treatment. Washington, DC: National Academy Press, 1999. 62. Buckley TA, Short TG, Rowbottom YM, Oh TE. Critical incident reporting in the intensive care unit. Anaesthesia 1997; 52(5): 403–409. 63. Stambouly JJ, McLaughlin LL, Mandel FS, Boxer RA. Complications of care in a pediatric intensive care unit: a prospective study. Intensive Care Med 1996; 22(10): 1098–1104. 64. One Liners, Issue 25, January 2004: http://devices.mhra.gov.uk/mda/mdawebsitev2.nsf/ e8be0ee313c493aa80256bbb00307b2e/33916896eddb6c3380256e2200535602/$FILE/ issue%2025.pdf [accessed 2005].

5 The Medical Product Surveillance Network (MedSun) Roselie A. Bright, Marilyn N. Flack, and Susan N. Gardner US Food and Drug Administration, Rockville, MD, USA

Historical motivation Medical Device Reporting was mandated in 1984 (a full description of the legal basis for device regulation is provided in Chapter 2 of this book). The concept was that manufacturers would be required to report to the US Food and Drug Administration (FDA) all instances of medical device malfunction (that could result in serious injury or death if it recurred), serious injury, and death related to any of its products. In 1985 and 1986 the General Accounting Office (GAO), an organization that reports to Congress, investigated how well mandatory reporting was working. The investigation followed information about specific events in a sample of hospitals through various chains of reporting to FDA. GAO selected 10 study devices and identified a random stratified sample of 2038 hospitals. GAO mailed each hospital a questionnaire about one device randomly chosen from the list of 10 devices, and asked for a complete report of an associated adverse event in that hospital in 1984. Most (1651) of the hospitals replied that they had had an adverse event on that list of 10 devices. Most of these (78%) responded to the follow up questions about where the adverse event was reported. Out of 1175 identified adverse events, less than 1% was reported directly to FDA, 46% (543) were reported to the manufacturer or distributor, 4% were reported to other organizations, and 49% were not reported at all. Of the 543 reports that hospitals said they had sent to manufacturers, those manufacturers found records of only 139 for this study, and only three were then reported to FDA. None of the other organizations to which the user facilities had sent reports forwarded those reports to Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

64

CH5

THE MEDICAL PRODUCT SURVEILLANCE NETWORK (MEDSUN)

FDA. In total, the study estimated that less than 1% of the initial 1175 adverse events known to the study hospitals were reported to FDA, either directly or indirectly. This is probably an overestimate of the overall reporting rate because there was no assurance that the responders for the hospitals in fact knew about all the adverse events relevant to the study. The authors recommended the remedy that ‘FDA explore the possibility of establishing a voluntary, postmarketing surveillance system involving a representative sample of hospitals that would report directly to device manufacturers’ [1,2]. Mandatory user facility (hospitals, ambulatory surgical facilities, nursing homes, outpatient treatment facilities, or outpatient diagnostic facilities which are not a physician’s office) reporting of device-related deaths directly to FDA and reporting of serious injuries to manufacturers first appeared as part of the Safe Medical Devices Act (SMDA) of 1990 [3] as a partial solution to the underreporting problem. The purpose of the law was to use the direct reports from the users as a check on whether reporting from manufacturers was complete [4]. The Congressional documents do not explain why mandatory reporting by all user facilities would be better than voluntary reporting by a representative sample of hospitals. In 1991, FDA issued a public notice stating that the final reporting requirements would be published in a forthcoming regulation, but that in the meantime, user facilities should begin reporting under SMDA by the end of 1991. The final requirements were published as regulation in 1995 [5]. It became apparent by 1994 that the problem of underreporting by users was persisting in spite of the new law because the numbers of death reports from user facilities was much lower than the numbers from manufacturers (both are required to report deaths associated with the use of medical devices to FDA). The idea of active, complete adverse event reporting by a sample of healthcare facilities was re-proposed in 1996 and developed into a pilot project, ‘DeviceNet’.

Initial considerations for the design of DeviceNet During the initial design phase in 1996, features of several adverse event collection systems current at that time were examined (Table 5.1). The existing systems (still structured the same way today) for reporting adverse medical device events (AMDEs) at the FDA were examined first. They were the Medical Device Reporting (MDR), User Facility Reporting (UFR), and MedWatch programs. Under MDR, manufacturers must report to FDA all serious injuries, deaths, and malfunctions that could have resulted in serious injury or death, regardless of the healthcare setting. Under UFR, user facilities must report deaths to FDA, as well as to the manufacturer. User facilities must also report serious injuries to manufacturers, or, if they do not know the manufacturer, to FDA. MedWatch continues to be a voluntary reporting system open to consumers and any person in any healthcare setting. Next, non-FDA adverse event reporting systems were examined. Private voluntary systems were the Medication Errors Reporting System, which is administered by private

Any healthcare setting

Any healthcare setting Anyone, usually pharmacists and healthcare professionals Usually biomedical or clinical engineers

FDA

FDA US Pharmacopeia and Institute for Safe Medical Practices

ECRI

Medical Device Reporting (MDR), Alternative Summary Reporting (ASR), User Facility Reporting (UFR)* MedWatch* Medication Errors Reporting System (MERS)[7] Medical Device Problem Reporting [15]

Source of adverse event data

Administered by

System name

Voluntary (vs. mandatory) reporting X

X X

X

X X

X

Detect unexpected AEs

Third party collects and de-identifies reports

Active (vs. passive) AE detection

Various surveillance systems in operation in 1996 and their features

X

X X

X

Detect rare AEs

Table 5.1

Find problems in ‘real’ users with multiple comorbidities or problems in ‘real-world’ settings X

X X

X

Reliable data on exposure

(Continued)

Allow calculation of the public burden imposed by AEs Complete capture of AEs

Statistical sample of emergency departments Any healthcare provider

Any aviation professional

CPSC

DVA through NASA

Federal Aviation Administration, using the National Aeronautics and Space Administration (NASA)

*These systems are defined further in Chapter 2 of this book. **NEISS is discussed further in Chapter 6 of this book.

Selected hospital services; data actively gathered by infection control officer

CDC

National Nosocomial Infections Surveillance System (NNISS) [6,7] National Electronic Injury Surveillance System (NEISS)** Pilot precursor of Patient Safety Reporting System [8] Aviation Safety Reporting System (ASRS) [9,10]

Source of adverse event data

Administered by

System name

X

X

Active (vs. passive) AE detection

(Continued)

X

X

X

X

Voluntary (vs. mandatory) reporting

Table 5.1

Third party collects and de-identifies reports X

X

Detect unexpected AEs X

X

X

X

Detect rare AEs X

Find problems in ’real’ users with multiple comorbidities or problems in ‘real-world’ settings X

X

X

Complete capture of AEs X

X

X

X

Allow calculation of the public burden imposed by AEs Reliable data on exposure

INITIAL CONSIDERATIONS FOR THE DESIGN OF DEVICENET

67

firms, the US Pharmacopeia (for adverse drug events and potential adverse drug events), the Institute for Safe Medical Practices (for drugs and drug-delivery devices), and Medical Device Problem Reporting, administered by ECRI?. The designers also considered the National Nosocomial Infections Surveillance System (NNISS) [6,7], administered by the Centers for Disease Control and Prevention. Additionally, the National Electronic Injury Surveillance System (NEISS), administered by the Consumer Product Safety Commission and described in Chapter 6, was evaluated. Both NNISS and NEISS involve collecting data from a relatively small group of hospitals; NNISS hospitals volunteer reporting from selected parts of the hospitals and NEISS hospitals are statistically selected to report emergency room visits. In both systems, specialists based in the hospitals are trained to collect and transmit the reports. The pilot precursor to the VA Patient Safety Reporting System (VA PSRS) was also examined [8]. In this system, any healthcare provider in the VA healthcare facilities can report to the National Aeronautics and Space Administration (NASA), which acts as a disinterested third party because they are neither involved in healthcare nor are they regulatory. This latter system was modeled on the Aviation Safety Reporting System (ASRS), where the Federal Aviation Administration has NASA collect, process, and de-identify reports from any aviation professional [9,10]. Experts were consulted regarding the factors that engendered and hindered success in the various systems that collect information on adverse events. Design features thought to be most important were those that improved the likelihood of reporting. Likelihood of reporting was felt to be directly related to the satisfaction of the reporters, which in turn was higher if reporting was among the following: Voluntary. The systems based on voluntary reporting were MedWatch, MERS, ECRI, NNISS, NEISS, the pilot precursor of the PSRS, and ASRS. Secure. All of these systems protected reports from public identification of the reporters, but only ASRS always protected reporter identity from the sponsoring agency. Simple. Simplicity was broken into two aspects: ease of reporting, which had been addressed by dedicated computer systems (NNISS, NEISS, MERS) or telephone numbers (MedWatch, ECRI); and simple content, which probably could not be addressed when reporting on complex technical adverse events. Useful. Reporters deduced the usefulness of their reports by direct feedback from the agency to which they had reported or by public notifications from these agencies. Several of the programs provide direct root cause analysis, when available, to the reporting facility. These features were incorporated in the design of DeviceNet.

68

CH5

THE MEDICAL PRODUCT SURVEILLANCE NETWORK (MEDSUN)

Implementation of pilot (‘DeviceNet’) The voluntary nature of reporting was felt to be so important that while designing DeviceNet, volunteer facilities were solicited. Security was addressed by limiting reporting to one or two liaisons from each facility and having them report to a third party (contractor to FDA), who removed the facility identifier from the report after the report was completed. Simplicity was built in by providing the choice of reporting by paper, fax, or phone; the contractor would help the reporter learn what was desired in the reports through interactive questions about the content of each report. The usefulness of the reporting system was addressed in this small pilot by the enhanced communications with the reporter. Newsletters were sent to the participating sites with useful feedback, including de-identifed information from the reports. The DeviceNet design was tested in a small Phase I pilot of 18 hospitals and six nursing homes during 1997–1998. The basic design described above was used, and the geographic area was limited to control travel costs. When the pilot was over, representatives from the participating hospitals and nursing homes were invited to an all-day briefing session that was used to interpret the findings and influence the design of the successor project. The findings were summarized in a report to Congress [11], as required by the Food and Drug Modernization Act of 1997 (FDAMA): ‘The proposed design of the national program [Medical Device Surveillance Network (MeDSuN)], which will be implemented by regulation following the large-scale Phase II study, is one aimed at improving the protection of the health and safety of patients, users, and others by: reducing the occurrence of medical device related events; serving as an advanced warning system from the clinical community; and creating a two way communication channel between FDA and the user-facility community. This system will allow the dissemination of data regarding newly emerging device problems to healthcare professionals, both in the Network and outside it, and to the public. It will allow FDA to apply the knowledge gained from the reported data to the device approval process and to prevention and control programs. The proposed Medical Device Surveillance Network will provide FDA with a setting within which research will be conducted on current device issues. Design features important to the success of the system are: assuring confidentiality to reporters through the use of a neutral third party; providing meaningful incentives to encourage participation; minimizing the burden of participation; and providing timely feedback to the participants that demonstrates the value of their reporting.’

FDAMA mandate As noted in the previous section, while DeviceNet was under way, the US Congress passed the 1997 Food and Drug Administration Modernization Act. The law said that the current User Facility Reporting system shall eventually be replaced by one that is limited to a ‘subset of user facilities that constitutes a representative profile of user reports’ [12]. FDA was pleased by this statutory change, since it reflected the concept FDA was already testing in DeviceNet.

MEDSUN BASIC DESIGN

69

MedSun basic design The MedSun design is based on the successes of the DeviceNet pilot. During the early design and implementation phases of MedSun, the goals were refined to the following: Increase the number of adverse event reports submitted by participating facilities. Ascertain the denominator for specific device use at the participating facilities upon request. Obtain high-quality reports, particularly regarding human factor issues, for each event. This would allow Center for Devices and Radiological Health (CDRH) staff to spend less time on improving report quality. Support CDRH actions to protect against and prevent AMDEs by: *

Providing relevant information to the FDA pre-market approvals group.

*

Taking regulatory action.

*

Promoting improved device design.

Disseminate device safety information to the public. The basic design is to use a convenience sample of volunteer hospitals and other facilities within strata determined by bed size, geographic region, and hospital type (academic, community, pediatric, etc.). The rationale for starting with a convenience sample is to refine the procedures and develop the program with a flexible and willing group of participants. The MedSun pilot started in 2002 with 25 facilities restricted geographically to the eastern USA. During 2003–2005 enrollment continued across the USA, with the latest group of participants coming from the west coast. Each facility agrees to designate two MedSun representatives, generally the Risk Manager and the Biomedical Engineer. Some sites have also included patient safety professionals, quality improvement professionals, and nurse managers. The representatives’ communications to the MedSun contractor are primarily adverse event reports and the return communications are useful feedback and education. At the time of enrollment, the facility representatives receive formal training from the MedSun contractor. Continued training occurs implicitly through follow-up of individual reports by the contractor. Evidence of the success of this training mode includes the contractor becoming adept at asking follow-up questions, facilities learning to think about and report about device problems in more depth, FDA learning what questions should be standard, and continuous improvement in report quality.

70

CH5

THE MEDICAL PRODUCT SURVEILLANCE NETWORK (MEDSUN)

Further ongoing implicit training is a byproduct of custom searches of FDA’s database of adverse medical device events performed by the MedSun contractor at the request of the participating sites. Explicit training is conducted in the form of annual conferences for members, quarterly Clinical Engineering tele-conferences, and training visits to the sites made by the contractor’s MedSun regional representatives. Additionally, MedSun has created various educational materials for representatives to use to train and motivate health professionals who use medical devices about the importance of recognizing device problems and reporting them within the site. These materials include customized slide presentations, videos, and posters. The slide presentation comes with a customizable script, provides examples of medical devices used by various medical specialists, provides information about how and why device problems occur, and emphasizes the reasons for reporting the problems. The educational video provides a fast-paced real-life example of how a device problem occurs in a hospital and what the response to such an issue should be. The posters are intended to remind healthcare providers what devices are (Figure 5.1), to think about the causes of adverse events (Figure 5.2), and to report adverse events (Figure 5.3). Feedback to the sites takes various forms, such as Public Health Notifications, weekly listing of device recalls, the monthly MedSun Newsletter, topic-specific emails, quarterly audio conferences on device safety-related topics, annual conferences, and phone calls from the MedSun contractor and FDA. The MedSun Newsletter consists of several pages of articles suggested by MedSun reports, as well as a listing of the adverse event reports received that month (patient, reporter, and reporting site identifiers removed). An FDA MedSun website is under development and will be launched in the Spring of 2007. Many of the MedSun educational tools as well as the MedSun Newsletter will then be publicly available. A recent addition to MedSun is the ‘Device Safety-Exchange’ (DS-X). DS-X provides a forum where the sites may share their ’success stories’ of quality improvement projects achieved in the area of medical devices. Sites may read each other’s projects and adopt those that may also meet the needs of their institutions. MedSun also uses various other methods to obtain information about possible problems with medical devices. In order to better understand a possible emerging device issue, MedSun reaches out to the participating sites through surveys and focus groups to determine the clinical community’s experience with a particular device, or class of devices. A few examples of issues that have been explored through MedSun include: failure of vena cava filters to properly deploy; problems with the use of surgical staplers (staples fail to deploy, fail to form, etc.); questions about hypersensitivity reactions and thrombosis events associated with drug-eluting stents; questions associated with the ease of insertion and ease of withdrawal of a ‘spiral’shaped pulmonary catheter; and problems with the use of heart-valve sizers to accurately select the correct size of heart valve. Information obtained from these small studies is qualitative and not quantitative. The data has been useful in helping FDA determine whether a problem warrants further exploration.

CURRENT STATUS

71

Figure 5.1 MedSun poster intended to remind healthcare providers of the scope of medical devices. Used with permission of the artist, Rich Lillash

Current status As of November 2005, MedSun (now called the Medical Product Safety Network) has 350 participating facilities (320 hospitals of 100 or more beds, 21 nursing homes, and nine ‘other’ facilities). During 2006, FDA plans to keep enrollment at this level, but will roll out facilities which have not been active reporters and replace them with new sites. As of the end of October 2005, the cumulative number of reports was 4516, including 111 deaths and 475 serious injuries. There were 3930 voluntary reports (minor injuries,

72

CH5

THE MEDICAL PRODUCT SURVEILLANCE NETWORK (MEDSUN)

Figure 5.2 MedSun poster that encourages healthcare providers to think about the causes of adverse events. Used with permission of the artist, Judy Guitteau

malfunctions, potential for harm). Releasable data from the reports are publicly available. All MedSun participants benefit: FDA gains ties to the clinical community: obtains increased number and higher quality reports; obtains timely information about emerging device issues through surveys and focus groups; and learns of clinical issues through DS-X. MedSun facilities obtain valuable feedback to improve internal processes: access to de-identified adverse event reports; access to timely device information in newsletters; obtain use of educational materials to promote reporting within their own facilities so

CURRENT STATUS

73

Figure 5.3 MedSun poster that encourages healthcare providers to report adverse events. Artwork copyright ß 2005 David Plunkert, used with his permission

they may address patient safety problems; access to annual conferences, etc.; and learn about initiatives at other institutions which may also benefit them. The 2005 MedSun Customer Satisfaction Survey indicated that 80% of the participating sites believed MedSun had improved patient safety in their facilities (see the following section). The US clinical community benefits from ‘lessons learned’. MedSun reports have contributed to numerous regulatory actions, including product recalls, manufacturer-site inspections, and Public Health Notifications sent out to inform the public of issues MedSun brought to light. In 2006 the public will also have access to the MedSun newsletter and educational tools.

74

CH5

THE MEDICAL PRODUCT SURVEILLANCE NETWORK (MEDSUN)

In order to obtain a more in-depth view of problems that occur in particular parts of hospitals, MedSun plans to develop three ‘networks’ within the MedSun network for 2006. These three more targeted networks will include hospital laboratories, pediatric intensive care units, and catherization and electrophysiology laboratories within the MedSun sites. The hypothesis is that FDAwill obtain more in-depth reports if the data are gathered directly from and fed back to the device users.

Is MedSun successful in promoting the safe use of medical devices? In addition to the regulatory successes cited in the final item of the list above, it is important to ask the MedSun participating facilities whether the feedback they receive has been useful to them. Every year since data collection began (2002), the MedSun representatives at each site have been asked to fill out a Customer Satisfaction Questionnaire. The results for the 2005 survey are presented here. Out of 650 eligible respondents, 357 completed the survey (55% response rate). The following shows the percentage of respondents who rated different feedback items as useful, very useful, or extremely useful: Newsletter, 78%. MedSun/MAUDE database searches done for them, 96%. Public Health Notifications sent to them, 85%. Product Recall information sent to them, 87%. MedSun recall database, 91%. MedSun posters, 71%. Engineering conferences, 81%. MedSun Annual Conference, 99%. The survey asks the respondents if they had used any MedSun products to improve patient safety: 66% said they had used the Newsletter at least once to improve safety in their facility. 51% of Engineering Audio Conference participants said they had used that information at least once to improve safety.

EPIDEMIOLOGIC CONSIDERATIONS

75

76% of Annual Conference attendees said they had used that information at least once to improve safety. 58% said they had been contacted by the contractor or FDA for follow-up information on reports and found the contact helpful in more fully understanding the root cause of the event. And finally, 84% answered ‘Yes’ to the question, ‘Has participation in MedSun had an impact on patient safety in your facility?’. Each respondent provided specific examples to each of their answers. While each example was specific to each site, the overall theme of answers indicated that the participants use MedSun feedback to become more proactive in their efforts to promote patient safety with medical devices in their facilities. By learning of problems that have happened at other MedSun facilities, MedSun participants can prevent those same problems from occurring in their own sites. It is not unusual for MedSun sites to share resolutions to device problems they have encountered, and the other sites may also then put this experience to use to solve similar problems. Now that MedSun understands the type of information that user facilities find useful, the program will place that information on the public MedSun website, noted earlier, so that all medical device users may improve patient safety.

Epidemiologic considerations As discussed in Chapter 4 of this book, on surveillance, there are several goals of a surveillance system that, when met, provide the epidemiologic basis of public health interventions. These goals are: To detect rare or unexpected adverse events. To find problems in ‘real’ users with multiple comorbidities (including vulnerable populations) in ‘real-world’ settings (including many years after initial device exposure). To have complete capture of adverse events and reliable information on device exposure. To allow full appreciation of the public health burden imposed by adverse events of specific natures or related to specific device types. MedSun has significant strengths related to these goals. Although MedSun is not big enough to provide reliable detection of rare adverse events, it can provide reliable detection of uncommon and unexpected adverse events. MedSun staff continue to

76

CH5

THE MEDICAL PRODUCT SURVEILLANCE NETWORK (MEDSUN)

develop, pilot, and distribute materials to assist the Risk Managers in generating adverse event reports in their facilities, which encourages full reporting. However, there is independent evidence that Risk Managers know about only a small proportion of adverse medical device events that occur in their facilities [13,14]. If each hospital (assuming 300 hospitals) only reported the numbers of adverse events that were discovered in the discharge codes of the hospital used for the Samore et al. study (about 1100 per year), MedSun would receive 330 000 reports per year [14]. That number of reports would overwhelm the system; hence, new initiatives in MedSun are designed to enhance surveillance by implementing targeted device-specific ‘networks’ within the MedSun network, supported by designated local clinical champions. There is much goodwill for the program among MedSun participants and a waiting list to join. This enthusiasm could be used to encourage specific sites to provide not only improved data capture for adverse device-related events but to provide data on device exposure as well. Innovations in record keeping, such as the use of uniform device codes and a new culture of device use documentation, will facilitate data collection for device safety. The potential is for MedSun to become the proving ground for innovations in device-related documentation; its achievement would be the attainment of complete capture of AMDEs and reliable information on device exposure. MedSun addressed the issue of member facilities representing national experience by choosing facilities from subgroups of national facility lists, based on factors such as number of beds and region of the country. However, this is not a statistical sample. Participation by a national statistical sample would allow estimates of the public health burden imposed by various device-related adverse events. However, the full realization of the benefit will depend on full and willing participation by the sampled hospitals, which should be enhanced by the desirability of MedSun participation.

Summary MedSun was developed in response to the recognition that FDA needed to foster reporting of medical device-related adverse events by healthcare facilities. The program has achieved higher reporting rates and significant popularity among its members. These invaluable strengths can be leveraged to meet the remaining challenges to member representatives, to discover accurate rates of adverse events and device use in their facilities, and to ensure that members are a representative statistical sample of all US facilities.

References 1. United States General Accounting Office. Medical Devices: Early Warning of Problems Is Hampered by Severe Underreporting. United States Government Publication GAO/PEMD 87–1, 1987.

REFERENCES

77

2. United States General Accounting Office. Medical Devices: Early Warning of Problems Is Hampered by Severe Underreporting. United States Government Publication GAO-T-PEMD 87–4, 1987. 3. The Safe Medical Devices Act of 1990, Public Law 101–629, signed 1990. 4. Safe Medical Devices Act of 1990 Report. US House of Representatives, Report 101–808. 5. Medical devices; medical device user facility and manufacturer reporting, certification and registration. Final rule. Federal Register 1995; 60 (237): 63577–63606. Available at Federal Register Online via GPO Access (wais.access.gpo.gov), DOCID:fr11de95–18. 6. Jarvis WR. Benchmarking for prevention: the Centers for Disease Control and Prevention’s National Nosocomial Infections Surveillance (NNIS) system experience. Infection 2003; 31(suppl 2): 44–48. 7. Patient safety reporting systems and research in health and human services (HHS). Fact Sheet. April 2001. Agency for Healthcare Research and Quality, Rockville, MD: http://www.ahrq.gov/ qual/taskforce/hhsrepor.htm [accessed July 2005]. 8. Patient Safety Reporting System (PSRS) program overview: http://psrs.arc.nasa.gov/program_briefing.htm [accessed January 2006]. 9. Aviation Safety Reporting System (ASRS) program overview: http://asrs.arc.nasa.gov/ overview_nf.htm [accessed January 2006]. 10. Aviation Safety Reporting System (ASRS) (discussed in Tamuz M, Understanding accident precursors: http://www.nae.edu/NAE/engecocom.nsf/0754c87f163f599e85256cca00588f49/ 85256cfb004759c185256dd60053b88d/$FILE/Tamuz.pdf). 11. Gardner S, Flack M. Designing a medical device surveillance network: Report to Congress, 1999: http://www.fda.gov/cdrh/postsurv/medsun.html [accessed November 2005]. 12. Food and Drug Administration Modernization Act, 1997: http://www.fda.gov/cdrh/ modact97.pdf [accessed July 2005]. 13. Samore MH, Evans RS, Lassen A, Gould P et al. Surveillance of medical device-related hazards and adverse events in hospitalized patients. JAMA 2004; 291: 325–334. 14. Bright RA, Anderson S, Wiessner P, Nebeker J et al. Medical device problems in intensive care units: dangers, detection, and diversity of types (manuscript, 2005). 15. About medical device problem reporting: http://www.ecri.org/Problem_Reporting/ Problem_Reporting.aspx [accessed July 2005].

6 The National Electronic Injury Surveillance System (NEISS) and medical devices Brockton J. Hefflin and Thomas P. Gross US Food and Drug Administration, Rockville, MD, USA

Thomas J. Schroeder US Consumer Product Safety Commission, Bethesda, MD, USA

Each year, hospital emergency departments (EDs) are witness to a great number and variety of injuries and other types of product-related adverse events. Many such incidents occur in association with products that are distributed throughout, and used by, the entire community. The National Electronic Injury Surveillance System (NEISS) has the ability to capture information from EDs on problematic events that are associated with the use of various products, and therefore provides an outstanding opportunity for the active surveillance of potential public health problems.

Description and history of NEISS NEISS is the US Consumer Product Safety Commission (CPSC)’s main source of product-related and recreational injury data; it is based on over 350 000 ED records collected each year. As of January 1 2005, NEISS consisted of a statistical sample of 97 hospital EDs representative of all EDs open to the general public in the USA and its

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

80

CH6

THE NATIONAL ELECTRONIC INJURY SURVEILLANCE SYSTEM

territories (over 5000). Information from ED records is searched by a CPSC-trained hospital representative (NEISS hospital coordinator), who then enters NEISS case reports onto a computerized abstraction form, which is made available electronically to the CPSC on a daily basis. Based on the sample design, an analysis weight is assigned to each case which allows for the determination of national estimates for any type of injury or group of interest [1]. NEISS methodology has been structured in this way almost since its inception, making it an effective public health tool for the past several decades. Historically, data from NEISS have been used for consumer product safety alerts to prevent injuries related to such issues as baby walker design, lack of bicycle helmet use, falls (of children) from shopping carts, and unsafe trampoline use [2–5]. There is a core set of variables entered into NEISS from the ED record for every NEISS case. The variables include the age, gender, and race of the injured person, the locale of the injury incident (e.g. home or school), the body part injured, the injury diagnosis, and the patient disposition (e.g. treated and released, or hospitalized). Up to two products associated with the injury can be coded in NEISS, and applicable cases are coded as work-related incidents. Additionally, a free text narrative is entered to describe verbatim the incident as recorded in the ED record [6]. An additional ‘second screen’ of variables are coded for cases that meet certain inclusion criteria at the hospital level (e.g. all children’s poisonings prompt a second screen that has variables related to the poisoning and treatment). The NEISS started in 1969 as a study by the National Commission on Product Safety. It was briefly under the jurisdiction of the FDA in 1970–1971 and included a statistical sample of 130 hospitals. With the advent of the CPSC in the early 1970s, the NEISS became the core of CPSC’s Bureau of Epidemiology. The NEISS is occasionally resampled, most recently in 1997, to reduce the sample size due to budget and statistical constraints. Starting in the 1980s, CPSC entered into individual agreements with other federal agencies to expand NEISS for the collection of data pertinent to those other agencies (e.g. data on violence, work-related injuries, and firearm injuries). In July 2000, with funding from the Centers for Disease Control and Prevention, NEISS was expanded to collect all trauma-related visits to the emergency department. This expansion was named the ‘All Injury Program’ (NEISS-AIP); budget constraints, however, have limited this expansion to a subsample of two-thirds of NEISS hospitals (i.e. all trauma cases are reported from a smaller sample of hospital EDs). National projections of trauma injuries from this subsample are still possible.

Potential uses and limitations of NEISS The NEISS can be used to produce national estimates and track trends annually for a wide variety of product-associated injuries. Any incident that can be routinely captured in an ED record can conceptually be captured by NEISS. An analysis of the data can estimate the magnitude of injuries and help determine hazard patterns and potential ways to reduce or minimize injuries in the future. Using NEISS for trend analysis after a problem has been addressed also helps to measure the effectiveness of the intervention.

UTILIZATION OF NEISS

81

For special studies, NEISS can be expanded beyond its basics to include the automated ‘second screen’, in which additional study-specific information is abstracted (e.g. worker’s occupation for work-related injuries, or seatbelt and airbag information for motor vehicle collisions). Additionally, telephone interviews can be conducted as part of NEISS to collect information directly from the injured person that would not routinely be included in an ED record (e.g. product brand names, model numbers, user experience). Like any other surveillance system, NEISS has limitations. Only incidents in which the injured party seeks treatment in an ED are captured. Treatment received at other medical sites, such as a family doctor’s office or health clinic, are not captured. Also, NEISS only collects first-time visits to the ED, and thus cannot be used as a measure of ED utilization frequency; data on follow-up visits and transfers are not collected. Since NEISS data are abstracted from ED charts, only those adverse events that are recognized and recorded by the treating physician can be identified. The NEISS is also subject to both sampling and non-sampling errors. Sampling errors exist because NEISS is a sample of 97 hospitals from the approximately 5000 in the USA; thus, every national estimate has a corresponding variability or confidence interval around it. Non-sampling errors result from coding mistakes, misunderstanding of reporting rules, missing or incomplete ED records, or misreading of handwriting on the records.

Utilization of NEISS to produce national medical device-associated adverse event estimates The postmarket surveillance of medical devices by the FDA is centered on the review of medical device-associated adverse event reports received by the agency through its Medical Device Reporting (MDR) system. By virtue of the fact that the MDR system is dependent on reports submitted by device manufacturers, device importers, user facilities, and the public, it cannot be used to establish national estimates of device-associated adverse events; therefore, the true public health burden of such events has long been unknown (see Chapter 4). The production of statistically valid estimates of deviceassociated adverse events has been a long-time goal of the medical device regulatory community, in order to: (a) determine the scope of problems related to medical devices; (b) better characterize the type of problems related to medical devices; (c) initiate targeted public health interventions; and (d) evaluate the effectiveness of the interventions. For these reasons, a pilot study was conducted to use NEISS to produce the firstever national estimates of medical device-associated adverse events [7].

Data collection and analysis An interagency agreement between FDA and CPSC enabled the collection of NEISS case reports involving medical device products for 1 year (July 1999–June 2000).

82

CH6

THE NATIONAL ELECTRONIC INJURY SURVEILLANCE SYSTEM

To aid in their recognition of medical device-associated cases during collection, NEISS hospital coordinators were provided with a medical device definition as well as a list of examples of the wide variety of medical devices regulated by the FDA. The NEISS hospital coordinators were instructed to select cases where the ED record indicated that a medical device contributed to, or may have contributed to, an incident resulting in a patient medical problem (including physical trauma, pain, and medical conditions such as hypersensitivity reaction or dyspnea). The abstraction process resulted in 10 395 medical device-associated NEISS case reports. Because of the expense, the NEISS second screen and telephone interview options were not used in this study. Based on the type of device(s) involved in a particular incident, each of the 10 395 NEISS case reports was assigned to one or two (some cases involved two devices) of 15 medical specialty groups (e.g. cardiology, radiology, orthopedics) used to classify medical devices by the FDA. Each medical specialty group was stratified by device categories (e.g. prostheses, fixation devices, orthopedic drills). National estimates for the number of annual cases were determined by medical specialty, device category, injury diagnosis, demographic characteristics, and patient disposition.

National estimate The total annual estimated number of adverse events was 454 383 (95% confidence interval, 371 156–537 610), involving over 60 medical device categories (from toothbrushes and tampons to sun lamps and pacemakers). Devices in the physical medicine specialty (including, for example, wheelchairs,crutches, specialized chairs) were identified in 40%of the total estimated number of cases. An unintentional traumatic event was the most common mechanism of injury, resulting most commonly in contusions/abrasions, punctures, and lacerations; 13% of the total estimated number of cases resulted in patient hospitalization. Occupationally-related incidents occurred within healthcare facilities; however, incidents occurred at home more frequently than any other location (about 42%).

Significance of medical device case national estimate The total estimated number of cases in the pilot study represented over 1% of the average annual number of all injury-related visits to hospital EDs in the USA for that year (n ¼ 39 029 000) [8]. More importantly, the pilot study estimate was over four times greater than the annual number of adverse event reports received by the FDA’s MDR system, which includes reports from several sources, as well as duplicate-incident and device malfunction-only incident reports. This suggests that the scope and severity of patient problems related to medical devices is probably greatly underestimated by MDR reports alone, and therefore deserves increased attention. This attention is especially warranted given that the home is apparently becoming an increasingly prevalent location for medical device use.

POTENTIAL FOR LONG-TERM UTILIZATION OF NEISS

83

Potential for long-term utilization of NEISS for medical device surveillance The promise of using the standard NEISS tool as a potentially important surveillance tool for medical device-related patient injury was established in the study by Hefflin et al. described above [7]. Although that exploratory study was limited by its ability to assess device-relatedness (a problem inherent to the analysis of MDR data as well), it nonetheless brought to light the significant potential public health burden associated with use of these products. Further efforts to refine and apply these lessons learned in the pilot study to support the long-term utilization of NEISS for medical device surveillance is presented and discussed here.

Refining and targeting the public health burden Device-related adverse events are those in which the medical device is considered to have caused or contributed to the event. This definition includes not only problems inherent in the physical device itself (e.g. hypodermic needle or heart valve), but also its conditions of use (e.g. use error, poor maintenance, and adverse environmental factors associated with its use) [9]. For some events, such as an unintentional puncture from a hypodermic needle, the device contribution to the event is ‘obvious’, whereas for many others, such as heart failure secondary to a paravalvular leak, it may not be. Further information is needed to differentiate events as device-related from those merely deviceassociated. Additional detailed information is also needed to assess the mechanism of injury. Further efforts, therefore, are being made to improve and adapt for these purposes the information collected through the standard NEISS tool. As noted by Hefflin et al. [7], these efforts include: (a) additional training of NEISS hospital coordinators on appropriate case ascertainment and gathering more specific information from the ED record if possible, e.g. more details on the device and the event to determine relatedness; (b) inhospital review of emergency department records and NEISS case report samples to assess data quality and completeness of case ascertainment; and (c) consideration of other methods, e.g. posters in the ED, to encourage and remind staff to collect targeted information on cases involving medical devices. The ED record is often vague or incomplete when describing the medical device involved in an incident, therefore, awareness of the importance of recording such information may need to be emphasized among ED staff if better quality data are to be collected. In concert with these efforts, special studies as described in other parts of this book may be instituted to target segments of particular interest in the medical device arena. As the home increasingly becomes the point-of-care for many illnesses, the public health burden of adverse events related to use of devices in the home is of increasing concern [9,10]. The FDA has recently focused its efforts in this area and will need to work with stakeholders to develop device intervention strategies to diminish the

84

CH6

THE NATIONAL ELECTRONIC INJURY SURVEILLANCE SYSTEM

potential for adverse impact of home-use devices [11]. Given the high proportion of adverse device events associated with home use [7,9], more refined NEISS data may be of particular use for both choosing interventions and measuring the interventions’ impacts.

Long-term applications The recent Institute of Medicine report, To Err is Human: Building a Safer Health System [12], allows these efforts to be put into context with other safety initiatives. The report, among other items, calls for enhanced reporting of medical product-related hazards as well as for agencies, such as the FDA, to increase its attention to public safety. Building upon and refining the work done in NEISS to date will contribute to these efforts. More refined and longer-term use of NEISS data will: (a) help to derive estimates of rates of underreporting of device-related events; (b) embellish knowledge of the safety profile of devices in targeted populations; and (c) provide data on specific trends cover time, to assess interventions. FDA is often called upon to provide estimates of the amount of underreporting to the Agency of device-related adverse events. Few robust estimates exist. A figure of less than 1% (of the true adverse event rate) is often quoted, based on a General Accounting Office report of device adverse events occurring in hospitals [13]. More recent, rigorous data suggest underreporting rates as high as 40-fold, when comparing ICD-9 discharge codes to voluntary incident reporting within hospitals [14]. Using NEISS data, one might be able to compare, for example, estimates of the number of life-threatening device-related events (by device type) to the number reported by caregivers to the FDA. Such efforts can help target intervention efforts by FDA to enhance timely reporting of certain devicerelated events of interest. Gathering more refined data through NEISS will also provide a more complete understanding of device hazards in target populations. Such information can complement information gathered through FDA’s spontaneous reporting systems to provide insight, for example, into problems with the home use of devices or particular issues related to the pediatric use of devices. Much of the latter is used ‘off-label’ [15], and thus further safety information may be of particular importance in this vulnerable population. Longer-term NEISS data can also provide a window into trends, both general and specific, of device-related hazards. For example, NEISS data was used recently by FDA to assess trends in reports of problems related to menstrual tampons. The identification of such trends can be of value in assessing the causal relationship between the use of medical devices and adverse events, or the effectiveness of public health interventions. Data gathered on all three fronts (i.e. on underreporting, safety profiles, and trends) will enhance FDA’s ability to positively intervene in important device-related public health issues. Such efforts are among those called for in the Institute of Medicine’s recent report on patient safety [16].

REFERENCES

85

References 1. Schroeder T, Ault K. The NEISS sample (design and implementation), 1997 to present. Washington DC: US Consumer Product Safety Commission, Division of Hazard and Injury Data Systems, June 2001. 2. Consumer Product Safety Commission. Consumer product safety alert: www.cpsc.gov/cpscpub/ pubs/5086.pdf [accessed June 2 2005]. 3. Consumer Product Safety Commission. Consumer product safety alert: www.cpsc.gov/cpscpub/ pubs/5002.pdf [accessed June 2 2005]. 4. Consumer Product Safety Commission. Consumer product safety alert: www.cpsc.gov/cpscpub/ pubs/5075.pdf [accessed June 2 2005]. 5. Consumer Product Safety Commission. Consumer product safety alert: www.cpsc.gov/cpscpub/ pubs/085.pdf [accessed June 2 2005]. 6. NEISS Coding Manual. Washington, DC: US Consumer Product Safety Commission, 2005. 7. Hefflin BJ, Gross TP, Schroeder TJ. Estimates of medical-device associated adverse events from emergency departments. Am J Prev Med 2004; 27(3): 246–253. 8. National Center for Health Statistics. Health, United States, 2002. Hyattsville, MD: National Center for Health Statistics, 2002. 9. ECRI. Medical device problem reporting for the betterment of healthcare. Health Devices 1998; 27(8): 277–292. 10. Ruggiero C, Sacile R, Giacomini M. Home telecare. J Telemed Telecare 1999; 5: 11–17. 11. US Food and Drug Administration. Centers for Devices and Radiological Health Home Health Care Committee, Rockville, MD, August 2004. Food and Drug Administration Internet Site: http://www.fda.gov/cdrh/cdrhhhc 12. Institute of Medicine. To Err is Human: Building a Safer Health System. Washington, DC: National Academy Press, 2000. 13. General Accounting Office. Medical Devices: Early Warning of Problems Is Hampered by Severe Underreporting. Washington, DC: GAO/PEMD 87–1, 1986: 41. 14. Samore MH, Evans RS, Lassen A et al. Surveillance of medical device-related hazards and adverse events in hospitalized patients. J Am Med Assoc 2004; 291(3): 325–334. 15. US Food and Drug Administration. Report to Congress: barriers to the availability of medical devices intended for the treatment or diagnosis of diseases and conditions that affect children, October 2004, pp. 1, 13, 17. 16. Institute of Medicine. Patient Safety: Achieving a New Standard for Care. Washington, DC: National Academy Press, 2004.

7 Medical device nomenclature Brockton J. Hefflin and Thomas P. Gross US Food and Drug Administration, Rockville, MD, USA

Elizabeth A. Richardson and Vivian H. Coates ECRI, Plymouth Meeting, PA, USA

A medical device nomenclature is a complex system of names, definitions, and codes utilized by the greater healthcare community to identify groups of medical devices for specific purposes. Those purposes may include product registration and adverse event reporting associated with regulatory agencies, product procurement and inventory associated with healthcare facilities, documentation of patient care by healthcare providers, or product research performed by public health investigators. The nomenclature greatly facilitates data management for medical device-group information, and therefore requires fastidious development and maintenance, and general acceptance.

Technical elements A medical device nomenclature intended for surveillance and epidemiology activities should be one that supports the exchange of information on medical devices between various stakeholders. To support such an exchange, the technical elements of a nomenclature should be designed to support the consistent development and application of the nomenclature over time. This may be challenging in the world of medical devices, as it is one that is constantly evolving: new devices are developed and enter routine use, some devices change over time, uses of devices evolve, and others become obsolete.

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

88

CH7 MEDICAL DEVICE NOMENCLATURE

A standardized, controlled nomenclature for medical devices begins with a clear definition of its purpose, as well as the overall scope of the terminology included. The scope and coverage of a particular medical device vocabulary may vary, depending on its key users. Technical elements of a medical device nomenclature should support an overall organization that helps users consistently code, capture, and retrieve information at the generic device level, e.g. a biphasic external defibrillator, as well as relate specific models/makes of devices to one another as members of a common group [1]. In addition, a controlled vocabulary needs to be comprehensive – gaps may lead to inappropriate use of existing concepts, which can adversely affect information retrieval and analysis [1,2].

Device naming and level of specificity Controlled medical device nomenclatures are typically concept-based, as opposed to term-based. A concept has a unique, identifiable meaning, and can be represented by many ‘terms’ or ‘names’ [1–4]. For example, the concept of ‘pacemaker’ can be represented by the terms ‘pacer’, ‘pulse generator’, or as a specific type of ‘stimulator’. Typically, as is the case with several of the established medical device nomenclatures available today, a ‘preferred term’ for a given concept is always identified. This preferred term is the formal name associated with the concept. In addition to the preferred term, many different ‘entry terms’ or cross-references, such as synonyms, lexical variants, etc., will be included [1,3]. Medical device nomenclatures that are rich in entry terms that capture alternative names for a given concept can greatly increase the usability of a nomenclature across different applications, including medical device surveillance and related activities [1,3,5]. Constructing preferred terms typically follows a consistent approach that supports key hierarchical relationships among broader categories of devices and the related, more specific technologies [1,5,6]. Many of the nomenclatures available today start with what is referred to as a ‘base concept’ (e.g. ‘Pacemaker’) and develop more specific terms, using descriptive qualifiers to designate the individual devices’ key characteristics (e.g. ‘Pacemaker, Cardiac’, ‘Pacemaker, Cardiac, Implantable’, ‘Pacemaker, Cardiac, External’, ‘Pacemaker, Cardiac, Implantable, Dual-Chamber’, and ‘Pacemaker, Cardiac, External, Transcutaneous’). Terms using natural language are typically incorporated into the nomenclature as ‘entry terms’ or cross-references (e.g. ‘Cardiac Pacemaker’, ‘Cardiac Pulse Generator’, ‘Dual-chamber Implantable Pulse Generator’, ‘Cardiac Stimulator’ and ‘Single-chamber Atrial Pacemaker’) [1,3,5,6]. Determining the appropriate level of specificity for the concepts included in a given medical device nomenclature can be a difficult task. A nomenclature that contains concepts that are too specific may be difficult to use from the perspective of indexing and coding information; on the other hand, a nomenclature that contains concepts that are too broad renders it useless for retrieving specific discrete sets of data for review and analysis [1,5]. Typically, when determining the desired level of specificity for various medical device concepts, one should view the overall scope and intended purpose of the nomenclature, as well as the key characteristics of the different groups of included medical devices [1].

TECHNICAL ELEMENTS

89

The level of specificity within a medical device nomenclature should support the users’ ability to code, capture, and retrieve information on unique categories of devices that share a common set of characteristics. A term should reflect, as far as possible, the characteristics of the concept which are given in the definition. This is referred to as the accuracy (motivation) of the term [3]. Characteristics serve as a basis for the classification of concepts, and are necessary for their differentiation. Given the heterogeneity of medical devices on the market, the common characteristics that form the basis for unique concepts may vary [1,5]. In general, most medical device concepts can be organized by what is referred to as: (a) extrinsic characteristics; or (b) intrinsic characteristics. Extrinsic characteristics can include those that describe intended use or function, while intrinsic characteristics typically describe shape, size, or material. When developing concepts and the associated terms, one can rely on the various important characteristics of a particular medical device that differentiates it from the related concepts in a group. For example, the concept of ‘endoscope’ may be too broad as a single concept; breaking out the various types of endoscopes by various characteristics of intended use (e.g. colonoscopy, bronchoscopy) and to some degree, design (e.g. flexible vs. rigid) may be a better approach [1,3,5].

Supporting relationships among concepts There are several types of relationships that are typically supported in a standard medical nomenclature. The first of these is the hierarchical relationships – a technically sound medical device nomenclature should use hierarchical relationships to illustrate key parent–child relationships. Hierarchical relationships provide obvious advantages for natural navigation of terms (for information retrieval and analysis), and allow users to identify broader groups containing more specific types of a particular medical device (e.g. various types of cardiac pacemakers, various types of endoscopes, etc.) [1,4,5]. To maximize usability, the medical device nomenclature should support the characteristic of ‘polyhierarchy’, i.e. multiple coexisting hierarchies rather than a single hierarchy. This allows users to view key parent–child relationships via one or more hierarchical paths and supports effective and efficient navigation of the nomenclature, as well as comprehensive information retrieval [5]. In addition to the traditional parent–child relationships supported by a hierarchical organization, there are other types of relationships that are useful in a medical device nomenclature. An important example of these types of relationships is the partitive relationship, which demonstrates a relationship between a whole and its parts. Many medical devices, such as complex imaging systems, kits/trays/sets, and total joint prostheses, are comprised of many different components [1,5]. When trying to retrieve and analyze information for medical device surveillance, for example, relationships that illustrate the link between the larger whole and its individual parts can be very useful [7]. Another type of relationship that should be supported, in order to optimize information retrieval and analysis, is that which shows that a given medical device

90

CH7 MEDICAL DEVICE NOMENCLATURE

is used in conjunction with another device(s) to accomplish a particular action. Often in medicine, a single intervention can involve dozens of devices designed to be used in conjunction with one another, and the relationships between these devices can be extremely useful, for example, when examining information related to the potential cause of a medical device error, or analyzing a particular trend in data related to medical device safety [7]. Finally, medical device nomenclatures may include other types of data, known as ‘attributes’, that provide users with a method to relate various concepts to one another as a member of a larger group. For example, attributes designed to indicate that a particular concept is related to a given regulatory class of devices, one or more clinical specialty(s) or service area(s), or a broad category of medical devices, such as ‘capital equipment’ or ‘hospital supplies’, have been used in the medical device nomenclatures currently available [1,5].

Unique concept codes and definitions A concept in a controlled vocabulary is typically assigned a unique identifier, or code, which in itself should be meaningless. Because theworld of medical devices is constantly evolving, how we categorize these concepts is likely to change. For this reason, the unique code assigned to a concept must not be inextricably bound to a hierarchy position in the terminology, so that it is not necessary to change the code as our understanding of the medical technology evolves. Changing the code may make historical data confusing or erroneous [1,2,4]. Unique codes must not be re-used when a term is obsolete or superseded. Consistency of data over time is not possible when concepts change, especially when this changes the meaning of the code [1,5]. This encompasses the key quality indicator for nomenclature known as ‘Concept Permanence’ – that is, the meaning of a concept, once created, should not change, and the concept itself must never disappear from the nomenclature. Even if a device-group becomes obsolete, the code that represents it should neither change meaning nor be removed from the nomenclauture. In addition, there should be no ambiguity among the various concepts in the nomenclature, that is, no formal concept identifier can be linked to more than one meaning [1,2,4,5]. The concepts in a controlled medical vocabulary should be defined formally through a complete definition that provides a comprehensive description of the key characteristics of a given medical device group (e.g. intended use and general physical description). The definition must distinguish a unique concept from others in the nomenclature, and ensure that there is no redundancy among two or more concepts [1,2,4,5]. Definitions can also provide the basis for the hierarchical and other key relationships between the medical device concept in question and the other concepts in the nomenclature [3,4]. In a medical device nomenclature, explicit definitions are critical, as they ensure that the meaning users assign to a given concept is identical to that which the authors of the nomenclature have also assigned to the concept. This supports the consistent use of the nomenclature when it comes to classification and organization of information [1,2,4,5].

CURRENT TERMINOLOGIES

91

Current terminologies Following is a brief background of prominent nomenclature systems for medical devices.

Classification names for medical devices and in vitro diagnostic products The US Food and Drug Administration (FDA)’s medical device classification was the first of its kind [8]. In 1969, President Nixon, in his consumer affairs message to Congress, directed the Department of Health and Human Services (HHS) Secretary to determine the scope and nature of additional legislative controls needed to protect the public against unreasonable risk of illness or injury from medical devices. About 2 years later, the HHS Secretary requested that the FDA Commissioner develop an inventory of existing medical devices. The FDA subsequently sent out questionnaires to the medical device industry to help with the identification of all medical device products; as a result, 8000 types of devices were identified. In 1973, FDA initiated a preliminary classification of medical devices, and 2 years later the 8000 devices resulting from the earlier survey of industry were each assigned a three-letter code (product code or ‘procode’) to facilitate their use in the computer systems of the time. The list of 8000 devices was eventually condensed to a smaller list of device groups with a product code assigned to each. After enactment of the 1976 Medical Device Amendments to the Food, Drug, and Cosmetic Act [9], the FDA began to assign the medical device product codes to one of three regulatory classes, each class associated with a different level of device-related health risk. Finally, to augment the management of medical device data, FDA (Center for Devices and Radiological Health) in 1977 created an internal device classification database to provide the generic names, codes, definitions, and regulatory requirements of all regulated medical device and radiological health products.

Universal Medical Device Nomenclature SystemTM (UMDNSTM) The UMDNS [10] is an international nomenclature for medical devices that has been maintained since the late 1970s by ECRI (a company name, not an acronym: a not-for-profit health services research institute). The UMDNS developed as a means to identify medical device groups to meet the research needs of ECRI, and to meet the various needs of its customers, which include hospitals, other nonprofit organizations, government agencies, device manufacturers, and e-commerce companies located around the world. The UMDNS has been translated into several European languages. Since 1991, the US National Library of Medicine has incorporated the UMDNS into its Unified Medical Language System (UMLS) [11], a group of databases and associated programs used to enhance or create electronic information systems for biomedical and health data or informatics research. In 2003, the UMDNS was recommended by the

92

CH7 MEDICAL DEVICE NOMENCLATURE

Committee on Data Standards for Patient Safety of the US Institute of Medicine as one of the core terminologies for the electronic health record.

Global Medical Device Nomenclature (GMDN) In 1993, the European Committee for Standardization (CEN) initiated an effort to develop a pan-European medical device nomenclature primarily for the exchange of regulatory information. The International Organization for Standardization (ISO) joined the effort in 1996, widening its scope of application, and the following year the project that led to the creation of the GMDN began with the aim of creating a single, internationally-developed, global nomenclature for medical devices. The GMDN was first formed through the merger of six established nomenclature systems that included current versions of FDA’s Classification Names for Medical Devices and In Vitro Diagnostic Products [21] and the UMDNS; also included were the nomenclature systems used in Norway, Japan, the European Diagnostic Manufacturers Association’s In Vitro Diagnostic Product Classification (EDMA) [12], and the International Organization for Standardization’s Technical Aids for Disabled Persons Classification (ISO 9999) [13]. After completion of the first version in 2001, the GMDN became an ISO and CEN standard [14,15], and since that time has been further developed and maintained by a maintenance agency composed of international members primarily from regulatory bodies and the medical device industry.

Other terminologies Names for groups of medical devices are also included in additional terminologies that are widely used, such as: Systematized Nomenclature of Medicine1 (SNOMED1) [16]. United Nations Standard Products and Services Codes1 (UNSPSC1) [17]. International Classification of Disease (ICD) [18]. Healthcare Common Procedure Coding System (HCPCS) [19]. These terminologies, in general, include medical device names as just one of a number of types of descriptors (e.g. may also include names for other types of products, procedures, or clinical terminology). They were developed by various organizations primarily for medical record or product inventory purposes. However, their medical device components fall short of being ‘nomenclatures’, as previously described, because the device names are often broader than the generic group level, lack definitions, and/or they do not include a rich navigation system for medical devices.

APPLICATIONS OF NOMENCLATURE

93

Applications of nomenclature The importance of nomenclature was evident more than 150 years ago. In relation to vital statistics, William Farr noted that, ‘the nomenclature is of as much importance in this department of inquiry, as weights and measures in the physical sciences, and should be settled without delay’ [20]. Today, nomenclatures continue to be essential ‘informatics’ tools in many fields, including medical device-related healthcare. To be of utmost use, relevant nomenclatures need to be standardized, ‘since the quality and resolution of those terminologies dictate the quality of healthcare information’ [2]. As noted in the introduction to this chapter, once developed, a robust medical devicerelated nomenclature may be put to use in a variety of healthcare-related applications. This section will discuss application of such a nomenclature to device regulation, device tracking and commerce, and device research.

Device regulation The US Food and Drug Administration (FDA) is responsible for regulating medical devices and radiation-emitting electronic products. As noted previously, to carry out its regulatory mission the FDA makes use of its own nomenclature, entitled Classification Names for Medical Devices and In Vitro Diagnostic Products [21]. This nomenclature is comprised of, among other items, generic device-type terms and their associated product codes. Prior to approval or clearance of a device for marketing in the USA, the FDA either creates a new term/code (e.g. for breakthrough technologies) or assigns a pre-existing term/code to that device (e.g. for ‘me-too’ devices). These terms/codes are then used for various regulatory purposes. They are used to identify and list devices produced or distributed for marketing in the USA by a variety of registered medical device-related establishments (e.g. original equipment or contract manufacturers, repackagers, relabelers, and initial importers) [22]. The product codes are also used by manufacturers, importers, and the FDA as additional product identifiers in device recalls. As true in the USA, the European Union (EU) and the European Economic Area (EEA) have seen the need for use of a nomenclature to regulate devices in the context of developing harmonized regulations for devices marketed in the EU/EEA [23,24]. They have provisionally adopted (pending translations) the use of the GMDN for purposes, among others, of: (a) notification of other regulatory authorities of placing devices on the market; (b) certification of devices prior to marketing; (c) recalling products from the market; and (d) exchange of adverse event (or vigilance) information related to marketed medical devices (http://www.gmdn.org/index.xalter). With regard to the latter, the USA is part of a multinational network of regulatory agencies that have agreed to exchange urgent and high-profile device-related adverse event and recall information. The GMDN is used in that exchange. Aside from the EU/EEA, other regulatory authorities representing Japan, Australia, and the Scandinavian bloc countries, among others, have provisionally adopted the GMDN as well.

94

CH7 MEDICAL DEVICE NOMENCLATURE

Device management and commerce A robust medical device nomenclature is also of utmost importance in supporting effective medical technology management within healthcare institutions. Nomenclature can be used in support of device acquisition and replacement, inventory archiving and monitoring, device inspection and maintenance scheduling, and monitoring and disseminating information on device hazards [25,26]. Purchasing departments have to work closely with other parts of the healthcare institution (e.g. heads of clinical departments, biomedical engineering) to meet institutional needs (patient care, financial, and other), and a device nomenclature is an essential tool in the process. Furthermore, safety-related information, in the form of manufacturer letters, FDA alerts/notifications/recalls, or other sources, must be effectively disseminated within institutions and across institutions (e.g. within hospital consortia, HMOs, and the like). Using the same device nomenclature (across departments or between institutions) will help avert missing important safety information and potentially compromising patient safety [26]. Biomedical engineering departments are also increasingly using software packages for device management, including handling and reporting of adverse event reports and product problems based on accepted device nomenclatures [25]. The commercial application of a device nomenclature is evident in the procurement process. Internet-based applications allow end users (i.e. user facilities such as hospitals and nursing homes) and device manufacturers to access valuable information [27]. End users can access up-to-date information on devices made by various manufacturers, and manufacturers can integrate their own websites into a centralized multi-lingual product catalog. Internet applications can also be used to support after-sales services.

Device surveillance and epidemiology The importance of a product-specific nomenclature and identifiers has long been recognized in the field of drug-related safety surveillance and epidemiology. The National Drug Code (NDC) Directory was originally established as an essential part of an out-of-hospital drug reimbursement program under Medicare. The NDC serves as a universal product identifier for human drugs and is widely incorporated in healthcarerelated databases. The NDC is viewed as a cornerstone in drug-related safety assessment and epidemiologic research. Establishment of a medical device nomenclature at the ‘device group’ level is a necessary, but not sufficient, step to conduct the kind and scope of safety surveillance and epidemiology seen in the drug sector. As with the NDC, more specific deviceidentifiers (e.g. manufacturer, brand, and/or model designation) linked to device group terms, would make for a more robust system. The importance of such unique product identifiers is quite evident in the postmarket arena, where information on specific devices and their relative performance in varying settings and population subgroups is at a premium and cannot generally be derived from existing healthcare and administrative databases. For instance, administrative and billing systems (such

FUTURE DEVELOPMENTS

95

as Medicare or Blue Cross/Blue Shield) use procedure-based claims, e.g. diagnosisrelated groups (DRGs) or common procedural terminology (CPT) that typically cannot be translated into device-specific information. By universally incorporating a robust device-specific nomenclature into healthcare-related databases, far more effective device safety surveillance and epidemiologic research could be accomplished. Examples of the use of device identifier information (at varying levels of specificity) will illustrate this potential. De novo collection of manufacturer-specific and model-specific device information was important in epidemiologic assessment of risk of connective tissue disease associated with breast implant rupture [28] and risk of meningitis associated with cochlear implants [29], respectively. Existing device group-level data from national cardiovascular registries were used to assess risk factors for mortality associated with transmyocardial revascularization (a procedure based on use of medical lasers) [30] and hemorrhagic complications associated with the use of hemostasis devices [31]. Similar group-level information, obtained from nationwide hospital discharge data, were used to assess short-term nationwide morbidity and mortality associated with receipt of tissue or mechanical heart valves [32]. Medicare hospital discharge data combined with chart abstraction were used to measure mechanical complications associated with use of central venous catheters [33]. Safety surveillance has also utilized varying levels of device-specific information. Safety profiles of devices of interest have been published as case series, based on devicespecific information to the model level, of reports received through FDA’s passive adverse event reporting system [34,35]. Data-mining efforts, i.e. the use of screening algorithms and computer systems to detect unanticipated data patterns, have also been applied to look for signals of potential drug safety problems in FDA’s passive reporting system [36], and similar efforts are being explored with use of device-related events (to the manufacturer level). More active surveillance methods, based on a nationwide network of emergency departments, have been used to estimate the US public health burden of non-hospital device-associated (at the device group level) adverse events [37]. Other active surveillance methods, using device group-level data contained in ICD-9 codes, have been used to assess the rates of device-related hazards and adverse events in a major healthcare facility [38]. Suffice it to say that the more robust and specific the device nomenclature, and the more widely incorporated, the more useful it could be in furthering safety surveillance and epidemiologic research.

Future developments Work towards the goal of a single medical device nomenclature utilized by all stakeholders worldwide continues. Such standardization would provide tremendous benefit in communicating and sharing medical device data for all points between, and including, the patient and the regulatory agency. Efforts are being made to consolidate the GMDN, UMDNS, and FDA terminologies to reach this goal. As a potential solution to its search

96

CH7 MEDICAL DEVICE NOMENCLATURE

for a national medical device terminology standard, the National Committee on Vital and Health Statistics (the legislatively mandated advisory committee for health data and health information policy to the Secretary of the US Department of Health and Human Services) anticipates this consolidation. To identify medical devices at a more specific level, as previously described, efforts are also being made to explore methods for the unique identification of medical devices (e.g. the application of a barcode), and for the routine documentation of medical device identifiers (e.g. in medical records). Of course, all such methods would require specific device identifiers to be associated with generic device groups for complete identification, making a global standard device nomenclature all the more significant.

References 1. ECRI. Guide to Constructing Terms. Plymouth Meeting, PA: ECRI, 1997; 16 pp. 2. Chute CG, Cohn SP, Campbell JR. A framework for comprehensive health terminology systems in the United States: development guidelines, criteria for selection and public policy. J Am Med Inform Assoc 1998; 5(6): 503–510. 3. International Organization for Standardization (ISO). International Standard ISO 704. Terminology Work – Principles and Methods. Geneva, Switzerland: ISO, 2000; 38 pp. 4. Cimino JJ. Desiderata for controlled medical vocabularies in the twenty-first century. Methods Inf Med 1998; 37(4–5): 394–403. 5. Testimony of Vivian Coates to the US Department of Health and Human Services. National Committee on Vital and Health Statistics, Subcommittee on Standards and Security, Panel of Medical Device Terminology Testifiers – UMDNS, August 19 2003: http://www.ncvhs.hhs.gov/ 030819tr.htm 6. Testimony of Brockton Hefflin to the US Department of Health and Human Services. National Committee on Vital and Health Statistics, Subcommittee on Standards and Security, Panel of Medical Device Terminology Testifiers – GMDN, August 19 2003: http://www.ncvhs.hhs.gov/ 030819tr.htm 7. Testimony of Mark Bruley, ECRI, Panel 3: Particular Systems Issues. Written Statement. National Summit on Medical Errors and Patient Safety Research, September 2000: http:// www.quic.gov/summit/wbruley1.htm 8. US Food and Drug Administration. FDA Medical Device Product Classification Database: http:// www.fda.gov/cdrh/procode.html [accessed April 12 2005]. 9. The Medical Device Amendments of 1976. 10. ECRI. Universal Medical Device Nomenclature System2 (UMDNS2). Plymouth Meeting (PA): ECRI, 2005: http://www.ecri.org [accessed April 14 2005]. 11. US National Library of Medicine. Unified Medical Language System: http://www.nlm.nih.gov/ research/umls/about_umls.html [accessed April 12 2005]. 12. About EDMA: http://www.edma-ivd.be/edma_fr01.htm [accessed April 14 2005]. 13. WHO. Technical Aids for Persons with Disabilities – Classification and Terminology (ISO 9999): http://www.who.int/classifications/icf/iso9999/en/ [accessed April 14 2005]. 14. International Organization for Standardization (ISO). ISO 15225:2000, Nomenclature – Specification for a nomenclature system for medical devices for the purpose of regulatory data exchange. 15. European Committee for Standardization (CEN). EN ISO 15225:2000, Nomenclature – Specification for a nomenclature system for medical devices for the purpose of regulatory data exchange.

REFERENCES

97

16. SNOMED1 International: http://www.snomed.org/ [accessed April 14 2005]. 17. UNSPSC1: http://www.unspsc.org/ [accessed April 14 2005]. 18. WHO/International Classification of Diseases: http://www.who.int/classifications/icd/en/ [accessed April 14 2005]. 19. Center for Medicare and Medicaid Services, Healthcare Common Procedure Coding System (HCPCS), level II coding procedures: https://www.cms.hhs.gov/medicare/hcpcs/codpayproc.asp [accessed April 14 2005]. 20. Farr W. First Annual Report of the Registrar-General of Births, Deaths, and Marriages in England. London, UK, 1839; p. 99. 21. US Food and Drug Administration. Classification Names for Medical Devices and In Vitro Diagnostic Products. Rockville, MD: US Food and Drug Administration. 22. US Food and Drug Administration. Registration and device listing: http://www.fda.gov/cdrh/ reglistpage.html [accessed June 3 2005]. 23. Council of European Communities, Council Directive 93/42/EEC. Official Journal of the European Commission, Brussels 1993; L-169. 24. Council of European Communities, Council Directive 98/79/EC. Official Journal of the European Commission, Brussels 1998; L-331. 25. Bliznakov Z, Pappous G, Bliznakova K et al. Integrated software system for improving medical equipment management. Biomed Instrument Technol 2003; 37: 25–33. 26. Agency for Healthcare Research and Quality (AHRQ). A model ‘best practice’ for patient safety focused medical technology management: a US Air Force–ECRI collaboration. In Advances in Patient Safety: From Research to Implementation. Rockville, MD: AHRQ, 2005. 27. Palamas S, Kalivas D, Panou-Diamandi O. An internet-based system for the commerce of medical devices. IEEE Eng Med Biol 2002; March/April: 26–32. 28. Brown SL, Pennello G, Berg WA et al. Silicone gel breast implant rupture, extracapsular silicone, and health status in a population of women. J Rheum 2001; 28(5): 996–1003. 29. Reefhius J, Honein MA, Whitney CG et al. Risk of bacterial meningitis in children with cochlear implants. New Engl J Med 2003; 349(5): 435–445. 30. Peterson ED, Kaul P, Kaczmarek RG et al. From controlled trials to clinical practice: monitoring transmyocardial revascularization use and outcomes. J Am Coll Cardiol 2003; 42: 1611–1616. 31. Tavris DR, Gallauresi B, Lin B et al. Risk of local adverse events following cardiac catheterization by hemostasis device use and gender. J Invas Cardiol 2004; 16(9): 459–464. 32. Astor BC, Kaczmarek RG, Hefflin BJ, Daley WR. Mortality following aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000; 70: 1939–1945. 33. Bright RA, Mermel L, Richards C, Eakle MR, Yoder D. Mechanical and allergic adverse events related to central vascular catheters: epidemiology in the Medicare in-hospital population, 2002. Value Health 2004; 7: 331–332. 34. Brown SL, Woo EK. Surgical stapler-associated fatalities and adverse events reported to the Food and Drug Administration. J Am Coll Surg 2004; 99: 1–8. 35. Brown SL, Reid MH, Duggirala HJ. Adjustable silicone gastric banding adverse events reported to the Food and Drug Administration. J Long-term Effects Med Implants 2003;13(6): 509–517. 36. Szarfman A, Machado SG, O’Neill RT. Use of screening algorithms and computer systems to efficiently signal higher-than-expected combinations of drugs and events in the US FDA’s spontaneous reports database. Drug Safety 2002; 25(6): 381–392. 37. Hefflin BJ, Gross TP, Schroeder TJ. Estimates of medical device-associated adverse events from emergency departments. Am J Prevent Med 2004; 27(3): 246–253. 38. Samore MH, Evans RS, Lassen A et al. Surveillance of medical device-related hazards and adverse events in hospitalized patients. J Am Med Assoc 2004; 291(3): 1–10.

8 Data sources for medical device epidemiology studies and data mining Danica Marinac-Dabic, Baoguang Wang, Brockton J. Hefflin, Hesha J. Duggirala and Tripthi M. Mathew US Food and Drug Administration, Rockville, MD, USA

Cara J. Krulewitch University of Maryland School of Nursing, Baltimore, MD, USA

Introduction The world of medical devices is rapidly expanding. It is estimated that 20–28 million Americans have some type of implanted medical device [1]. This estimate would be astonishingly higher if it included patients with any type of medical device (both implanted and non-implanted). Finally, if the number included estimates of exposure to diagnostic medical devices including in vitro diagnostics, the proportions of exposed patients would increase dramatically. For the majority of medical devices (i.e. Class I and II devices), very few or no clinical data are required before marketing. For approximately 1% of new medical devices that are marketed in the USA each year (i.e. Class III devices), clinical trial data are required to demonstrate the safety and effectiveness of these devices. These data are usually collected from a highly selective patient population at facilities where healthcare professionals are well trained. In addition, these clinical trial data tend to contain relatively short follow-up, although patients and caregivers are interested in the safety Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

100

CH8

DATA SOURCES FOR MEDICAL DEVICE

and effectiveness of these devices for the entire life span of patients. Once a medical device is on the market, it is generally used in a broader patient population for a much longer period of time than that for clinical trials [2–4]. Caregivers’ training and experience in using the medical devices could also vary greatly. Therefore, it is crucial to monitor continuously the safety, effectiveness and reliability of medical devices after they are permitted onto the market. Thus, epidemiologic studies are essential tools to estimate both prevalence and incidence of utilization and adverse events associated with marketed medical devices, and to continue to evaluate the effectiveness and risk:benefit ratio of medical devices. Medical device epidemiologic studies may address numerous public health issues related to device use [3], such as: (a) description of characteristics of patients who use medical devices including demographic profile, underlying disease, medical history, and other factors that might affect the performance of medical devices; (b) estimation of the prevalence and incidence of utilization of medical devices and adverse events associated with medical devices; (c) assessment of safety and effectiveness and public health impact; (d) evaluation of risk:benefit ratio; and (e) comparative analysis of different medical devices or comparison of current medical devices with other medical interventions. The characteristics of a specific study depend among other things on the questions that need to be answered, the urgency of answering these questions, and available resources for conducting the study. The medical device epidemiology program at the Center for Devices and Radiological Health (CDRH) uses a variety of internal and external databases, explores new methodologies of surveillance, including enhanced and active surveillance, expands existing device registries and develops new ones, contributes supplemental questions to the national surveys, assesses published data and explores data-mining methodology as a tool for signal detection. We present an overview of data sources used in epidemiologic studies of medical devices. Additionally, we discuss their strengths and limitations and their significance for the CDRH postmarket program. Finally, we identify the challenges and needs for future use of databases in medical device epidemiology.

Data sources Identification of optimal data sources for medical device epidemiology continues to be a challenge. Medical device epidemiology data typically come from: (a) health-related databases; and (b) device-specific studies. This very broad classification of data sources has its limitations because very often there is no clear boundary between the two, as databases may offer the framework for nested device-specific studies. A database has been defined as ‘an organized set of data or collection offiles that can be used for a specified purpose’ [5]. Here we broadly define a health-related database as a ‘collection of demographic and health-related data for a specific public health purpose’. According to the methodological means and purpose of data collection, we further classify these databases to: (a) surveillance databases; (b) registries; (c) administrative

SURVEILLANCE DATABASES

101

databases; (d) survey-related databases; and (e) technology assessment databases. Medical device databases can be sponsored by governmental agencies, professional organizations, manufacturers of medical devices, health insurance companies, academia, non-profit organizations; and other interested parties. Device-specific epidemiologic studies are a heterogeneous category. They may involve individual study (designed to answer specific question) or the synthesis of existing studies (designed to enhance the current knowledge about a specific device). The individual studies may be retrospective or prospective, experimental or observational (cohort, case-control, case-cohort, cross sectional, quasi-experimental) and may be sponsored by industry, professional organizations, academia, government, etc. We first present an overview of major types of databases used for epidemiologic studies of medical devices and we illustrate each category with specific examples. Where applicable, we further classify these major categories according to the sponsoring party. Finally, we broadly describe potential application of the data-mining technique in the study of medical device databases. For more details regarding device-specific studies as data sources (including their designs), the reader should consult the relevant chapters of this book.

Surveillance databases The World Health Organization has defined surveillance as ‘an ongoing systematic collection, analysis and interpretation of health related data for the purpose of planning, implementation and evaluation of public health programs’ [6]. The main objective of medical device surveillance is to detect patterns of actual or potential medical device adverse events. An effective surveillance program should be built on epidemiologic principles, allowing estimates of public health impact of device specific adverse events. The detailed methodological discussion of epidemiologic surveillance programs is provided in Chapter 4.

FDA surveillance systems MAUDE database The US Food and Drug Administration (FDA) maintains a database containing data on device-related adverse event reports (see Chapter 2). This database is currently known as the Manufacturer and User Facility Device Experience Database (MAUDE). Presently administered as part of the FDA’s MedWatch program, the program accepts information voluntarily submitted to the FDA by healthcare providers and consumers [7]. In 1973, the FDA began a voluntary reporting program for adverse events associated with medical devices. In 1984 CDRH implemented mandatory reporting as part of the Medical Device Reporting Regulation, 21CFR803 [8]. Under this regulation,

102

CH8

DATA SOURCES FOR MEDICAL DEVICE

manufacturers and importers are required to submit reports of device-related deaths, serious injuries, and malfunctions. Since enactment of the Safe Medical Device Act (SMDA) in 1990 [9] and Medical Device Amendments of 1992 [10], the mandatory universal reporting of adverse events by device user facilities was initiated. Facilities that use medical devices subject to mandatory reporting include hospitals, nursing homes, outpatient diagnostic and treatment facilities, ambulatory surgery centers, ambulance services, and home healthcare service providers. Private offices of physicians, dentists, or other healthcare professionals are exempt from the FDA’s mandatory reporting requirements, but practitioners can report problems voluntarily to the FDA. The database currently houses over 1.8 million reports. Approximately 95% of these reports are from industry, and the remaining 5% from healthcare facilities and providers. The number of reports submitted has continued to increase, with approximately 180 000 reports submitted per year. A public version of this database is available on the Internet [11]. A second database, which is the predecessor to MAUDE, the Medical Device Report Database (also known as the Device Experience Network Database), is also available on the Internet [12]. The data elements per event include the manufacturer, model-specific device, event and receipt dates, and patient and device problem codes. The FDA uses the information reported through these programs to assist in the early identification and characterization of emerging medical device problems and related public health issues. The reports are used for health hazard evaluations, product assessments, trend analysis, regulatory actions, or for developing effective education programs and timely feedback to healthcare practitioners and medical device manufacturers. Case series studies conducted using MAUDE data looked at the adverse events associated with various devices, including breast implants [13–15], gastric band device [16], pulmonary artery catheterization [17], gloves [18], surgical staplers [19, breast pumps [20], and infusion pumps [21]. The MAUDE database contains nationwide and non-US data on device-related adverse events and it is one of a few sources for detecting previously unknown or rare adverse events. Data are collected using a standardized form and are readily accessible to the public. However, the MDR system is a passive reporting system and, as a result, reports are often incomplete or difficult to understand. Adverse events are underreported due to lack of detection and and/or attribution of device to event, lack of knowledge about reporting system, liability concerns, and lack of motivation to report. Data reflect reporting bias associated with severity or uniqueness, publicity, and litigation. Because of the lack of the denominator data, the estimation of prevalence and incidence are not possible. Finally, because of the incompleteness of the data, it is very difficult to determine the causality. See Chapter 4 for a full discussion of the disadvantages of MAUDE as a surveillance database. While MAUDE is a useful tool to monitor the safety of medical devices on the market by detecting signals of adverse events related to medical devices, it contains limited information for epidemiological studies aimed at examining associations between particular medical devices and adverse events. Other sources of data, therefore, need to be explored to obtain necessary information for such studies.

SURVEILLANCE DATABASES

103

Medical device surveillance network (MedSun) MedSun was created in response to the section of the Food and Drug Administration Modernization Act (FDAMA) [22], which required FDA to move from its mandatory user facility reporting program to surveillance reporting by a subset of clinical facilities. An overall objective was to increase the quantity and quality of user facility reporting through training and education of staff in participating user facilities. The MedSun network currently consists of aproximately 350 healthcare institutions nationwide. MedSun reports come directly from the facilities and often contain a more detailed description of the event, leading to easier assessment of the problem. Because the majority of MedSun reports are of minor injuries and device problems in which no patient injury occurs, the emphasis is on taking appropriate measures to prevent injuries. The MedSun program is described in more detail in Chapter 5.

International vigilance surveillance system A process for the global exchange of vigilance reports between National Competent Authorities (NCAs) has been established through the Global Harmonization Task Force (GHTF), which was established in 1992. Standardized reports about potentially highrisk issues for which action is to be taken are submitted electronically to a shared listserver. The GHTF is a voluntary international consortium of public health officials, responsible for administering national medical device regulatory systems and representatives from regulated industry. Currently, the program exchanges approximately 150 reports per year that include those equivalent to US Class I and high-level Class II recalls, all public health notifications, and special public health concerns (e.g. high index of preventability, or particularly vulnerable populations).

US consumer product safety commission surveillance systems National Electronic Injury Surveillance System (NEISS) NEISS is a sentinel system operated by the US Consumer Product Safety Commission (CPSC), designed to capture information on product-related injuries that result in emergency department (ED) visits (see Chapter 6). Information in NEISS is abstracted from the ED patient record by data management-trained hospital personnel to produce a NEISS case report, which consists of coded variables and a brief narrative description of the adverse event incident. Each of the hospitals in the NEISS sample is selected to represent many similar hospitals across the nation (stratified random sample of all US hospitals with at least six beds that provide a 24 hour emergency service; included are four strata that represent hospital size, measured by number of ED visits, and an additional stratum that contains children’s

104

CH8

DATA SOURCES FOR MEDICAL DEVICE

hospitals). Therefore, case reports incorporated into NEISS from individual hospitals are differentially weighted to produce national estimates and associated variances for the number of adverse events related to specific products [23]. NEISS was used to produce the first-ever, statistically valid national estimates of adverse events associated with medical device products [24].

Centers for Disease Control and Prevention surveillance systems The Centers for Disease Control and Prevention (CDC) maintain multiple surveillance programs that provide opportunities to study specific medical devices.

National Nosocomial Infections Surveillance System (NNISS) This voluntary surveillance system was established in early 1970 to monitor the incidence of nosocomial infections and their associated risk factors [25]. This is the only national system that tracks nosocomial infections. It currently receives reports from approximately 300 hospitals, high-risk nursery and surgical units. The surveillance data are used to detect and monitor adverse events, examine risk and protective factors, design and evaluate preventive measures and help implement effective prevention strategies. The system is being redesigned to include a web-based knowledge management and to make adverse events reporting that will be available to all US hospitals, long-term-care facilities, and other healthcare organizations as a National Healthcare Safety Network. This surveillance program can potentially be used to study various healthcare-related infections associated with medical devices [26–29]. Dialysis Surveillance Network In 1999, the CDC created the Dialysis Surveillance Network as a voluntary national network of adult and pediatric dialysis facilities. The purpose was to provide a method for individual hemodialysis centers to record rates of vascular access infections, other bacterial infections, disease control and prevention measures, and to provide rates for comparisons among various dialysis centers. Participating centers enter the data in a user-friendly form or use the Internet-based system for data entry. A computer algorithm then determines whether the infection case definitions are met. The participation in this system is free and the reports can be generated as frequently as the participating center needs. The summary data are released, but the data from the individual centers are only released to the dialysis center reporting it [30,31]. The information in these surveys regarding devices used during dialysis is so general that they are of very limited utility for medical device epidemiology or surveillance.

REGISTRIES

105

Registries The literature indicates that one of the first public registries was established in England and was known as The Domesday Book, which dates back to 1086 [32]. In the world of life sciences, a ‘registry’ usually refers to a health-related database. There are numerous definitions for a registry in the scientific literature. The basic definition of the registry describes it as an ‘observational database designed to reflect current patterns of practice without influencing the treatments or interventions’ [33]. Solomon et al. (1991) defined a registry as ‘a database for the collection of a specific set of demographic and health data on identifiable persons with a specific public health purpose’ [32]. The last couple of decades have seen a dramatic increase in the use of registries, primarily due to the increased need for the evaluation of healthcare-related information and technological development in data collection and storage at the registry framework. Nowadays, registries can be used for public health monitoring of various diseases/ conditions, treatment effectiveness (drugs/devices), treatment compliance, quality of life measurements, or incidence, prevalence and survival data. Registries can be observational (descriptive and analytical) or interventional. They can be designed to prospectively or retrospectively address specific question. In addition, once the framework of the registry is developed, studies with various designs can be conducted within the registries (cohort, case-cohort, case-control, cross-sectional, quasi-experimental). The design of registries may or may not include a comparison group; if a comparison group exists, it can be internal, external or historical. The sample size calculations of the registries depend on the study question, study design and scale of measurement of the variables in the study. Sampling and allocation schemes that may be utilized in a registry include matching, restriction, probability sampling, and non-probability sampling. The general strengths of registries include the ability to collect data on large numbers of people rapidly and efficiently. Registries are useful for manufacturers to monitor their products’ performance within the study population. They enable clinicians and researchers to compare their patient population with those of other researchers [32]. Registries also can have limitations. The development of a registry is a complex process that includes implementation and documentation plan, determination of the data variables, data collection, processing procedures and access policies, case definition and ascertainment procedures, quality check/control procedures and a method for disseminating the registry data/findings to the stakeholders [32]. The implementation of a registry has challenges, being time consuming and expensive to maintain. In addition, poor design and rationale of registries can lead to misleading results. Although registries are usually large, they are never fully representative of the population. Issues that may affect the validity of studies and should be controlled include information bias, selection bias, and loss to follow-up. Also, registries are often limited due to over-control in the dissemination of registry findings by the sponsors.

106

CH8

DATA SOURCES FOR MEDICAL DEVICE

Registries sponsored by professional organizations American College of Cardiology (ACC) National Cardiovascular Data Registry (NCDR) The ACC NCDR was developed in the late 1990s and is now expanding to over 700 institutions nationwide who submit cardiac catheterization data to the registry. These data include over 140 core elements needed for measuring the clinical management and outcomes of patients undergoing diagnostic cardiac catheterizations and percutaneous coronary interventions. CDRH was the first governmental entity to do collaborative research with ACC by using their NCDR. The largest observational studies of hemostasis devices in general, and manufacturer-specific devices, have recently been completed in a collaborative effort between CDRH and ACC [34,35]. As new device-related issues emerge, the CDRH continues to consider potential collaborations with ACC. National ICD Registry The National ICD Registry is a collaborative effort of ACC and the Heart Rhythm Society. The registry is designed to be multicenter and national in scope and collect information from hospitals on patients undergoing implantable cardioverter defibrillator implantation procedures. The CDRH entered the collaboration with ACC in 2005 to study the experiences of patients with implanted ICDs and associated risk factors, such as demographics, preoperative risk factors, previous interventions, preoperative cardiac status, preoperative hemodynamics, operative procedure, complications and postoperative status. In late 2005 the CMS announced that this registry would be a reporsitory for information on 1300 hospitals nationwide that perform ICD implantation procedures, and those hospitals are expected to report all procedures among Medicare beneficiaries and are encouraged to report all ICD-related procedures [36]. The strengths of the National ICD Registry include its national scope and potential to enroll over 100 000 patients annually. Comparisons will be enabled nationally and between the hospitals. This registry is audited under, and compliant with, the Health Insurance Portability and Accountability Act of 1996 (HIPAA) to ensure the protection of patient confidentiality and privacy [37]. The quarterly and annual benchmark reports and quarterly newsletters issued by the registry are important tools for disseminating information on comparisons to the national standards. One of the major limitations of the registry is the lack of the longitudinal data collection that can be used beyond the device implantation [36]. Society of Thoracic Surgeons (STS) National Adult Cardiac Surgery Registry The STS registry was developed in 1990 and currently collects data from approximately two-thirds of all US cardiothoracic hospitals (over 700) and contains detailed data on

REGISTRIES

107

patient demographics, clinical profile, and acute outcomes on more than 2.1 million procedures. CDRH collaborated with the Duke Clinical Research Institute (the data coordinating center) in use of STS data to assess the use and outcomes of transmyocardial revascularization (TMR) [38].

National Clearinghouse of Plastic Surgery Statistics (sponsored by the American Society of Plastic Surgeons) The American Society of Plastic Surgeons (ASPS) has been collecting data on breast augmentation and reconstruction, along with other plastic surgery procedures, since 1992. ASPS has partnered with DataHarbor Inc., a healthcare industry data management and technology development company, to collect, analyze, and present the data which are kept in the National Clearinghouse of Plastic Surgery Statistics [39]. This database is based on surveys sent to more than 17 000 board-certified physicians (mostly in the USA and Canada), and is focused on type of surgery procedures, including procedures with breast implants. No information on adverse events was collected. The CDRH used the registry to obtain information on statistical trends of cosmetic and reconstructive plastic surgery.

Cosmetic Surgery Telephone Survey (sponsored by the American Society of Plastic Surgeons) ASPS has also collected data on breast augmentation and reconstruction along with other plastic surgery procedures with national telephone surveys conducted by TeleNation, a service of MarketFacts, Inc. Each week TeleNation completes two national telephone surveys, including the Cosmetic Surgery Telephone Survey [40]. The survey consists of a minimum of 1000 interviews with adults 18 or older (500 males and 500 females) in the contiguous USA. TeleNation interviews are conducted over a 3 day period in the MarketFacts National Telephone Center. The Cosmetic Surgery Telephone Survey contains information on demographics (age, gender, household income, marital status, location of residence, education), view of cosmetic surgery (approve of cosmetic surgery for myself and others; approve of cosmetic surgery for others but not myself; undecided; and disapprove of cosmetic surgery), views of cosmetic surgery compared with 10 years ago (same attitude toward cosmetic surgery now as 10 years ago; more favorable now than 10 years ago; and less favorable now than 10 years ago), type of procedures considered (thighs; facial wrinklers; buttocks; breasts; nose; none of the above) and factors affecting the decision (being content with your current appearance; money; fear of surgery; what others think; none of the above). This database is limited to data on demographics and trends in cosmetic and reconstructive surgery procedures, including breast implants, but contains little information for medical device epidemiology research.

108

CH8

DATA SOURCES FOR MEDICAL DEVICE

Cosmetic Surgery National Data Bank (sponsored by the American Society for Aesthetic Plastic Surgery) The American Society for Aesthetic Plastic Surgery (ASAPS) has been collecting data on cosmetic surgery procedures, including breast implants, since 1997. In 2005, silicone breast implants were added to the list of procedures of this database, Cosmetic Surgery National Data Bank. The data collection was based on physician self-administered mail survey. Similar to the data collected by ASPS, this database focus on information on demographics and type of procedures, but not information on adverse events or complications. Hence, it has limited value for medical device epidemiology studies [41].

Fibroid Registry for Outcomes Data (FIBROID) registry The FIBROID registry represents a unique example of collaboration between a professional medical society, academic investigators, industry and government. The registry was a conducted through cooperative venture between the Society for Interventional Radiology (SIR) Foundation and Duke Clinical Research Institute (DCRI). The representatives of the FDA Center for Devices and Radiological Health (CDRH) were involved in the initial planning of the registry and as the members of the external advisory committee to research questions and to define data points of interest for inclusion. The goals of the FIBROID registry were to: capture high-quality patient safety and effectiveness data for uterine artery embolization (UAE); demonstrate patient volume and rates of acute outcome events for patients undergoing UAE; collect and quantify longitudinal functional and clinical outcomes of patients undergoing UAE; assess and benchmark clinical practice patterns; facilitate data collection for postmarket surveillance and randomized clinical trials (using infrastructure in place for the Registry); collect and quantify use of resources for patients undergoing UAE; develop and refine standards of care for use of UAE for leiomyomas; and monitor risk-adjusted procedural outcomes and provide data to support development and design of randomized clinical trials. The strengths of this registry are its size, over 3000 enrolled patients undergoing a procedure, and the volume of the data collected, including demographic data, reproductive and gynecologic history, and medical history, using the validated Uterine Fibroid Symptom and Quality of Life questionnaire, details of pelvic imaging, procedure-related details, and follow-up for 36 months. The major limitations of this registry include its voluntary nature, potential loss to follow-up, and lack of a comparison group [42].

Government-sponsored registries National Institutes of Health (NIH)

New Approaches to Coronary Interventions registry In 1990, the National Heart, Lung, and Blood Institute (NHLBI) funded a multicenter voluntary registry, New

REGISTRIES

109

Approaches to Coronary Intervention (NACI) [43]. This registry was essentially the continuation of the original (1977–1981) [44] and second (1985–1986) [45] percutaneous transluminal coronary angioplasty (PTCA) registries in evaluating the initiation and evolution of conventional PTCA. The NACI registry prospectively collected patient and lesion information on patients treated with one of eight new devices then entering investigation for coronary intervention, in order to evaluate objectively the safety and effectiveness of these new devices. The procedures and devices studied were directional coronary atherectomy, rotational atherectomy, transluminal extraction atherectomy, Palmaz–Schatz stent, Gianturco–Roubin stent, Advanced Interventional Systems (AIS) excimer laser, Spectranetics excimer laser, and the Spears laser balloon. Common definitions were used for data collection across all devices, and device-specific forms are used to record procedural details. The database structure allowed for data analysis at the patient, procedure, lesion, and device levels, as required to perform in-depth analyses of the immediate and long-term success of new devices. The registry structure also allowed expeditious planning and performance of randomized trials comparing a new device to conventional PTCA. In the mid-1990s, with funding from the FDA Office of Women’s Health, CDRH epidemiologists collaborated with University of Pittsburgh (the registry data coordinating center) to investigate gender differences after coronary angioplasty with the Palmaz–Schatz stent. Both short- (30 day) and long-term (1 year) outcomes were investigated among equal numbers of men and women (about 500 each) from registry data collected during 1990–1994 [46].

Dynamic Registry Following the NACI registry, NHLBI funded the ‘Dynamic Registry’. This multicenter registry is the only formal registry of consecutive percutaneous coronary intervention (PCI)-treated cases that captures both in-hospital and longterm patient outcomes, while characterizing initial procedural strategy and outcome in great detail at the patient and lesion level. Data were collected on approximately 2000 consecutive patients undergoing PCI during three recruitment ‘waves’ across 20 clinical centers, which incorporated an enriched sample of women and minorities. Patients were contacted via telephone interview at 1 year by trained nurse coordinators to assess symptoms, coronary events, and medication status. Informed consent was obtained from all patients, and the study protocol was approved by institutional review boards at the respective clinical sites and at the University of Pittsburgh data coordinating center. The Registry was extended through June 2007 to continue to collect data on prior ‘waves’ of registry patients, including 4 year annual follow-up of 2000 Registry patients who undergo PCI with the first generation of drug-eluting stents and 1-year follow-up of an equal number of patients who will undergo PCI at a time when subsequent generations of drug-eluting stents have penetrated clinical practice. There are no exclusion criteria and data are available for four sequential recruitment ‘waves’ of consecutive patients (3630 patients with 5173 attempted lesions). Data collection included demographics, medical history and risk factors, and detailed coronary angiographic and procedural information. Severe concomitant noncardiac disease included cerebrovascular disease, peripheral arterial disease, renal insufficiency, pulmonary disease, or cancer. Angiographic findings were recorded according to previously reported definitions [47].

110

CH8

DATA SOURCES FOR MEDICAL DEVICE

Armed Forces Institute of Pathology

American Registry of Pathology (ARP), sponsored by the Armed Forces Institute of Pathology (AFIP) The ARP was begun in February 1993 and received final administrative approval by the AFIP Research Committee on June 13 1993 [48]. Departments of Pathology from military and civilian hospitals were invited to participate in this registry. The most valuable data in this registry are its pathology results of a representative variety of explanted breast implants received by the registry. There are more than 175 explanted breast implants in the registry. Capsular tissues are available in approximately 75% of these cases [.]. The data in this registry can be used to address some unanswered questions, such as whether silicone migrates in the body, etc. [48].

Centers for Medicare and Medicaid Services

CMS-mandated National Registries CMS has recently issued national coverage decisions, calling for the collection of prospective data via national registries, on left ventricular assist devices (LVADs), carotid stents and ICDs [36]. As part of the decision, Medicare will cover those patients who meet certain criteria and who are entered into the registry. CDRH will use the data from these registries to better assess the continuing safety, effectiveness and reliability of these devices. The CDRH’s access to the data and relationship with CMS/NIH data monitoring and analysis is under discussion. Manufacturer-sponsored registries Registries are useful to provide additional data on postmarket experiences in support the safety and effectiveness of device in a large number of patients. An example of manufacturers sponsored registry is examined.

Adept1 Registry for Clinical Evaluation Adept1 Registry for Clinical Evaluation (ARIEL) was sponsored by Innovata plc (formerly known as ML Laboratories plc). It was established in the UK in 2000 soon after the company marketed its product, Adept1 Adhesion Reduction Solution (device trade name) or Absorbable Adhesion Barrier (4% icodextrin solution) (device generic name). The objective of ARIEL was to gather and share information on surgeons’ experiences in the use of Adept1 and to monitor adverse events in patients treated with Adept1. Participation in this registry was voluntary, with 253 centers (150 gynecologic surgery centers and 103 general surgery centers) in six European countries. The registry used a five-page physician-administered data collection form to collect patients’ demographics and medical history, surgical procedures, use of Adept1, surgeons’ satisfaction with handling Adept1 and their clinical observations, and complications and adverse

REGISTRIES

111

events during and after surgery prior to discharge. Between February 2000 and December 2003, the ARIEL Registry captured data on 4620 patients who were treated with Adept1 (2882 in the Gynecological Surgery Registry and 1738 in the General Surgery Registry. These 4620 patients represent 8.3% of 55 802 patients treated with Adept1 at the time of closure of this registry. While the registry data covered multiple domains, there was no information on race/ethnicity and post-discharge adverse event data were not collected. Participation of the Registry was voluntary and initially targeted to surgeons from leading centers. There was no information on percentages of patients treated at the leading centers or the small centers, and no information on what constitutes a leading center or a small center. The ARIEL Registry data represented 8% of all patients treated with Adept1 and the data were collected primarily from surgeons at leading centers. The rates of adverse events reported to the ARIEL Registry were much lower that those observed in the clinical trials, which may suggest underreporting. Information to allow evaluation on an association between Adept1 and reported adverse events were incomplete [49–52].

National registries from other countries Many countries outside the USA have established disease-specific national registries. For example, Sweden has over 40 years of experience with these types of registries, such as the National Cancer Registry, the Hospital Discharge Registry, the Medical Birth Registry and the Cause of Death Registry [53]. More recently, registries designed to capture specific procedures involving medical devices have evolved. We present some examples of such registries.

Danish Registry for Plastic Surgery of the Breast In addition to the major databases of breast implants as part of cosmetic and reconstructive surgery, described above, there are breast implant registries in the UK and Scandinavian countries. As a result of a centralized healthcare system and population data collection system in those countries, data collection is much more complete than that in the USA, and the data can be linked to other population data as needed. One of those registries is the Danish registry for Plastic Surgery of the Breast [54]. This registry was established in 1999 and provides plastic surgeons with a nationwide system for the collection of preoperative, perioperative, and postoperative data on women undergoing breast implantation, breast reduction, or mastopexy. The objective of this registry is to examine short-term and long-term local complications and possible health effects, and to contribute to an ongoing evaluation of surgical results and surveillance of the products. Data collection for this registry relies on a self-administered questionnaire focusing on medical history, demographics, and behavioral factors. Preoperative blood samples are drawn for storage. Surgical data, postoperative results, and complications are registered following surgery and at postoperative visits. As of 2002, the registry included 24 private

112

CH8

DATA SOURCES FOR MEDICAL DEVICE

and public clinics, representing more than 80% of the plastic surgery clinics in Denmark. As of November 2001, a total of 1472 women with breast implants were included in the registry. The strengths of this registry include its 80% representation of plastic surgery in the country, preoperative and postoperative results, postoperative complications, and storage of blood samples, which is very critical to address unanswered question about whether silicone breast implants cause long-term adverse events, such as connectives tissue diseases, signs and symptoms. While the database may not be available to the public, its data published in the literature can be used as references or comparisons. The major limitation of this registry is that the data collection is based on a self-administered questionnaire, which may introduce selection bias, recall bias, and information bias. But with centralized healthcare data and population data, some of these biases could be verified and corrected.

Finnish Arthroplasty Registry Data from arthroplasty operations in Finland have been collected since 1980, when the Finnish Orthopedic Association founded the Arthroplasty Register (AR). The Finnish Health Authorities have been responsible for the AR since 1987 and the National Agency for Medicines (NAM) has been in charge of maintaining the central register since 1994. The main goal of the registry is to develop a computerized database to register total joint replacement operations, to enable structured and detailed recording of total joint arthroplasty, and to serve as tool in patient follow-up and data analyses. Regular postoperative follow-ups are at 1, 3, 5, 7, and 10 years. Preoperatively, patient demographics are recorded, including social security number, gender, BMI, diagnosis, previous operations, Harris hip score etc. Perioperatively, data on surgical technique, implant and manufacturers codes, surgical approach, cementing technique, bank bone, complications, etc., are collected. Postoperatively, subjective contentment, late complications, and Harris hip score are recorded. From the collected data, different statistical reports are produced, such as distribution of diagnosis, age, gender, etc. Survival analyses of different implant types can be made accordingly. The central national register provides general survival data to hospitals, which thus are able to follow up. This system can be adapted in all hospitals performing total joint replacement surgery, but it could also be extended for data collection of the arthroplasties in larger communities or the whole country [55].

Automated large administrative databases Large databases can provide useful information, such as background estimates of prevalence and incidence of specific diseases or conditions in the general population (e.g. data collected by health insurance companies), utilization and cost of specific procedures (Healthcare Cost and Utilization Project, Medical Expediture Panel Survey, Medicare and Medicaid data).

AUTOMATED LARGE ADMINISTRATIVE DATABASES

113

Private healthcare databases Databases maintained by managed care organizations can be used for limited types of epidemiologic studies of medical devices. Examples of these databases include Harvard Pilgrim Healthcare [56], Kaiser Permanente Medical Care Program, United Health Group and others [57]. These large databases allow for rare events to be detected. The claims can be longitudinally linked to follow the enrolled patients over time and across the type of procedure or service they receive. Most of the epidemiologic research using healthcare databases was conducted to study prescription medications, and many healthcare plans have cooperative agreements with FDA to study pharmacoepidemiology. Medical devices to be studied are limited to those that can be inferred from a procedure code, which often are not specific to particular devices (see Chapter 3). The potential of these databases to contribute to the study of medical deices epidemiology can be illustrated with several examples from the cardiovascular arena, especially pacing leads, pacemakers, and heart valves [3].

Government agencies databases Centers for Medicare and Medicaid Services (CMS) databases CMS maintains a database of over 30 million Americans who have been enrolled in traditional fee-for-service Medicare, Parts A and B, since 2003 [58]. Part A contains the records for individual medicare hospitalizations, dates of admission and discharge, admitting diagnosis, procedures, and comorbid conditions. All procedures and diagnoses are coded using the International Classification of Diseases (ICD-9 CM) [59,60]. Part B collects information on physician visits, and surgical and diagnostic procedures [58]. These clams also contain unique patient and physician identifiers and billable service that uses Current Procedural Terminology (CPT) Codes [60]. CMS also keeps an enrollment file with demographic information (sex, race, place of residence, date of death) for all enrollees that is updated monthly. One of the strengths of these data is that these linked files essentially constitute a person-specific registry of medical histories [58]. So long as the patient remains in the enrollment system, the records are kept and updated and there is no loss to follow-up. There are many general limitation of the claims-based data. Identification of devices made by particular manufacturers can not be obtained. Claims data do not contain specific details about the procedures (e.g. location and size of the implanted device, side of the body where the device was placed). Other limitations in using these data are caused by the time gap between the procedure (e.g. implant placement) and the time when the data are available for analysis. This gap of time is used for claims processing and adjudication and shortening it would dramatically increase the value of the CMS system for studying the postmarket performance of medical devices.

114

CH8

DATA SOURCES FOR MEDICAL DEVICE

Agency for Healthcare Research and Quality (AHRQ) databases

Healthcare Cost and Utilization Project (HCUP), Nationwide Inpatient Sample (NIS) NIS is the largest all-payer inpatient care database in the USA. It contains data from approximately 8 million hospital stays each year [61]. The NIS is a stratified probability sample of 994 non-federal hospitals in 37 states participating in HCUP, and is designed to approximate a 20% national sample of hospital discharges. NIS includes more than 100 clinical and nonclinical variables for each hospital stay. These include primary and secondary diagnoses, primary and secondary procedures, admission and discharge status, patient demographics (e.g. gender, age, race, median income for ZIP Code), expected payment source, total charges, and length of stay. Demographic and hospital information, as well as ICD-9 CM codes for diagnoses and procedures, are used to estimate in-hospital morbidity and mortality rates associated with various procedures performed in non-federal hospitals in the USA. CDRH has used these data to estimate patient characteristics and inhospital mortality rates associated with aortic valve replacement (both tissue and mechanical) [62] and pacemaker implantation [63]. Because of it large size, NIS allows the analyses of rare conditions, rare procedures and special patient populations. NIS is the only national hospital database with charge information on all patients, regardless of payer, including persons covered by Medicare, Medicaid, private insurance, and the uninsured. For most states, NIS includes hospital identifiers that permit linkages to the American Hospital Association’s database and county identifiers that permit linkages to the Area Resource File. The NIS contains clinical and resource use information included in a typical discharge abstract, with safeguards to protect the privacy of individual patients, physicians, and hospitals (as required by data sources). The NIS can be weighted to produce national estimates. The major limitation to medical device studies is that the procedure codes are the only indication of device use. Healthcare Cost and Utilization Project (HCUP); Kids’ Inpatient Database (KID) KID is a database of hospital inpatient stays for children. As with NIS, the KID is the only all-payer inpatient care database for children in the USA and includes data from 2–3 million hospital discharges for children, from 2500–3500 US community hospitals from 36 states. The KID includes charge information on all patients, regardless of payer, including children covered by Medicaid, private insurance, and the uninsured. The KID contains clinical and resource use information included in a typical discharge abstract, with safeguards to protect the privacy of individual patients, physicians, and hospitals (as required by data sources). The KID can be weighted to produce national estimates. The KID is composed of more than 100 clinical and nonclinical variables for each hospital stay. These include primary and secondary diagnoses, primary and secondary procedures, admission and discharge status, patient demographics (e.g. gender, age, race, median income for ZIP Code), expected payment source, total charges, length of stay, and hospital characteristics (e.g. ownership, size, teaching status) [64].

NATIONAL SURVEYS

115

Academic databases Many universities are exploring the use of public databases to create specific databases suitable for epidemiologic studies. Those databases are often linked to large federal and state administrative databases useful for conducting pharmacopedemiology studies [57]. Their use in the medical devices arena is still limited.

National surveys Centers for Disease Control and Prevention: National Center for Health Statistics The National Center for Health Statistics (NCHS) is responsible for administering various surveys at the national level. Some of these surveys are examined here.

National Health Interview Survey (NHIS) This national survey provides information on the health of the US civilian population [65]. It is a cross-sectional survey with a multistage area probability design to achieve a representative sample of housholds. The questionnaire consists of basic health questions, demographics and current health questions. The CDRH used the 1988 Medical Device Supplement to the to the NHIS to provide prevalence of various implanted devices [66], such as artificial knees [67], internal orthopedic fixation devices [68], intraocular lenses [69], heart valves [70], pacemakers [71], aortic valves [72], breast implants [73], and tympanostomy tubes [74]. National Mortality Followback Survey (1993) The Mortality Followback Program began in 1961 and was administered by the NCHS. The National Mortality Followback Survey (NMFS) collects additional information from a sample of US residents who die in a given year to supplement the death certificate with information from the next of kin or another person familiar with the decedent’s personal history. The acquired information (from telephone or personal interviews conducted by trained field representatives of the US Bureau of the Census) provides an opportunity to study the etiology of disease, demographic trends, or other health issues [75,76]. In 1993, the NMFS included specific questions on the use of several medical devices. As part of the data collection, a study was conducted to evaluate the characteristics of persons aged 65 years or more who received an initial cardiac pacemaker during their final year of life. The goal of the study was to determine how prudently pacemakers were being utilized in this group. This analysis of NMFS data suggested that older persons who had initial pacemakers implanted during their final year of life were

116

CH8

DATA SOURCES FOR MEDICAL DEVICE

not terminally ill, inactive pacemaker candidates, in general, but relatively independent, physically functional candidates who frequently died abruptly. It appeared that the physical, mental, and life expectancy factors recommended for consideration by expert guidelines for the implantation of cardiac pacemakers were generally applied to persons in this group [76].

1988 NMIHS Survey The 1988 National Maternal and Infant Health Survey (NMIHS) offered a unique opportunity to explore the relationship between birth outcomes and medical device use on a population-based nationally representative sample of births that occurred in 1988. It was unique because it contained extensive self-reported information about a woman’s behaviors, her medical and obstetric history and child outcome linked to both the infant’s birth certificate information and the medical and hospital records of her pregnancy and the post-partum period [77]. The 1988 NMIHS was conducted to study factors related to poor pregnancy outcomes in the USA. It drew stratified random samples from infant birth and death certificates and fetal death certificates in 48 states, the District of Columbia, and New York City that occurred in 1988. The final sample included three components: 9953 women who had live births; 5332 women who had infant deaths; and 3308 women who had late fetal deaths. Blacks were oversampled in all three components. Lowand very low-birthweight infants were oversampled in the birth component. An additional component included a Hispanics oversample from Texas, since a third of 1988 Texas births were to Hispanics. In 1988 there were 3 898 922 resident live births that were included in the sampling frame, 38 910 infant deaths, and approximately 15 000 late fetal deaths of 28 weeks or more gestation. Therefore, the overall probability of survey selection was 1 in every 354 live births, 1 in every 6 infant deaths and 1 in every 4 late fetal deaths. Data from the NMIHS were weighted to reflect national estimates. Content included in the 35-page questionnaire included: prenatal care and health habits; delivery; hospitalizations before and after delivery; previous and subsequent pregnancies; mother’s and father’s characteristics; family income; and baby’s health. Questionnaires were also sent to up to seven prenatal care providers, the hospital of delivery, and three hospitals to which mother and baby were admitted before or after delivery. Data collected from hospital and provider questionnaires included all prenatal procedures (including an extensive list of medical device use), hospitalization for delivery, maternal hospitalization before and after delivery, and health status of infant, as well as infant rehospitalization after delivery. The completed NMIHS data link information from four sources: vital records (birth certificate, report of fetal death, infant death certificate); mothers’ questionnaires; prenatal care providers’ questionnaires; and hospital questionnaires. The CDRH used the NMIHS data to study the utilization of home pregnancy tests [78], electronic fetal monitoring and diagnostic ultrasound during pregnancy [79,80], and possible adverse events related to prenatal ultrasound [81].

NATIONAL SURVEYS

117

National Healthcare Survey The National Healthcare Survey (NHCS) is a combination of several healthcare provider surveys that collect information on healthcare services and characteristics of the patients served [82]. The following surveys are routinely administered: National Hospital Discharge Survey (NHDS) and National Survey of Ambulatory Surgery (NSAS) (the only medical device-related data are procedure codes) [83]; National Ambulatory Medical Care Survey (NAMCS) and National Hospital Ambulatory Medical Care Survey (NHAMCS) (the only medical device-related data are procedure codes) [84]; National Nursing Home Survey (NNHS) (includes data on assistive devices) [85]; and the National Home and Hospice Care Survey (NHHCS) (includes data on assistive devices) [86]. These surveys are based on a multistage sampling design that includes healthcare facilities or providers and patient records. Because the data are collected directly from providers, they provide more accurate and detailed data on diagnosis and treatment, as well as on the characteristics of the institutions. These surveys offer data for studying the impact of new medical technologies (including medical devices) changes and to monitor specific diseases/conditions and overall use of healthcare resources, within the tight limits of data available on medical device use.

The National Home and Hospice Care Survey (NHHCS) A national probability sample survey of home health and hospice care agencies, this survey was first conducted by NHCS in 1992 and repeated in 1993 and 1994. The survey was fielded again in 1996, 1998, and most recently in 2000. The NHHCS was implemented as a result of changing trends in alternative sources of care for individuals and families facing long-term and end-of-life healthcare needs. In 2000, the sample consisted of about 1800 home health and hospice agencies and a sample of six current patient records and six discharged patient records from those agencies. The survey includes all types of agencies that provided home health and hospice care, regardless of whether they were Medicare or Medicaid. The data collected depict both the characteristics of these healthcare providers and the people they serve. Agency and patient items include, for example: type of ownership and affiliation; Medicare and Medicaid certification; patient demographics and functional status; diagnoses; services received; types of service providers; patient living arrangements and caregiver; expected sources of payment; and reason for discharge. Data are obtained through personal interviews with agency administrators and staff primarily responsible for the sampled patients’ care. Respondents also refer to patient medical and other records, as necessary. In the past, CDRH has added questions to this survey to look at certain traditional medical devices that are being used in the home, such as ventilators, oxygenators, nebulizers, catheters, infusion pumps, blood glucose monitors, and restraints [86,87]. The general strength of the surveys is that they are population-based, which allows for national estimates of various health issues simultaneously. Another advantage of the surveys is that they can be linked with each other [87]. The major weaknesses of the

118

CH8

DATA SOURCES FOR MEDICAL DEVICE

surveys include high cost, length of the planning stage, and time lag. The use of the surveys to study medical devices is still limited to the most commonly used devices and those mentioned in the data.

Technology assessment databases In the Institute of Medicine Report, technology assessment is defined as ‘any process of examining and reporting properties of a medical technology used in healthcare, such as safety, efficacy, feasibility, and indications for use, cost, and cost-effectiveness, as well as social, economic, and ethical consequences, whether intended or unintended’ [88]. Various aspects of technology assessment are performed by managed care organizations (MCOs), federal and state agencies and non-profit organizations [3]. Some of the databases that are useful for technology assessment are described here.

ECRI technology assessment databases The Emergency Care Research Institute (ECRI) is a nonprofit health services research organization that focuses on healthcare technology, healthcare risk and quality management, and healthcare environmental management. It provides healthcare-related services to a wide variety of customers through its more than 30 databases, publications, information services, and technical assistance services. ECRI gathers and investigates reports of incidents involving medical devices (including capital equipment, reusable and disposable instruments, reagents, etc.). Information is collected from healthcare providers, patients, and manufacturers around the world. As a result of ECRI’s investigations, many manufacturers have recalled or modified their devices [89].

AHRQ technology assessment program The technology assessment program at the Agency for Healthcare Research and Quality (AHRQ) provides technology assessments for the Centers for Medicare and Medicaid Services (CMS) [90]. These technology assessments are used by CMS to inform its national coverage decisions for the Medicare program as well as provide information to Medicare carriers. AHRQ’s technology assessment program uses state-of-the-art methodologies for assessing the clinical utility of medical interventions. Technology assessments are based on a systematic review of the literature, along with appropriate qualitative and quantitative methods of synthesizing data from multiple studies. When available, technology assessment topics are linked to corresponding information on the CMS website. Several recent technology assessments were focused on medical devices technologies, such as cardiac catheterization, magnetic resonance imaging (MRI), positron emission tomography (PET), acupuncture, and screening essays for various diseases.

DATA MINING

119

Data mining Data mining is a statistical technique for extracting meaningful, organized information from large complex databases. There has been increasing interest in the use of datamining technology to enhance the FDA’s ability to monitor the safety of products after they have been approved for use. The software attempts to detect strong, consistent associations that occur at higher than expected frequencies. The Center for Devices and Radiological Health (CDRH) receives more than 100 000 adverse event reports annually [4]. CDRH has begun to use a standard statistical tool to detect signals among these reports and improve upon its current ability to identify adverse event patterns in postmarket safety databases. Data collected from suspected device-related adverse event reports and other electronic medical information could aid the FDA in identifying ‘signals’ of adverse events and the patterns in which they occur. The more effective the tools FDA can use to detect such patterns, the faster it can evaluate and act on these reports as appropriate. For example, more effective data mining might allow the agency to identify a pattern of adverse events in a specific population of patients implanted with a device, and the agency could then communicate this knowledge sooner to medical professionals and patients, so preventing more adverse events. The intent of the data-mining software is to calculate signal scores that represent ‘relative reporting rates’ for adverse events. There are several software applications available for data mining. These include SAS1 Enterprise MinerTM (SAS Institute Inc, Cary, NC), SPSS Clementine1 (SPSS Inc, Chicago, Il.), IBM1 DB21 Intelligent Miner for Data (International Business Machines Corp., White Plains, NY), Microsoft1 SQL ServerTM 2000 (Microsoft Corp., Redmond, WA), DBMiner1 (DBMiner Technology, Inc., Vancouver, BC, Canada), and WebVDMETM (Lincoln Technologies, Inc., Wellesley Hills, MA). These applications provide features such as data preparation to mining and presentation, a visual rapid modeling environment for data mining and support decision trees, clustering, and third-party algorithms. FDA’s Center for Devices and Radiological Health, Center for Drug Evaluation and Research and Center for Biologics Evaluation and Research have all utilized the webVDME approach for data mining. The software has been developed by Lincoln Technologies, a software development and consulting company focused on support of the pharmaceutical industry drug development process. They provide systems and services related to applications in clinical development, pharmacovigilance, electronic data submission, and regulatory review to industry sponsors and to FDA and other public health agencies. The webVDME environment includes support for designing and executing runs of the Multi-Item Gamma Position Shrinker (MGPS) data-mining algorithm, as well as specialized graphical visualization tools and simulation-based facilities for evaluating safety signal detection methodologies. The MGPS utilizes empirical Bayesian methodology to create a model across device–event combinations to reduce the variance for each estimate of relative reporting rate. An empiric Bayes geometric mean (EBGM) is generated, which is an estimate of the relative reporting rate or ratio of observed to expected counts [91].

120

CH8

DATA SOURCES FOR MEDICAL DEVICE

Future use of databases for medical device epidemiology Expanding medical device technology New technologies and innovations continue to lead to rapid advances in medical devices. Medical devices are becoming more sophisticated than ever. We are seeing more devices that are computer-based and information-rich, combined with one or more mechanisms, more precise, less invasive, smaller in size, and easier for patients to use at home. In addition, the population using medical devices is getting broader, owing to the aging population in economically developed countries and progressively expanding globalization. The advances and changes in the world of medical devices pose many challenges to conducting medical device epidemiology studies. As technology advances, the need for databases, especially registries, will increase in order to maximize medical expertise and FDA and industry resources in data gathering and analysis. As the use of databases for adverse event prevention as well as health promotion are more fully appreciated, interest in database development and active participation grow.

Growing database capabilities There is a growing trend to provide more sophisticated programming logic within the database structure. Initially, databases only controlled what kind of data could be put in the fields. Databases then became more sophisticated and features such as triggers, cascading updates, etc. were developed. As technology advances, database management will continue to advance as well. New systems will be more flexible, comprehensive and easier to use.

Increased privacy and confidentiality concerns The global need for information sharing in the area of epidemiology will impose additional challenges in protecting the privacy and confidentiality of patients and providers contributing data to medical device databases. Many countries are in the process of adopting more rigorous requirements regarding data privacy. However, the proper balance must be found between privacy protection and the obligations of the research community to share the findings of epidemiologic studies in a timely fashion.

References 1. Marvick C. Implant recommendations (news). J Am Med Assoc 2000; 283: 269. 2. Brown SL, Bright RA, Tavris DR. Medical device epidemiology and surveillance: patient safety is the bottom line. Expert Rev Med Devices 2004; 1: 89–90.

REFERENCES

121

3. Shatin D, Bright RA, Astor BC. Databases for studying the epidemiology of implanted medical devices. In The Bionic Human: Health Promotion for People with Implanted Prosthetic Devices, Johnson FE, Virgo KS (eds). Totowa, NJ: Humana, 2005: 115–132. 4. Bright RA. Pharmacoepidemiology studies of devices. In Pharmacoepidemiology, 4th edn, Strom BL (ed.). New York, NY: Wiley, 2005: 487–500. 5. Last J. A Dictionary of Epidemiology, 3rd edn, Oxford University Press (USA), 1995. 6. World Health Organization. Public health surveillance: http://www.who.int/immunization_monitoring/burden/routine_surveillance/en/ [accessed May 15 2006]. 7. Kessler DA, Kennedy DL. MedWatch: FDA’s new medical products reporting program. J Clin Eng 1993; 18: 489–492. 8. Medical Device Reporting Regulation. 21CFR803. 9. Safe Medical Device Act of 1990. Public Law No. 101–629, 104, Stat 4511, November 28 1990. 10. Medical Device Amendments of 1992. Public Law No. 102–300, June 16 1992. 11. MAUDE database: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfMAUDE/search.CFM [accessed May 2006]. 12. Medical device report database: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfmdr/ search.CFM) [accessed May 2006]. 13. Brown SL, Parmentier CM, Woo EK, Vishnuvajjala RL, Headrick ML. Silicone gel breast implant adverse event reports to the US FDA, 1984–1995. Publ Health Rep 1998; 113: 535–543. 14. Brown SL, Todd JF, Luu HM. Breast implant adverse events during mammography: reports to the Food and Drug Administration. J Women’s Health 2004; 13: 371–378. 15. Brown SL, Hefflin B, Woo EK, Parmentier CM. Infections related to breast implants reported to the Food and Drug Administration, 1977–1997. J Long Term Eff Med Implants 2001; 11: 1–12. 16. Brown SL, Reid MH, Duggirala HJ. Adjustable silicone gastric banding adverse events reported to the Food and Drug Administration. J Long Term Eff Med Implants 2003; 13: 9–17. 17. Kaczmarek RG, Liu CK, Gross TP. Medical device surveillance: gender differences in pulmonary artery rupture following pulmonary artery catheterization. J Women’s Health 2003; 12: 931–935. 18. Dillard S, Heflin BJ, Kaczmarek RG, Petsonk L, Gross TP. Health effects associated with medical glove use. Association of Operating Room Nurses Journal 2002; 76: 88–96. 19. Brown SL, Woo EK. Surgical stapler associated fatalities and adverse events reported to the Food and Drug Administration. J Am Coll Surg 2004; 199: 374–381. 20. Brown SL, Bright RA, Dwyer D, Foxman B. Breast pump adverse events: reports to the Food and Drug Administration. J Hum Lactation 2005; 21: 169–174. 21. Brown SL, Morrison A. Local anesthetic infusion pump adverse events reported to the Food and Drug Administration. Anesthesiology 2004, 100: 1305–1307. 22. Section 213 of the FDA Modernization Act of 1997, amended §519 (b) of the Food, Drug and Cosmetic Act, 21 USC §360i(b). 1997. 23. Schroeder T, Ault K. The NEISS Sample (Design and Implementation), 1997 to Present. Washington DC: US Consumer Product Safety Commission, Division of Hazard and Injury Data Systems, June 2001. 24. Hefflin BJ, Gross TP, Schroeder TJ. Estimates of medical device-associated adverse events from emergency departments. Am J Prevent Med 2004; 27: 246–253. 25. National Nosocomial Infections Surveillance System: http://www.cdc.gov/ncidod/dhqp/ nnis_pubs.html [accessed May 2006]. 26. Richards MJ, Edwards JR, Culver DH, Gaynes RP. Nosocomial infections in pediatric intensive care units in the United States. National Nosocomial Infections Surveillance System. Pediatrics 1999; 103: e39.

122

CH8

DATA SOURCES FOR MEDICAL DEVICE

27. Bronchoscopy-related infections and pseudoinfections – New York, 1996 and 1998. Morb Mortal Wkly Rep 1999; 48: 557–560. 28. Emori, TG, Edwards JR, Culver DH et al. Accuracy of reporting nosocomial infections in intensive care unit patients to the National Nosocomial Infections Surveillance System: a pilot study. Infect Control Hosp Epidemiol 1998; 19: 308–316. 29. National Nosocomial Infections Surveillance System. Nosocomial infection rates for interhospital comparison: limitations and possible solutions. Infect Control Hosp Epidemiol 1991; 609–612. 30. National Surveillance of Dialysis Associated Diseases in the US. 2002 Report. Semin Dialysis 2005. 31. Electronic reporting of infections associated with hemodialysis. Nephrol News 2005. 32. Solomon DJ et al. Evaluation and implementation of public health registries. Publ Health Rep 1991; 106(2). 33. Kennedy L, Craig AM. Global registries for measuring pharmaco-economic and quality-of-life outcomes. Focus on design and data collection, analysis and interpretation. Pharmacoeconomics 2004; 22(9): 551–568. 34. Tavris DR, Gallauresi BA, Lin B, Rich SE et al. Risk of local adverse events following cardiac catheterization by hemostasis device use and gender. J Invasive Cardiol 2004; 16: 459–464. 35. Tavris DR, Dey S, Albrecht-Gallauresi B, Brindis R et al. Risk of local adverse events following cardiac catheterization by hemostasis device use: Phase II. J Invasive Cardiol 2005; 17: 644– 650. 36. Hammill S, Phurrough S, Brindis R. The National ICD Registry: now and into the future. Heart Rhythm 2006; 3:470–473. 37. The HIPAA Privacy Rule. Medical privacy – national standards to protect the privacy of personal health information: http://www.hhs.gov/ocr/hipaa/privacy.html [accessed May 2006]. 38. Peterson ED, Kaul P, Kaczmarek RG, Hammill BG et al. Society of Thoracic Surgeons. From controlled trials to clinical practice: monitoring transmyocardial revascularization use and outcomes. J Am Coll Cardiol 2003; 42: 1611–1616. 39. American Society for Plastic Surgery. National clearinghouse of plastic surgery statistics: http:// www.asps.org [accessed May 2006]. 40. American Society for Plastic Surgery. Cosmetic surgery telephone survey: http://www.asps.org [accessed May 2006]. 41. American Society for Aesthetic Plastic Surgery. Cosmetic surgery national data bank statistics: http://www.asaps.org [accessed May 2006]. 42. Spies JB, Myers ER, Mulgund J, Hume KM, Strain C. The FIBROID Registry: report of structure, methods and initial results (for the FIBROID Investigators under Contract No. 03R00027101D.) Registry Publication No. 05(06)-RG008. Rockville, MD: Agency for Healthcare Research and Quality, October 2005. 43. Steenkiste AR, Baim DS, Sipperly ME, Desvigne-Nickens P et al. The NACI Registry: an instrument for the evaluation of new approaches to coronary intervention. Cathet Cardiovasc Diagn 1991; 23: 270–281. 44. Detre K, Holubkov R, Kelsey S, Cowley M et al. Percutaneous transluminal coronary angioplasty in 1985–1986 and 1977–1981: the National Heart, Lung, and Blood Institute Registry. N Engl J Med 1988; 318: 265–270. 45. King SB III, Yeh W, Holubkov R, Baim DS et al. Balloon angioplasty versus new device intervention: clinical outcomes. A comparison of the NHLBI PTCA and NACI registries. J Am Coll Cardiol 1998; 31: 558–566. 46. Marinac-Dabic D, Kennard ED, Detre MD, Torrence ME. Baseline characteristics of men and women treated with Palmaz–Schatz coronary stent. Pharmacoepidemiol Drug Safety 1997; 6(suppl 2): S54.

REFERENCES

123

47. Williams DO, Holubkov R, Yeh W, Bourassa MG et al. Percutaneous coronary intervention in the current era compared with 1985–1986: the National Heart, Lung, and Blood Institute Registries. Circulation 2000; 102: 2910–2914. 48. Centeno JA. American Registry of Pathology AFIP Silicone Breast Implant Registry progress report: http://www.afip.org/Departments/environmental/registries/afipsbirpr.html [accessed May 2006]. 49. diZerega GS, Verco SJ, Young P et al. A randomized, controlled pilot study of the safety and efficacy of 4% icodextrin solution in the reduction of adhesions following laparoscopic gynaecological surgery. Hum Reprod 2002; 17: 1031–1038. 50. Menzies D, Hidalgo M, Walz MD et al. Use of icodextrin 4% solution in the prevention of adhesion formation following general surgery: experience from the multicentre ARIEL Registry. Ann R Coll Surg Engl 2006 (in press). 51. Sutton C, Minelli L, Garcia E, Korell M et al. Use of icodextrin 4% solution in the reduction of adhesion formation after gynaecological surgery. Gynecol Surg 2005; 2: 287–296. 52. US FDA. FDA clinical review of Adept@ Adhesion Reduction Solution. Presented at the Obstetrics and Gynecology Device Panel 71st Meeting, Gaithersburg, MD, March 27 2006: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfAdvisory/details.cfm?mtg¼662 [accessed May 2006]. 53. Rosen M, Ericson A. Healthcare registries, a community asset: centralized registries of healthcare data can save life and improve quality of life. Lakartidningen 1999; 96: 3668– 3673. 54. Henriksen TF, Holmich LR, Friis S, McLaughlin JK et al. The Danish Registry for Plastic Surgery of the Breast: establishment of a nationwide registry for prospective follow-up, quality assessment, and investigation of breast surgery. Plast Reconstr Surg 2003; 111: 2182–2189. 55. Nordic Orthopedic Federation 50th Congress, Tampere, Finland, June 7–10 2000: http:// www.actaorthopscand.org/PDFs/NOF_0006.pdf [accessed May 2006]. 56. The relation of household income to mammography utilization in a prepaid healthcare system. J Gen Intern Med 2001; 16: 206–207. 57. Strom BL (ed.). Pharmacoepidemiology, 4th edn. New York, NY: Wiley, 2005. 58. Malenka DJ, Kaplan AV, Sharp SM, Wennberg JE. Postmarketing surveillance of medical devices using Medicare claims. Health Aff (Millwood) 2005; 24: 928–937. 59. ICD-9-CM: International Classification of Diseases, 9th Revision, Clinical Modification. Los Angeles, CA: Practice Management Information Corporation, May 1999. 60. American Medical Association. Current procedural terminology, CPT 2001. Chicago, IL: American Medical Association Press, 2001. 61. Overview of the Nationwide Inpatient Sample (NIS). Agency for Healthcare Research and Quality, May 19 2006: http://www.hcup-us.ahrq.gov/nisoverview.jsp [accessed May 2006]. 62. Astor BC, Kaczmarek RG, Hefflin B, Daley WR. Mortality after aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000; 70(6): 1939–1945. 63. Daley WR, Kaczmarek RG. The epidemiology of cardiac pacemakers in the older US population. J Am Geriatr Soc 1998; 46(8): 1016–1019. 64. Overview of the Kids’ Inpatient Database (KID), May 19 2006: http://www.hcup-us.ahrq.gov/ kidoverview.jsp [accessed May 2006]. 65. Adams PF, Hardy AM. Current estimates from the National Health Interview Survey, United States, 1988. Hyattsville, MD: National Center for Health Statistics. Vital Health Stat 1989; 10(173): http://www.cdc.gov.nschs/data/series/sr_10/sr10_173.pdf [accessed June 2004]. 66. Moss AJ, Hamburger S, Moore RM, Jeng LL, Howie LL. Use of selected medical device implants in the United States, 1988. Advance Data from Vital and Health Statistics, No. 191. Hyattsville, MD: National Center for Health Statistics, 1990.

124

CH8

DATA SOURCES FOR MEDICAL DEVICE

67. Silverman BG, Gross TP, Kaczmarek RG, Hamilton PM. The epidemiology of knee implantation in the United States. Knee 1995; 2: 95–102. 68. Moore RM, Bright RA, Jeng LL, Sharkness CM et al. The prevalence of internal orthopedic fixation devices in children in the United States, 1988. Am J Publ Health 1993; 83: 1028–1030. 69. Sharkness CM, Hamburger S, Kaczmarek RG, Hamilton PM et al. Racial differences in the prevalence of intraocular lens implants in the United States. Am J Ophthalmol 1992; 114: 667–674. 70. Garver DA, Kaczmarek RG, Silverman BG, Gross TP, Hamilton PM. Epidemiology of prosthetic heart valves in the United States. Texas Heart Inst J 1995; 22: 86–91. 71. Silverman BG, Gross TP, Kaczmarek RG, Hamilton PM, Hamburger S. Epidemiology of pacemaker implantation in the United States. Publ Health Rep 1995; 110: 42–46. 72. Astor BC, Kaczmarek RG, Hefflin B, Daley WR. Mortality after aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000; 70: 1939–1945. 73. Bright RA, Jeng LL, Moore RM. National survey of self-reported breast implants: 1988 estimates. J Long Term Eff Med Implants 1993; 3: 81–89. 74. Bright RA, Moore RM, Jeng LL, Sharkness CM et al. The prevalence of tympanostomy tubes in children in the United States, 1988. Am J Publ Health 1993; 83: 1026–1028. 75. National Center for Health Statistics: National Mortality Followback Survey, August 23 2005: http://www.cdc.gov/nchs/about/major/nmfs/nmfs.htm [accessed April 8 2005]. 76. Hefflin BJ. Final-year-of-life pacemaker recipients. J Am Geriatr Soc 1998; 46: 1396–1400. 77. Sanderson M, Placek PJ, Keppel KG. The 1988 National Maternal and Infant Health Survey: design, content, and data availability. Birth 1991; 18: 26–32. 78. Jeng LL, Moore RM, Kaczmarek RG, Placek PJ, Bright RA. How frequently are home pregnancy tests used? Results from the 1988 National Maternal and Infant Health Survey. Birth 1991; 18: 11–14. 79. Moore RM, Jeng LL, Kaczmarek RG, Placek PJ. Use of diagnostic ultrasound, X-ray examinations, and electronic fetal monitoring in perinatal medicine. J Perinatol 1990; 10: 361–365. 80. Marinac-Dabic D, Moore RM Jr, Bright RA, Kaczmarek RG, Placek PJ. The use of prenatal ultrasound in the United States: results from the 1988 National Maternal and Infant Health Survey. J Perinatol 1993; 13: 169. 81. Marinac-Dabic D, Krulewitch CJ, Moore RM. Birth weight in relation to frequent prenatal exposures. Am J Epidemiol 1994; 139: S62. 82. National Center for Health Statistics. The National Healthcare Survey, March 2 2006: http:// www.cdc.gov/nchs/nchs.htm [accessed May 2006]. 83. National Center for Health Statistics. The National Hospital Discharge and Ambulatory Surgery Data, February 2 2005: http://www.cdc.gov/nchs/about/major/hdasd/hdasd.htm [accessed May 20 2006]. 84. National Center for Health Statistics. Ambulatory Healthcare Data, April 21 2006: http:// www.cdc.gov/nchs/about/major/ahcd/ahcd1.htm [accessed May 20 2006]. 85. National Center for Health Statistics. National Nursing Home Survey (NNHS), August 23 2005: http://www.cdc.gov/nchs/nnhs.htm [accessed May 2006]. 86. National Center for Health Statistics. National Home and Hospice Care Survey (NHHCS), August 23 2005: http://www.cdc.gov/nchs/nhhcs.htm [accessed May 2006]. 87. Torrence ME. Data sources: use in the epidemiologic study of medical devices. Epidemiology 2002; 13(3 suppl): S10–14. 88. Institute of Medicine. Assessing Medical Technologies. Washington, DC: National Academy Press, 1985. 89. ECRI, 2006: http://www.ecri.org/ [accessed May 2006].

REFERENCES

125

90. Agency for Healthcare Research and Quality. Technology Assessments: http://www.ahrq.gov/ clinic/techix.htm [accessed May 2006]. 91. Almenoff JS, DuMouchel W, Kindman LA, Yang X, Fram D. Disproportionality analysis using empirical Bayes data mining: a tool for the evaluation of drug interactions in the postmarketing setting. Pharmacoepidemiol Drug Safety 2003; 12: 517–521.

9 Ethical requirements and guidelines for epidemiological studies of medical devices Danica Marinac-Dabic and Suzanne C. Fitzpatrick US Food and Drug Administration, Rockville, MD, USA

Introduction More than two decades ago in his book chapter on ethics and environmental health, Corbett wrote that ‘Essences of right and wrong can be hard to calibrate, hard to determine. However, there is a gauge that not only evokes the best in human spirit but is as practical as any ethical gauge can be. It is called the Golden Rule. Essentially, it tells us to treat others as we treat our loved ones. It tells us not only to do our best, but also assert ourselves if we find someone else who has done ill’ [1]. A recent Institute of Medicine Report stated that, ‘conducting research with human participants is a privilege granted by willing volunteers’ [2]. Clinical research studies must be conducted according to the principles of value and scientific validity and respect for rights, dignity and safety of research subjects. Given the nature and scope of epidemiology and human subject’s involvement in research studies, the ethical concerns are very diverse, and constantly changing with the advent of technology and research methodological developments. Regulations governing the protection of human subjects involved in research help to ensure the procedural application of ethical principles in the research practice. Such regulations have been established for most US government agencies and other organizations that conduct research involving human subjects. For example, if studies involving human subjects are supported by the US Department of Health and Human Services Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

128

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

(HHS), they must be conducted in accordance with HHS regulations. In addition, if the studies involve Food and Drug Administration (FDA)-regulated products (e.g. medical devices), they must also follow FDA regulations. Many professional associations also have established guidelines for the conduct of clinical research studies. We first present a brief overview of bioethics principles and foundations, the applicable international and national regulations, and existing guidelines of some professional associations involved in epidemiologic research studies. Given the diversity of medical devices and research methodologies, and the wide range of risk-related considerations associated with them, we then present regulatory and ethical provisions governing clinical studies of medical devices. Finally, we attempt to identify the areas where further work is needed to sustain and encourage good ethical practice in the field of medical device epidemiology.

Bioethics foundations The beginnings of the development of principles of bioethics date back to the sixteenth century. In 1520 the Royal College of Physicians of London created a penal code for physicians that was in 1543 renamed an ‘ethical code’ [3]. Shortly after the American Medical Association was founded in 1847, the Royal College Code was adopted, with two subsequent revisions in 1957 and 1980 [3]. In 1947, the Nuremberg Code was created as a set of standards by which to judge the human experiments conducted by the Nazis. This code established many basic principles governing the ethical conduct of research involving human subjects [4]. In 1948, the World Medical Association adopted the Declaration of Geneva [5], binding the physician with the words, ‘the health of my patient will be my first consideration’. The International Code of Medical Ethics declared that, ‘a physician shall act only in the patient’s interest when providing medical care which might have the effect of weakening the physical and mental condition of the patient’ [6]. The World Medical Association adopted in 1964 its Declaration of Helsinki: Recommendations Guiding Medical Doctors in Biomedical Research Involving Human Subjects. This declaration consists of ethical principles that provide guidance to physicians and other participants in medical research involving human subjects. This declaration was subsequently revised many times and the latest revision was adopted in Edinburgh in 2000 [7]. It contains 27 basic ethical principles for all medical research and an additional five principles for combining medical research with medical care.

US government human subjects protection regulations General considerations Basic regulations governing the protection of human subjects in research supported or conducted by HHS (then the Department of Health, Education and Welfare) were first

US GOVERNMENT HUMAN SUBJECTS PROTECTION REGULATIONS

129

published in 1974. On July 12 1974, the National Research Act (Public Law 93-348) was signed into law, creating the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research [8]. The National Commission was specifically tasked to identify the basic ethical principles for the conduct of biomedical and behavioral research involving human subjects, and to develop guidelines for conducting human research in accordance with those principles. In 1979, the Commission published Ethical Principles and Guidelines for the Protection of Human Subjects of Research, also known as the Belmont Report, named after the Belmont Conference Center where the Commission met when preparing the report. The Belmont Report identifies three fundamental ethical principles for all human subjects research—respect for persons, beneficence, and justice [9]: Respect for persons. The principle of respect for persons incorporates two different ethical requirements: the requirement to acknowledge each person’s autonomy and the requirement to protect those with diminished autonomy. It is important to keep this principle in mind, because not every human being is capable of self-determination and because some individuals lose this capacity because of illness, disability, or circumstances that restrict liberty. In addition, the immature and the incapacitated may require extra protection until they mature or while they are incapacitated. In research involving human subjects, this principle translates into a requirement that subjects enter into the research voluntarily and with adequate information. Beneficence. This principle demands that persons be treated in an ethical manner and that the researcher make every effort to assure their well-being. Two general rules guide this principle. The first is ‘do not harm’ and the second involves maximizing possible benefits and minimizing possible harms. However, during the process of learning about what is harmful and what is beneficial, persons may be exposed to risk of harm. The question then becomes, what is a justifiable level of risk while searching for benefits and determining what level of risk exceeds the benefits? Clinical investigators are obliged to carefully examine the benefits and the reduction of risk that might occur from their research. In addition, this principle is applicable to the larger society, and therefore it is obligatory to recognize the longer-term benefits and risks that may result from the improvement of knowledge and from the development of novel medical, psychotherapeutic, and social interventions. Justice. This principle refers to the fairness of distribution of benefits and risks of research investigations. Principles of fairness dictate for each person an equal share, to each person according to individual need, to each person according to individual effort, to each person according to societal contribution, and to each person according to merit [9]. The selection of research subjects therefore needs to be carefully reviewed to determine whether some groups of people (e.g. particular racial and ethnic minorities, or persons confined to institutions) are being systematically selected simply because of their easy availability or their compromised position, rather than for reasons directly related to the problem being studied. The principle of justice also

130

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

requires that when the research is performed using public funds, it provide advantages to all members of society.

HHS human subjects protection regulatory requirements Based on the Belmont Report, HHS expanded its regulations for the protection of human subjects in the late 1970s and early 1980s. The HHS regulations are codified at 45 CFR part 46, subparts A–D [10]. The regulations incorporated in 45 CFR part 46 are based in large part on the Belmont Report and were written to offer basic protections to human subjects involved in both biomedical and behavioral research conducted or supported by HHS. The HHS regulations (45 CFR part 46) apply to research involving human subjects, conducted by the HHS or funded in whole or in part by HHS. In 1991, 17 Federal departments and agencies adopted a uniform set of rules for the protection of human subjects, identical to subpart A of 45 CFR part 46 of the HHS regulations. This uniform set of regulations is the Federal Policy for the Protection of Human Subjects, informally known as the ‘Common Rule’ [10]. This regulation defines the key terms related to human subjects protection research [10], as follows: Research is a systematic investigation, including development, testing, and evaluation, designed to develop or contribute to generalizable knowledge [11]. Human subject is a living individual about whom an investigator conducting research obtains (a) data through intervention or interaction with the individual, or (b) identifiable private information [12]. Identifiable private information includes information about behavior that occurs in a context in which an individual can reasonably expect that no observation is taking place, and information which has been provided for specific purposes by an individual and which the individual can reasonably expect will not be made public (e.g. a medical record) [13]. Each institution engaged in (non-exempt) HHS-supported human subjects research must provide a written Assurance of Compliance, to the Office for Human Research Protections (OHRP), that it will comply with the HHS human subjects regulations [14]. Institutions conducting non-exempt HHS-supported human subjects research must provide certification to the supporting (funding) agency that the research has been reviewed and approved by an Institutional Review Board (IRB) designated under an OHRP-approved Assurance [15]. Except where the IRB specifically approves a waiver in accordance with HHS regulations, no investigator may involve a human being as a subject in non-exempt research unless the investigator has obtained the legally effective informed consent of the subject, or the subject‘s legally authorized representative. The meaning of ‘legally effective’ and ‘legally authorized’ is determined in part by applicable state law.

FDA HUMAN SUBJECTS PROTECTION REGULATIONS

131

Informed consent must include the eight basic information elements described in the regulations. Information must be presented in language understandable to the subject or the subject’s legally authorized representative [16]. Informed consent must be documented with a written form approved by the IRB and signed by the subject or the subject’s legally authorized representative [17].

Exempt activities There are six categories of research that are exempt from these regulatory requirements. The three relevant to the subject of this chapter include the following: 1. Research conducted in established or commonly accepted educational settings, involving normal educational practices, such as: (a) research on regular or special education instructional strategies; or (b) research on the effectiveness of the comparison among instructional techniques curricula, or classroom management methods [18]. 2. Research involving the use of educational tests (cognitive, diagnostic, aptitude, or achievement), survey procedures, interview procedures, or observation of public behavior, unless: (a) information obtained is recorded in a manner that human subjects can be identified, directly or through identifiers linked to the subjects; and (b) any disclosure of the human subjects’ responses outside the research could reasonably place the subjects at risk of criminal or civil liability or be damaging to the subjects’ financial standing, employability, or reputation [19]. 3. Research involving the collection or study of existing data, documents, records, pathological specimens, or diagnostic specimens, if these sources are publicly available or if the information is recorded by the investigator in a manner that subjects cannot be identified, directly or through identifiers linked to the subjects [20]. Under HHS regulations [21], an IRB may approve a waiver or alteration of informed consent requirements where it finds and documents that: (a) the research involves no more than minimal risk to subjects; (b) the waiver or alteration will not adversely affect the rights and welfare of subjects; (c) the research could not practicably be carried out without the waiver or alteration; and (d) where appropriate, the subjects will be provided with additional pertinent information after participation. The IRB may also approve a waiver of the requirement for written documentation of informed consent under limited circumstances [22].

FDA human subjects protection regulations Although DHHS and FDA regulations have many elements in common, they also have significant differences (Tables 9.1, 9.2). These two sets of regulations, as related, define

132

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

Table 9.1 Section numbers §56.102 (FDA) §46.102 (HHS)

§56.104 (FDA) §46.116 (HHS)

§56.105 (FDA) §46.101 (HHS)

§56.109 (FDA) §46.109 (HHS) §46.117(c)(HHS)

§56.110 (FDA) §46.110 (HHS)

§56.114 (FDA) §46.114 (HHS)

§56.115 (FDA) §46.115 (HHS)

§56.115(c) (FDA)

§56.120(FDA) §56.124 (FDA)

HHS and FDA IRB regulations [24]

Description FDA definitions are included for terms specific to the type of research covered by the FDA regulations (test article, application for research or marketing permit, clinical investigation). A definition for emergency use is provided in the FDA regulations FDA provides exemption from the prospective IRB review requirement for ‘emergency use’ of test article in specific situations. HHS regulations state that they are not intended to limit the authority of a physician to provide emergency medical care FDA provides for sponsors and sponsor-investigators to request a waiver of IRB review requirements (but not informed consent requirements). HHS exempts certain categories of research and provides for a Secretarial waiver Unlike HHS, FDA does not provide that an IRB may waive the requirement for signed consent when the principal risk is a breach of confidentiality because FDA does not regulate studies which would fall into that category of research. (Both regulations allow for IRB waiver of documentation of informed consent in instances of minimal risk) The FDA list of investigations eligible for expedited review (published in the Federal Register) does not include the studies described in category 9 of the HHS list because these types of studies are not regulated by FDA FDA does not discuss administrative matters dealing with grants and contracts because they are irrelevant to the scope of the Agency‘s regulation. (Both regulations make allowances for review of multi-institutional studies) FDA has neither an assurance mechanism nor files of IRB membership. Therefore, FDA does not require the IRB or institution to report changes in membership whereas HHS does require such notification FDA may refuse to consider a study in support of a research or marketing permit if the IRB or the institution refuses to allow FDA to inspect IRB records. HHS has no such provision because it does not issue research or marketing permits FDA regulations provide sanctions for non-compliance with regulations

‘human subject’ and ‘research’ differently, and they also have different waivers and exceptions. For device studies, the practical implications of these differences relate to whether there is need for IRB review and approvals and whether Informed Consent is necessary [11,12,23].

OTHER PROFESSIONAL ETHICAL GUIDELINES

Table 9.2

133

HHS and FDA regulations on informed consent [24]

Section numbers

Description

§50.23 (FDA)

FDA, but not HHS, provides for an exception from the informed consent requirements in emergency situations. The provision is based on the Medical Device Amendments of 1976, but may be used in investigations involving drugs, devices, and other FDA-regulated products in situations described in §50.23 HHS provides for waiving or altering elements of informed consent under certain conditions. FDA has no such provision because the types of studies which would qualify for such waivers are either not regulated by FDA or are covered by the emergency treatment provisions (§50.23) FDA explicitly requires that subjects be informed that FDA may inspect the records of the study because FDA may occasionally examine a subject’s medical records when they pertain to the study. While HHS has the right to inspect records of studies it funds, it does not impose that same informed consent requirement FDA explicitly requires that consent forms be dated as well as signed by the subject or the subject’s legally authorized representative. The HHS regulations do not explicitly require consent forms to be dated

§46.116(c) and (d) (HHS)

§50.25(a)(5) (FDA) §46.116(a)(5) (HHS) §50.27(a)

Other professional ethical guidelines More than a decade ago, the National Academy of Science report entitled Responsible Science noted that ethical issues received little attention in professional societies and called for greater efforts to enhance ethical sensitivity by these groups [25]. Many professional organizations also recognized the need for the development of guidelines focusing on the ethical issues and procedures that govern the conduct of human subjects research. Below is a brief description of one of these guidelines.

World Health Organization (WHO) Good Clinical Practice guidelines The purpose of WHO’s guidelines for Good Clinical Practice (GCP) for trials on pharmaceutical products was to set internationally applicable standards for the conduct of such biomedical research on human subjects. They are based on the national guidelines of several countries, including Australia, Canada, the European Community, Japan, Denmark, Finland, Iceland, Norway, Sweden, and the USA. The overall objective was to provide a complementary standard that could be applied globally without replacement of national policies. These guidelines target clinical investigators; ethics review committees, pharmaceutical manufacturers and other sponsors of research, as well as regulatory bodies [26].

134

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

Epidemiologic ethical guidelines Epidemiology, defined as the ‘study of the distribution and determinants of healthrelated states or events in specified populations, and the application of this study to control of health problems’ [27], is often referred to as a ‘mother science of public health’ [28]. In the late 1980s Soskolne stated that ‘epidemiology is at best in its adolescence and that code of conduct may have to await until the epidemiology reaches its maturity’ [29]. While major developments have occurred in the field, the ethics issues are still not adequately addressed. A potential reason for this is that epidemiology practitioners come from a number of disciplines, bringing to the practice their differences in training, making ethical considerations more diverse. Some of the fields contributing to the epidemiology work force are medicine, nursing, epidemiology, public health, anthropology, and statistics, and their focus and nature of activities are very different from each other. Although some of the professions (e.g. medical) have their code of ethics to which the members must subscribe, the oath to the medical code of ethics (as noted by Soskolne [29]) focuses more on the healthcare practitioner–patient relationship and is not directly relevant to the practice of public health- and population-based research. The first textbook chapter on ethical issues in the field of epidemiology did not appear until 1986 [30]. During the late 1980s it was increasingly clear that there were insufficient ethical guidelines for epidemiologic research [29,31,32]. During the last 15 years, considerable efforts have been made by epidemiologists and ethicists to produce professional ethics guidelines for epidemiologists. As noted by Coughlin [33], these developments have occurred in conjunction with major public health developments, such as improvements in regulatory requirements for protection of human subjects, rising concerns about the confidentiality of medical records, the emerging AIDS epidemic, and advances in genetic epidemiology. During this time, guidelines were prepared by the Industrial Epidemiology Forum (IEF) in 1989 [34], the International Epidemiological Association in 1990 [35], the Council for International Organizations of Medical Sciences (CIOMS) in 1990 [36], and for the International Society for Environmental Epidemiology in 1996 [37]. Recently, the American College of Epidemiology (ACE) asked its Ethics and Standards of Practice (ESOP) Committee to produce ethics guidelines [38]. Each of these guidelines are briefly summarized below.

Industrial Epidemiology Forum (IEF) guidelines These guidelines recognize and describe the obligations of epidemiologists to research subjects, society, funding agencies, employers, and professional colleagues [34]. Obligations to research subjects include: protecting their welfare; obtaining informed consent; protecting privacy; maintaining confidentiality; and reviewing research protocols. Obligations to society include: avoiding conflicting interests; avoiding partiality; widening the scope of epidemiology; pursuing responsibilities with due diligence; and

OTHER PROFESSIONAL ETHICAL GUIDELINES

135

maintaining public confidence. Obligations to those funding research and employers as well as those to colleagues are similarly specified [34].

International Epidemiologic Association (IEA) guidelines IEA ethics guidelines are available only in draft form. They are made up of nine distinct sections. The first two include the definition and purposes of epidemiology and the core values of epidemiology. Following these is a section on basic principles of biomedical ethics – autonomy, beneficence, nonmaleficence, and justice – which also mentions the Helsinki Declaration. The following sections center around obligations to individuals, obligations to communities, and access to information. The last sections discuss scientific integrity, professional standards, and cultural variations in values [35].

Council for International Organizations of Medical Sciences (CIOMS) guidelines Adopted in 2002, these are the third biomedical research ethical guidelines issued by CIOMS since 1982. This document consists of 21 guidelines with commentaries. It also provides lists of items to be included in the research protocol submitted for scientific and ethical review and clearance. Appendices include also the World Medical Association’s Declaration of Helsinki. The guidelines discuss the following topics: ethical justification and scientific validity of research; ethical review; informed consent; vulnerability of individuals, groups, communities and populations; women as research subjects; equity regarding burdens and benefits; choice of control in clinical trials; confidentiality; compensation for injury; strengthening of national or local capacity for ethical review; and obligations of sponsors to provide healthcare services [36]. The 2002 CIOMS guidelines are designed to help define national policies on the ethics of biomedical research involving human subjects, apply ethical standards in local circumstances, and establish or improve ethical review mechanisms. Particular attention was given to the conditions and the needs of developing countries involved in research [36].

International Society for Environmental Epidemiology (ISEE) guidelines These guidelines were prepared by ISEE but not officially adopted. The foundation of these guidelines is in the IEF guidelines, with the addition of core values and a definition of environmental epidemiology. They also provide additional components to the general obligations featured in the original IEF guidelines, such as publishing methods and results [37].

136

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

American College of Epidemiology (ACE) guidelines These guidelines were prepared by the Ethics and Standards of Practice (ESOP) Committee on behalf of the American College of Epidemiology (ACE) [38]. They examine the core values, duties (obligations), and virtues that should serve as the basis for sound judgment. This document is divided into four components. The first part provides an overview of core values, duties, and virtues in epidemiology and provides concise definitions of these concepts. The second part provides general statements of the obligations that epidemiologists have to various parties. The third part is a more detailed discussion of these guidelines. The fourth part provides a summary and conclusions. Envisioned as a set of ethical and professional obligations to be followed by members of the American College of Epidemiology, these guidelines describe ethical rules and professional standards in the field of epidemiology: 1. Professional role of epidemiologists. 2. Minimizing risks and protecting the welfare of research participants. 3. Providing benefits. 4. Ensuring an equitable distribution of risks and benefits. 5. Protecting confidentiality and privacy. 6. Obtaining informed consent. 7. Submitting proposed studies for ethical review. 8. Maintaining public trust. 9. Avoiding conflicts of interest and partiality. 10. Communicating ethical requirements and confronting unacceptable conduct. 11. Obligations to communities. All these guidelines are built on the same bioethical foundations and are essentially very similar and very much in agreement with each other. They continue to play an important role in ensuring ethical conduct of studies conducted in the field of epidemiology.

Ethical requirements for medical devices epidemiologic studies Medical device studies involving human subjects fall into two major categories: experimental and observational. Experimental studies such as randomized clinical trials

ETHICAL REQUIREMENTS FOR MEDICAL DEVICES EPIDEMIOLOGIC STUDIES

137

typically involve administration of an intervention (e.g. a medical device) to one arm of the study and comparison with a control group (not receiving an intervention ) with respect to the outcome of interest. Epidemiological studies are also referred to as observational studies. In these studies, the investigator only observes the experience of study participants, without influencing it. Examples of observational studies include cohort studies, cross-sectional studies, and case-control studies. The data can be collected using surveys, focus groups, large existing databases, review of medical records, or various plans for prospective or retrospective data collection using other methodological tools. Because in epidemiologic studies the researcher does not influence the experience of the subjects, they are almost always associated with less risk than experimental studies.

Studies of ‘significant’ vs. ‘non-significant’ risk devices Federal requirements that govern investigations of medical devices were enacted as a part of the Medical Device Amendments of 1976 and the Safe Medical Devices Act of 1990. These statutes define the regulatory requirements for medical device development, testing, approval, and marketing [39,40]. Premarket regulatory review of medical devices has a risk-based component. Thus, all clinical investigations undertaken to evaluate the safety and effectiveness of a medical device must be conducted in accordance with the Investigational Device Exemption (IDE) regulation [41]. Certain clinical investigations of devices may be exempt from the IDE regulation [42], except in certain cases [43]. Unless exempt from the IDE regulation, the risk, i.e. ‘significant risk’ (SR) or ‘nonsignificant risk’ (NSR), of the investigational device must be determined. A SR device study is defined [44] as a study of a device that presents a potential for serious risk to the health, safety, or welfare of a subject and: (a) is intended as an implant; or (b) is used in supporting or sustaining human life; or (c) is of substantial importance in diagnosing, curing, mitigating or treating disease, or otherwise prevents impairment of human health; or (d) otherwise presents a potential for serious risk to the health, safety, or welfare of a subject. Examples of SR devices include respiratory ventilators, implantable cardioverter defibrillators, vascular hemostasis devices, bone morphogenic proteins, cochlear implants, hemodialyzers, sutures, infusion pumps, falloposcopes, extended wear contact lenses, and implantable prostheses (e.g. ligament, tendon, hip, knee, finger). Investigators studying these devices must submit an IDE to FDA, and the study may not begin until FDA and the IRB approve the study. A NSR study is one that does not meet the definition for a SR study. Some examples of NSR devices include dental filling materials, menstrual tampons (cotton or rayon only), and conventional gastroenterology and urology endoscopes. NSR device studies have fewer regulatory controls than SR studies and are governed by abbreviated requirements [45], including the requirement for IRB approval, but do not need FDA approval. The major differences are in the approval process, record-keeping and reporting requirements. Continuing review of both SR and NSR studies by the IRB is also required, as is the requirement for informed consent [46]. It is important to make a distinction between

138

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

the NSR device studies and the ‘minimal risk’ studies, a term utilized in the IRB regulation to identify certain studies that may be approved through an ‘expedited review’ procedure. By definition, SR studies present more than minimal risk, and thus, full IRB review is necessary for these studies. Some NSR studies may also qualify as ‘minimal risk’ studies, and thus may be reviewed through an expedited review procedure [47]. In general, however, IRB review at a convened meeting is required when reviewing NSR studies.

Clinical epidemiologic studies Although methods differ from experimental studies, these epidemiological studies should use the general scientific and ethical review standards applied in experimental studies. Ezekiel et al. [48] published an overview of requirements that must be considered and met to ensure that clinical research, in general, is ethical. These requirements include: 1. Value. The research must result in enhancements of health or knowledge. 2. Scientific validity. The research must be methodologically sound and rigorous. 3. Fair subject selection. Scientific objectives (and not other criteria) should determine communities selected as study sites and the inclusion criteria for individual subjects. 4. Favorable risk:benefit ratio. Within the context of standard clinical practice and the research protocol, risks must be minimized, potential benefits enhanced, and the potential benefits to individuals and knowledge gained for society must outweigh the risks. 5. Independent review. Unaffiliated individuals must review the research and approve, amend, or terminate it. 6. Informed consent. Individuals should be informed about the research and provide their voluntary consent. 7. Respect for enrolled subjects. Subjects should have their privacy protected, the opportunity to withdraw, and their well-being monitored. As pointed out by the authors [48], these requirements are universal, although they must be adapted to the health, economic, cultural, and technological conditions in which clinical research is conducted. The authors also make a distinction between the essential principles, such as value, validity, fair subject selection, favorable risk:benefit ratio, and respect for subjects are essential ethical principles (without which the research would be considered unethical) and the procedural requirements, such as independent review and

ETHICAL REQUIREMENTS FOR MEDICAL DEVICES EPIDEMIOLOGIC STUDIES

139

informed consent (which are intended to minimize the possibility of conflict of interest, maximize the coincidence of the research with subjects’ interests, and respect their autonomy, but other procedure may also achieve these results) [48]. We examine these ethical requirements below: 1. Value. As in clinical research, the exposure of participants to risks of research in medical device epidemiologic studies requires that the study bewell designed and that enhanced knowledge will lead to improvements to health or well-being. The exposure to risks of epidemiologic research (even if the risk is minimal) can be justified only if the whole society or certain populations will gain knowledge through dissemination of study findings. 2. Scientific validity. For epidemiological research (or any other clinical research) to be ethical, the research must: have a clear scientific objective; be valid and practically feasible; have sufficient power to test the objective; and offer a plausible data analysis plan [48]. The appropriateness of the hypothesis should be assessed prior to, and independent of, the specific research methods [48]. Medical device studies, regardless of the funding source (academia, industry, government), should be designed according to these principles. Very often, conflicting interests of stakeholders have the potential to adversely influence the scientific validity of the study. 3. Fair subject selection. As in clinical research studies [9,48,49], the subject selection for the epidemiologic studies of medical devices must be fair. This means that the selection should be based on scientifically justified criteria, not simply on the convenience of the investigators/sponsors or availability of certain groups of subjects. 4. Favorable risk:benefit ratio. By the nature of epidemiologic studies, the risks to the subjects are less than in clinical research. In cases of epidemiologic research studies of medical devices, the risks mostly center around confidentiality and privacy concerns. However, as with the clinical research studies, the potential benefits to individual subjects must be enhanced, and must outweigh the risks [9,48,49] for the research to be ethical. 5. Independent review. In recent years, IRB review responsibilities have been examined at the national level [50–52]. Recent reports agree that the degree of review and oversight of studies should be proportionate to the degree of risk posed by the study [52–55]. The recent IOM report called for the development of a risk stratification scheme for the review of human research [2]. Minimal risk studies should be reviewed carefully but in an expedited manner. Expedited review procedures generally allow the IRB Chair and one or more members of the IRB to review and approve research protocols without convening the full committee. Expedited review does not mean, however, there should be a decrease in human subject protection in the conduct of the research itself. The review of high-risk studies should be done with additional level of scrutiny. An independent review is a requirement because the parties affiliated with

140

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

the research study can have conflicts of interest. For some research with few or no risks, independent review may be expedited, but for much of clinical research, review should be done by a full committee of individuals with a range of expertise who have the authority to approve, amend, or terminate a study. Independent review of clinical research is also important for social accountability because clinical research often imposes risks on subjects for the benefit of society. In the USA, independent evaluation of research projects is conducted by granting agencies, local IRBs, data- and safety-monitoring boards, or FDA. 6. Informed consent. The purpose of obtaining informed consent in epidemiologic research is to ensure that research participants are fully informed of the purpose and the nature of the study, are aware of who the investigators and sponsors are, know the possible benefits and risks of their participation, the scientific methodology used, any anticipated discomfort, the voluntary nature of participation, and have the opportunity to withdraw at any time without penalty. Each of these elements is necessary to ensure that individuals make rational and free determinations of whether the research trials are consonant with their interests [48]. However, for many medical devices epidemiologic studies conducted solely under HHS Regulations, requirements to obtain the informed consent of research participants may be waived. For example, it is not feasible to obtain the informed consent of individuals in some epidemiologic studies and surveillance programs involving the use of existing large databases maintained for other purposes. In such situations, confidentiality safeguards and other measures should be employed to ensure that no harm can result from the research. 7. Respect for potential and enrolled subjects. Respecting potential and enrolled subjects means that: (a) their privacy must be respected by managing the information in accordance with confidentiality rules; (b) their withdrawal must be permitted without penalty; (c) subjects must be provided with new information of additional risks or benefits discovered in the course of research; (d) the welfare of subjects should be carefully monitored throughout their research participation; and (e) what was learned from the research must be communicated to the subjects, if applicable [48].

Epidemiologic studies using medical records, surveys and existing databases Privacy and confidentiality are key issues in every research study involving human subjects. They are particularly important in epidemiologic studies because these studies often involve the use of medical records, surveys and existing databases. As noted in the previous section, many such studies can be exempted under specific criteria. Martin and Eastwood [56] wrote that technological advances have created a ‘gradual erosion of the concept of confidentiality of medical records’. Confidential medical records that identify individuals are essential to epidemiologic research and practice,

ETHICAL REQUIREMENTS FOR MEDICAL DEVICES EPIDEMIOLOGIC STUDIES

141

including those used in studies of medical devices. As suggested in detail in the Council for International Organizations of Medical Sciences (CIOMS) International Guidelines for Ethical Review of Epidemiological Studies, information about research participants is generally divisible into: 1. Unlinked information, which cannot be linked, associated or connected with the person to whom it refers. In this situation, since this person is not known to the investigator and cannot be known, confidentiality is not at stake; or 2. Linked information, which may be anonymous, non-nominal or nominal. It is considered anonymous when the information cannot be linked to the person to whom it refers except by a code or other means known only to that person, and the investigator cannot know the identity of the person; it is non-nominal when the information can be linked to the person by a code (not including personal identification) known to the person and the investigator; and it is nominal or nominative when the information can be linked to the person by means of personal identification, usually the person’s name [36]. The enactment of the Health Insurance Portability and Accountability Act of 1996 (HIPAA) [57] added additional responsibility to healthcare providers and health plans in the area of human subject research. This Privacy Rule went into final effect in 2003 and is applied to identifiable health information held by health plans, practitioners and databases of medical information [58,59]. In order to assure confidentiality, epidemiologists should restrict access to personal information and store this information in secure environments (e.g. locked file cabinets). These precautions are also applicable to back-up documents stored in off-site locations. Epidemiologists should also gather, store and present data in such a manner as to prevent identification of study participants by third parties. No potentially identifying information should be given to third parties without the express written permission of the participant, unless required by law. Federally funded research must comply with both Privacy Rule and Common Rule, while private research that is not subject to Common Rule must comply with the Privacy Rule. Researchers also must comply with state laws. In the USA, researchers can request certificates of confidentiality from the Department of Health and Human Services agency that funded the research (or, if the research is not federally funded, from the National Institutes of Health). The certificate relieves the investigators from the obligation to comply with some categories of legal demands for disclosure, such as court subpoenas for individual research records, thus enhancing the promise of confidentiality [60]. Computerization has increased public concerns about confidentiality issues in research using various data analyses and linking. We strongly support Thurston’s statement that ‘research curiosity and the convenience of database research cannot justify the suspension of moral concerns about privacy and confidentiality that affect us all as patients and citizens’ [59].

142

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

Computerization has created possibilities to transport large quantities of data for epidemiologic analysis and increased the speed of data analysis and the capacity of data storage. Recent advances in computer technology that helped develop large datasets, and created the ability to link various datasets which contain personal identifiers, have increased concerns about maintaining and protecting the confidentiality of personal information related to an individual’s health. In response to such concerns, various governmental bodies have enacted laws regarding the confidentiality of health information. Medical device epidemiologists should comply with state and national laws regarding confidentiality and privacy, including those pertaining to data sharing or pooling of data. Investigators should unlink personal identifiers as soon as they are no longer needed. If personal identifiers must be kept, a clear justification should be provided to the review committee, along with a detailed description of how confidentiality will be adequately protected.

Future ethical requirements for epidemiologic studies Future policy statements and guidelines should refine ethical requirements for the longterm retention of data in data archives and data sharing, ethical issues in placebocontrolled trials, consideration of broader social and environmental consequences of medical device epidemiologic research, and surveillance systems for medical devices. Future guidelines should also address unique approaches to ethical concerns in subspecialties of epidemiology, including medical device epidemiology, pharmacoepidemiology, molecular epidemiology, genetic epidemiology, etc. The guidelines development should keep pace with rapidly changing technology of maintaining and manipulating data to assure confidentiality and privacy. With an increasing global approach to clinical research through multinational research projects, the guidelines for such research must ensure that ethical issues are maintained, regardless of where the study is being conducted. Finally, education in ethics must become and remain an integral part of the curricula of schools of public health and other education programs in the field of epidemiology.

References 1. Corbett TH. Ethics and Environmental health. In Legal and Ethical Dilemmas in Occupational Health, Lee JS, Rom WN (eds). Ann Arbor, MI: Ann Arbor Science Publishers, 1982; 451–459. 2. Institute of Medicine. Responsible Research. A Systems Approach to Protecting Research Participants. National Academies Press: Washington, DC, 2003. 3. Ad hoc Committee on Medical Ethics, American College of Physicians. Position paper, American College of Physicians Ethics Manual. Part 1. History of medical ethics, the physician and the patient, the physician’s relationship to other physicians, the physician and society. Ann Intern Med 1984; 101: 129–137.

REFERENCES

143

4. Nuremberg Code. In Trials of War Criminals Before the Nuremberg Military Tribunals Under Control Council Law, No. 10, Vol 2, Nuremberg, October 1946–April 1949. Washington, DC: US Government Printing Office; 181–182. 5. World Medical Association. International code of medical ethics. World Med Assoc Bull 1949; 1(3): 109, 111. 6. Declaration of Geneva. Adopted by the General Assembly of World Medical Association at Geneva, Switzerland, September 1948. 7. World Medical Association. Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Subjects (revised). Ferney-Voltaire, France: World Medical Association, 2000. 8. National Research Act (Public Law 93-348), July 12 1974. 9. National Commission for the Protection of Human Subjects of Biomedical or Behavioral Research. The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects of Research. Washington, DC: Government Printing Office, 1979. 10. 45 CFR 46, Subpart A. 11. 45 CFR 46.102(d). 12. 45 CFR 46.102(f)(1)(2). 13. 45 CFR 46.102(f)(2). 14. 45 CFR 46.103(a). 15. 45 CFR 46.103(f). 16. 45 CFR 46.116(a),(b). 17. 45 CFR 46.117. 18. 45 CFR 46.101(b)(1). 19. 19. 45 CFR 46.101(b)(2). 20. 45 CFR 46.101(b)(4). 21. 45 CFR 46.116(c)(d). 22. 45 CFR 46.117(c). 23. 21.CFR 56. 102. 24. Significant differences in FDA and HHS regulations for protection of human subjects: http:// www.fda.gov/oc/ohrt/irbs/default.htm [accessed January 6 2006]. 25. Institute of Medicine. Responsible Science. Washington, DC: National Academies Press, 1996. 26. World Health Organization. Guidelines for good clinical practice for trials on pharmaceutical products. In The Use of Essential Drugs. Geneva: WHO, 1995; Appendix 3. 27. Last JM. A Dictionary of Epidemiology, 3rd edn. New York, NY: Oxford University Press, 1995. 28. Institute of Medicine, Committee for the Study of the Future of Public Health. The Future of Public Health. Washington, DC: National Academies Press, 1988. 29. Soskolne CL. Epidemiology: questions of science, ethics, morality, and law. Am J Epidemiol 1989; 129; 1–18. 30. Tancredi L (ed.). Ethical Issues in Epidemiologic Research. Series in Psychosocial Epidemiology, vol 7. New Brunswick, NJ: Rutgers University Press, 1986. 31. MacMahon B. A code of ethical conduct for epidemiologists? J Clin Epidemiol 1991; 44: 147S–149S. 32. Last J. Professional standards of conduct for epidemiologists. In Ethics and Epidemiology, Coughlin SS, Beauchamp TL (eds). New York, NY: Oxford University Press, 1996; 53–75. 33. Coughlin SS, Ethics in epidemiology at the end of the twentieth century: ethics, values and mission statements. Epidemiol Rev 2000; 22: 169–175. 34. Beauchamp TL, Cook RR, Fayerweather WE, Raabe GK et al. Ethical guidelines for epidemiologists. J Clin Epidemiol 1991; 44: 151S–169S. 35. International Epidemiological Association Guidelines on Ethics for Epidemiologists. In Epidemiology Section Newsletter. Washington, DC: American Public Health Association, 1990.

144

CH9 ETHICAL REQUIREMENTS AND GUIDELINES

36. Council for International Organizations of Medical Sciences (CIOMS). International Ethical Guidelines for Biomedical Research Involving Human Subjects. Geneva: CIOMS, 2003. 37. Soskolne CL, Light A. Towards ethics guidelines for environmental epidemiologists. Sci Total Environ 1996; 184: 137–147. 38. Weed DL, Coughlin SS. New ethics guidelines for epidemiology: background and rationale. Ann Epidemiol 1999; 9: 277–280. 39. Medical Device Amendments, 1976. 40. Safe Medical Device Act, 1990. 41. 21 CFR 812. 42. 21 CFR 812.2(c). 43. 21 CFR 812.119. 44. 21 CFR 812.3(m). 45. 21 CFR 812.2(b). 46. 21 CFR part 50. 47. 21 CFR 56.110. 48. Ezekiel EJ, Wendler D, Grady C. What makes clinical research ethical? J Am Med Assoc 2000; 283: 2701–2711. 49. Levine RJ. Ethics and Regulation of Clinical Research, 2nd edn. New Haven, CT: Yale University Press, 1988. 50. General Accounting Office (GAO). VA research protections for human subjects need to be strengthened. Report No. GAO/HEHS-0000-155.Washington, DC: GAO, 2000. 51. Office of Inspector General (OIG), US Department of Health and Human Services (DHHS). Institutional Review Boards: a time for reform. Report No. OEI-01-97-00193. Washington, DC: DHHS OIG, 1998. 52. Office of Inspector General (OIG). Protecting human research subjects: status of recommendations. Report No.OEI-01-97-00197. Washington, DC: DHHS OIG, 2000. 53. Association of American Universities (AAU). Task Force on Research Accountability. Report on individual and institutional financial conflict of interest. Washington, DC: AAU, 2001. 54. General Accounting Office (GAO). Scientific research: continued vigilance critical to protecting human subjects. Report No. GAO/HEHS-96-72. Washington, DC: GAO, 1996. 55. National Bioethics Advisory Commission (NBAC). Ethical and Policy Issues in Research Involving Human Participants, vol 1. Bethesda, MD: NBAC, 2001. 56. Martin BA, Eastwood MR. The confidentiality of medical records: the right to privacy versus the public interest. Can J Psychiat 1980; 25(6): 492–495. 57. Health Insurance Portability and Accountability Act (HIPAA) of 1996. Public Law 104–191, August 21 1996. 58. 45 CFR 160. 59. 45 CFR 164. 60. Certificates of Confidentiality Kiosk: http://grants1.nih.gov/grants/policy/coc/index.ht [accessed May 20 2006]. 61. Thurston WE, Burgess MM, Adair CE. Ethical issues in use of computerized databases for epidemiologic and other health research. Chron Dis Canada 1999; 20(3): 127–131. 62. Pedersen R. Inside information. Occup Health Daftey Mag 1994; 17(4): 6. 63. Krever H. Report of the Commission of Inquiry into the Confidentiality of Health Information, vols I–III. Toronto: J. C. Thacher, Queen’s Printer, 1980.

10 An industry perspective: medical device epidemiology and surveillance Martha A. Feldman Drug and Device Development Co. Inc., Redmond, WA, USA

Introduction Medical devices have been integral to medical care since ancient times (7000 BC in Europe, 2000 BC in South America). For example, scalpels, slings, splints, and crutches were used in these historical eras. Some of the ancient surgical tools found are made of bronze and obsidian (Figure 10.1). Evidence of successful use of these early medical devices, in brain surgery for example, can be seen in the healed skull trephinations in skeletal remains. As technology and metallurgy have grown, so have the complexity and composition of medical devices. With the relatively recent addition of software to medical devices, their potential uses have grown. Currently the only way a biotechnology-derived therapeutic can be delivered is with a medical device, so this leads to even more potential uses of medical device products. The world of medical devices is simultaneously simple and complex. In some respects, it is easier to get a medical device on the market than to get a drug approved. For one thing, in the USA not all medical device products require clinical studies to support a performance claim in a Food and Drug Administration (FDA) premarket submission. Chemical, electrical, or mechanical performance testing in laboratories can be done before going into clinical testing. This laboratory testing is beneficial because it could reduce the number of human subjects needed in a clinical study. Sometimes the clinical history of a device (from studies outside the USA) allows for a waiver of clinical studies altogether. Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

146

CH10

AN INDUSTRY PERSPECTIVE

Figure 10.1 Bronze surgical knives, Egypt and Mesopotamia, ca. 600–200 permission from the Science Museum/Science and Society Library

BC.

Reproduced with

When the user has hands-on operation of a medical device, market clearance or approval for medical devices may also be more difficult. Early in the device design/ development, it is important to plan the device to avoid possible use error. If this has not been done early, the average user may not be able to use the device properly. Sometimes poor design or other device flaws are discovered only when the device is actually being used in its intended setting. That is one reason why an adverse event may occur,

INTRODUCTION

147

sometimes after the device is already on the market. One has to separate user or operator error from medical device error or malfunction. The issue of operator error may make a technically feasible device less safe if the average operator is not able to use the device properly. Note that some use errors are due to the manufacturer providing inadequate training and/or poorly written instructions for use. This chapter explores the good, the bad, and the ugly steps in the late life cycle of medical devices, i.e. the postmarket (PM) period, which includes: A description of the market for medical devices. A description of device life cycle: *

Preclinical studies.

*

Clinical studies.

*

Condition of approval (CoA) studies.

*

522 Postmarket studies.

*

Surveillance studies.

*

Epidemiologic (observational) studies.

This chapter discusses, with examples, why the continuation of some product assessment in the postmarketing period has become more important in recent years.

Market for medical devices The medical device market in the USA is frequently regarded as a stepchild when compared to the pharmaceutical and biotechnology industries. Consequently, the importance of the medical device industry is frequently underrated [1]: ‘The medical device industry may not grab as many headlines as the pharmaceutical industry and other subsectors of the life sciences field, but an argument can be made that it is the field’s backbone. Estimates place the US industry at US $43 billion in annual sales. The medical device and equipment industry is the largest life sciences subsector in the US, employing more than 320 000 at nearly 6200 locations, with an average annual wage of $52 000.’

In an article in the Journal of New England Technology, Dyke Hendrickson estimates a much higher figure for annual sales in the USA – $70 billion [2]. Another article [3] has similar information: ‘Medical device prognosis growth of 8% per annum is anticipated in the US, where the market is estimated to be worth $75.2 billion US dollars in 2005. Worldwide, the market is estimated over $140 billion’.

148

CH10

AN INDUSTRY PERSPECTIVE

In addition to medical devices being valued as commodities, the medical device industry appears to be resistant to economic slowdown. This may be because the Baby Boomers (those born after World War II, between 1946 and 1964) are more concerned about staying fit than previous generations, and are also more receptive to hightechnology solutions, e.g. having a stent implanted rather than life-long pharmaceutical use. There will also be more alternative treatment sites as patient treatment and care is moved out of the traditional hospital setting to the home, assisted living facilities, and regional treatment centers. While this has become common for some medical devices (e.g. diagnostic devices at regional laboratories and remote reading of radiological images), as technology develops, the types of alternative treatment locations will grow even more.

Potential for growth A market survey of medical equipment in the USA by the Freedonia Group is summarized here. Between 2001 and 2006, growth is put at 8%, to reach US $105 billion [4]. This growth occurs because there are continuous innovations as well as constantly evolving technologies. Austin Weber [5] states that ‘during the next decade, the industry will continue to grow at double-digit rates, thanks to the constant introduction of new designs and novel products that offer better healthcare to patients, reduce hospitalization stays and reduce the cost burden on patients and insurers’.

US export of medical devices From the US Census of 2002 Annual Survey of Manufacturers, I selected six fields that were related to medical devices and looked at the value of their exports. The six fields were: electromedical and electrotherapeutic apparatus; surgical and medical instruments; irradiation apparatus; surgical appliances and supplies; dental equipment and supplies; and ophthalmic goods. For each of these fields, I selected the five highest amounts of exports (Table 10.1). The summary table following Table 1 shows the total amount of exports in billions of US dollars for those six fields in descending order by country. Japan is the largest importer at almost $3 billion, with Canada, The Netherlands, and Germany not far behind.

The product’s life cycle: premarket (preclinical and clinical) and postmarket (PM) studies This is a brief review of the testing that might be done on a medical device. The premarket studies include preclinical (bench and sometimes animal testing) and, when required, clinical studies to add to the understanding of the performance and safety profile for the particular medical device. PM studies are performed after the medical device has been cleared or approved for marketing.

149

THE PRODUCT’S LIFE CYCLE

Table 10.1a

Five biggest export destinations for devices in different therapeutic areas

Therapeutic area

Five biggest export destinations (billions of dollars)/total annual exports (%)

Electrometrical and electrotherapeutic Surgical and medical instruments Irradiation apparatus Surgical appliance and supplies Dental equipment and supplies Ophthalmic goods

Japan 0.80/15.1%

Germany 0.61/11.5%

Netherlands 0.46/8.7%

Canada 0.36/6.9%

UK 0.31/5.8045%

Netherlands Japan 1.23/15.7% 1.04/13.3%

Mexico 0.79/10.1%

Canada 0.78/9.9%

Germany 0.51/6.5%

Germany 0.38/15.5%

Japan 0.36/14.7%

France 0.25/9.9%

Canada 0.17/6.7%

China 0.14/5.8%

Ireland 1.09/21.3%

Canada Japan 0.57/11.12% 0.57/11.1%

UK 0.35/6.9%

Netherlands 0.27/5.2%

Canada 0.18/21.7%

Germany 0.13/16.0%

Japan 0.05/6.4%

France 0.4/4.4%

Mexico 0.4/4.3%

Canada 0.24/20.3%

UK 0.21/17.7%

Japan 0.15/12.3%

Singapore 0.10/8.2%

Mexico 0.06/5.5%

US International Trade Statistics (http://www.census.gov/econ/census02/naics/sector31/). Value of Exports – 2004 NAICS 334510 - Electromedical and Electrotherapeutic Apparatus (mfg.) Value of Exports – 2004 NAICS 339112 – Surgical and Medical Instruments (mfg.) Value of Exports – 2004 NAICS 334517 – Irradiation Apparatus (mfg.) Value of Exports – 2004 NAICS 339113 – Surgical Appliances and Supplies (mfg.) Value of Exports – 2004 NAICS 339114 – Dental Equipment and Supplies (mfg.)

Preclinical studies Preclinical (or ‘nonclinical’) testing consists of bench (or laboratory) testing and sometimes animal testing. Bench tests are conducted to provide valid scientific evidence that the medical device is ready to be tested in animals, in the clinical setting, or in both (Table 10.2). The types and depth of studies performed depend on the intended use of the product and the risk involved in its use. For example, a mechanical product will have different test requirements than an electrical device or an implantable device. Table 10.1b

Total exports in six device areas1

Total exports (billions of dollars) to the 10 largest importing countries2 Japan 2.97 Canada 2.3 Netherlands 1.96 Germany 1.63 Mexico 1.25

Ireland 1.09 UK 0.87 France 0.65 China 0.14 Singapore 0.1

1 Electromedical and electrotherapeutic; surgical and medical instruments, irradiation apparatus; surgical appliances and supplies, dental equipment and supplies, ophthalmic goods. 2 USInternationalTrade Statistics:http://www.census.gove/econ/census02/naics: source,US Census Bureau, 2002 Economic Census. Annual Survey of Manufacturers.

150

CH10

AN INDUSTRY PERSPECTIVE

Table 10.2 Examples of some types of bench testing Components: materials tested for biocompatibility if in contact with patient Stability: to determine expiration dating and optimal storage conditions Electromagnetic interference: to assess potential for harm if device is used near other medical devices Electrical leakage/shielding: to ensure safety for operator and patient Software: do verification and validation testing to work out bugs; as features are added, continue testing to ensure integration will not affect operation

The animal species used in testing is also an important consideration. One consideration is that a determination must be made as to whether rodent-only studies suffice or whether non-rodent species testing is also needed. Depending on the type of device, animal testing is not always possible. For example, an animal model may not exist for the disease or clinical condition for which the device is intended. Some medical devices are made of materials which will be placed on or in the body, and other device materials may be liquids which will be placed on or in the human body. In these cases toxicity and biocompatibility testing may be needed. The decision on what testing is needed depends on several factors, for example, how long the product will be in contact with the body (very short, intermediate, or permanently) and what the contact will be (topical, internal, subcutaneous, intramuscular, intravascular, intrathecal, etc.).

Clinical studies Clinical studies may also be needed for some devices (Table 10.3). The design of the study depends partly on the reason for doing the study, for example, for supporting a submission, for marketing purposes, or for fund-raising. For instance, I had a client who was a physician at a hospital who invented a medical device that would make treating Table 10.3 Some reasons for conducting clinical studies Guidance document for the type of product specifies the need for clinical studies Unresolved issues in preclinical studies that need further clarification, perhaps of a mechanism of action or a principle of operation Safety concerns that cannot be ascertained from animal studies Building a safety profile for the product Demonstration of clinical utility Predicate product submission contained clinical data; subsequent applicants must provide similar clinical data Feasibility (proof of concept) studies may be desired by the company itself to ensure that its product’s design and intended use has merit Marketing evaluation and/or fundraising are other reasons for a company to do a clinical evaluation, even if a clinical study is not required Reimbursement support data

THE PRODUCT’S LIFE CYCLE

151

patients in his specialty easier. We filed an investigational device exemption (IDE) with the FDA for the purpose of obtaining data on three to five patients. These data were used to generate interest in funding a larger study to seek approval for this device. A sponsor of a product may decide whether clinical studies are needed to gain clearance or approval for their product, basing that decision on common sense and on the guidelines that FDA provides. For many devices, clinical validation provides sufficient information for the sponsor to assess clinical utility, for instance, a study that will show usability of the device but is not required for clearance. Some devices and indications require clinical data to support the indication for use in the submission. Once the bench and the animal studies have been completed, the sponsor (note that ‘sponsor’ and ‘manufacturer’ are used interchangeably in this chapter) must review all the data to determine whether the medical device is ready to be tested in humans. Depending on the intended use of the product, the study may be conducted in normal volunteers or in patients with the clinical disease or condition. Study plans are developed based on guidance documents, upon review of predecessor products’ summaries in premarket submissions, and upon studies described in peer-reviewed journals. The minimum contents of an investigational plan are listed in 21 CFR Part 812.25. In general, there are three suggested stages of clinical study for medical devices: feasibility (or proof of concept), pilot, and pivotal. If the product is a nonsignificant risk medical device, only Institutional Review Board (IRB) review and approval of the study is needed to proceed with human testing. For this type of product, and depending on the intended use, perhaps only feasibility and pivotal studies are needed. If the product is a significant risk medical device, then both an Investigational Device Exemption (IDE) and the IRB review and approval are needed, and all three stages are recommended. Some objectives of an early clinical study may be to: (a) assess safety of use of the product; (b) assess effectiveness of the product; (c) assess training requirements and how to address training and validation of training; (d) evaluate the effect on the patient in the event that the medical device operates partially or not at all; and (e) determine whether the protocol is ‘doable’ as written and will provide the kind of data needed to support a submission. The study design will vary based partly upon the risk of the product. It could simply be research into comparable medical devices that may be considered to be ‘substantially equivalent’ (i.e. to one already on the market). A study of a medical device that has no predicate would need to demonstrate that the device is safe and effective for its intended use. The environmental conditions under which medical devices are tested may have an impact on the performance and safety of the product, and therefore must be taken into account when designing the study. Examples of these are humidity, temperature, and altitude. Where the medical device is going to be used may affect the study design. For instance, the wide variety of settings in which medical devices are used may be a consideration in where the study is conducted: hospital in-patient (including OR, ER, laboratories, e.g. interventional cardiology/radiology procedures); outpatients in clinical physicians’ offices; field hospitals (e.g. battlefields, natural disasters); on-site medical treatment (e.g. accidents, sporting events, riots); emergencies (e.g. use of automatic defibrillators

152

CH10

AN INDUSTRY PERSPECTIVE

in public places, use of oxygen in airplanes); and home use. In addition, the training requirements may be different for the same device used in different locations. One examplewould be a chest tube used in an operating room vs. the same chest tube used by a physician in a field situation, where there may be limited antiseptic conditions and suboptimal light. Another example would be an automatic external defibrillator used in a hospital vs. the same device used in an airport.

Reasons for postmarket follow-up While preclinical and clinical testing is important, for various reasons, this type of testing is not always reflective of the performance of the device once it is on the market [6]. The clinical studies may be of short duration and in populations that are different from those in whom the product will actually be used. The environment in which premarket tests are performed may be very different from the environment in which the product will actually be used. Also, the premarket testing may not have assessed the long-term effects of an implantable medical device. These differences between pre- and postmarket use of products may occur even if the device is being used as intended (not off-label). Two examples of the effects of environmental differences are: 1. A company making strips for blood glucose meters performed their preclinical studies as required. The test results from the company’s laboratory showed excellent correlation of blood glucose readings obtained with its strips, as compared to the standard laboratory test for blood glucose. The company received clearance for its product. The day after its first shipment, the company had its first recall. All of the preclinical tests were performed in the laboratory under controlled environmental conditions. When consumers purchased these strips and used them under a wide variety of conditions (e.g. cold/warm, humid/dry) the strips turned black when tested under conditions out of the narrow range of the environmental conditions in the laboratory. 2. A pediatric ventilator was cleared and performed to its specification for several months. Four deaths were reported – all from hospitals in cities that were at least 5000 feet above sea level. Investigation found failure of some O-rings used in the device, as a result of use at high altitude. The device had not been tested for use at all possible use altitudes. These examples provide confirmation that PM follow-up on medical devices is very important for the product’s sponsor. What a sponsor doesn’t know about their product can come back to haunt them. This is a particular problem for US manufacturers because of the highly litigious society in which we live. It is in the sponsor’s best interest to intimately understand their product, and this means performing adequate PM studies of their product in order to understand its performance in the real world. The issuewill not always be whether problems were known, but might be whether the sponsor should have

POSTMARKET STUDIES

153

known about them. If a sponsor has not shown diligence in performing PM follow-up on their device, a jury of the injured party’s peers will rarely be sympathetic to the manufacturer’s plight. Not to be totally negative, PM studies also offer a manufacturer the opportunity to show how their product is better than a competitor’s or to improve their product so that it will be superior to competing products.

Postmarket studies A brief discussion follows of PM condition of approval studies (ordered at the time of approval), Section 522 studies, medical device report (MDR) surveillance, and epidemiological (observational) studies on the use, performance, and safety of medical devices. The goal of PM surveillance is ‘to discover additional safety information on marketed medical and radiation-emitting devices. This can be done by detecting adverse events (both real and potential), estimating adverse event frequencies (both absolute and relative), and by identifying groups at risk’ [7].

Uses for PM surveillance PM testing may be for delineation of an important unanswered question about a marketed device, since premarket testing cannot address all device-related concerns. The PM information can complement premarket data by looking at issues that were not completely answered during premarket testing. PM data collection is a tool that regulatory agencies can use to allow for more rapid approval/clearance of a product if the questions can be addressed in the PM period. PM surveillance can also be used to address new concerns that have arisen from different sources – product complaints, medical device reports, the medical literature, or the news media. The PM studies can evaluate the nature, severity, frequency, and/or distribution in the population of suspected problems. PM studies can also look at malfunction and/or failure of a device after long-term use or incidents of latent sequelae resulting from device use. PM surveillance can also be used to expand the original indication for use by augmenting premarket data, for example, for new or expanded conditions of use. PM studies can be used to assess significant changes in device characteristics or technology. PM testing allows for longer-term follow-up, which may help to evaluate what were rare events during the limited premarket phase, but which may become quite common as the device ages and fails. Most of the attention on PM surveillance usually focuses on the high-risk, Class III devices. However, there may also be serious adverse events associated with devices that are classified as low-risk. Many of these devices are Class II devices that did not require clinical data for support of a premarket notification, i.e. a 510(k) application. These adverse events may be due to a flawed design and/or lack of adequate validation.

154

CH10

AN INDUSTRY PERSPECTIVE

PM condition of approval studies Thomas P. Gross, Director, Division of Postmarket Surveillance at CDRH, gave a presentation (AdvaMed Device Submissions Workshop, May 25 2005) in which he listed the reasons why condition of approval (CoA) studies are needed (Table 10.5). While the premarket data can be used to make the initial decisions about safety and effectiveness and to gain understanding about the risk:benefit ratio on the use of the device, it is not possible, considering the limited population in which these studies are conducted, to develop a complete safety profile on the basis of premarket studies alone. Postmarketing data can improve the understanding of this profile. As part of the premarket review of a product, a determination should be done to see whether a postmarketing plan is needed. The plan should focus on any remaining safety and effectiveness questions that were not fully assessed in the premarket period.

Section 522 studies Another type of PM study is the 522 study. In the USA, the legal authority for this type of PM surveillance can be found in Section 522 of the Federal Food, Drug and Cosmetic Act. As stated in the document ‘Guidance for industry, review staff, and the clinical community: guidance on criteria and approaches for postmarket surveillance’ (Nov 2 1998) [8], the Agency can require a manufacturer to conduct PM surveillance for: 1. ‘Any device of the manufacturer that is Class II or Class III, the failure of which would be reasonably likely to have serious. adverse health consequences or which is intended to be:

Table 10.4 Many available approaches for postmarket surveillance 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

A detailed review of complaint history and scientific literature Nonclinical testing of the device Telephone or mail follow-up of a defined patient sample Use of pre-existing secondary datasets, such as Medicare data Use of registries, such as the Society for Angiography and Interventional Cardiology (SAIC) stent registries, or internal registries or tracking systems Case-control study of patients implanted with or using devices Consecutive enrollment studies Cross-sectional studies (multiple cohorts), non-randomized controlled cohort studies Randomized controlled trials Epidemiologic (observational) studies

This table was constructed using multiple sources, including a guest lecture given by Larry Kessler, ScD, Office Director, Officeof Science and Engineering Laboratories, CDRH, on Postmarket Surveillance of Medical Devices, in the Biomedical Regulatory Affairs Certificate course at the University of Washington on May 7 2002.

POSTMARKET STUDIES

155

Table 10.5 Need for postmarket condition of approval (CoA) studies Address premarket data limitations Gather essential postmarket information * Longer-term performance including effects of re-treatment and product changes * Community performance (clinicians and patients) * Effectiveness of training programs * Subgroup performance * Outcomes of concern: real and potential Balance premarket burdens Address Panel recommendations This table was based on a presentation by Thomas P. Gross, Director, Division of Postmarket Surveillance in the Office of Surveillance and Biometrics at CDRH (AdvaMed Device Submissions Workshop, May 25 2005)

2. Implanted in the human body for more than one year, or: 3. Is a life-sustaining or life-supporting device that is used outside a device user facility.’ The 522 studies are used for cause, that is, if there is a potential issue with the device that the FDA may require the sponsor to study in more depth. Some approaches of PM studies are shown in Table 10.4.

Challenges to performing postmarket surveillance studies through MDRs and adverse event reports One of the biggest challenges in permitting marketing of an innovative product is to assess the continuing safety profile of the product during the PM period. There are problems unique to studying the adverse events caused by medical devices during the PM period. There is no standard nomenclature for device adverse effects comparable to MedRA, the standard nomenclature used for adverse drug reactions. Another problem is the lack of a denominator (i.e. on how many people a device was actually used). A third unique issue is operator involvement, which includes training issues and human factors engineering. The user or operator may not want to report a use error caused by a misunderstanding of instructions use or by the lack of adequate training. These use errors could be caused by poor device design. Another issue is that there are multi-device situations, which makes the evaluation of an adverse event especially complicated. One example of this is the use of an electromagnetic surgical device in a room with portable lights, anesthetic administration machines and electric cautery. A second example would be the use of an implanted defibrillator in a person with an implanted insulin pump. These may be disincentives to reporting an adverse event. Another challenge for surveillance through adverse event reporting has to do with the fact that, as with many products in the general marketplace today, it is rare that one company manufactures all parts of its products, and it is uncommon for all manufacturing activities to take place in one facility. For example, with the more complex devices,

156

CH10

AN INDUSTRY PERSPECTIVE

subassemblies are outsourced, and final assembly is usually performed at one site. For manufacturers of medical devices that are sold worldwide, it is less expensive to have manufacturing sites, especially for the final assembly, closer to the regions in which the final product will be delivered. It is less expensive for the company to manufacture and ship the product to nearby customers. However, it is more difficult to get a complete picture of the safety profile of a medical device when it is manufactured and assembled at several facilities. Adverse event reports coming from several worldwide facilities may not be adequately or accurately coordinated as they come into the manufacturer’s headquarters. As part of the FDA’s field inspection program, a sponsor’s complaint files can be reviewed to determine whether the sponsor failed to report complaints that met the requirement(s) of a Medical Device Report (MDR) [9] that should have been reported. In addition, the FDA may learn of other potential adverse events from anecdotal sources, the news media, or other sources (such as medical examiners, morticians, or consumers).

Epidemiologic studies There are also situations where the device has been marketed for some time and is in general use when some adverse events occur as the product is used in ways that were not anticipated. In these cases, stepped-up surveillance, including performance of observational or epidemiologic studies, will give the sponsors insight into how their products are being used, whether there really is a problem, the nature of the problem, and potential solutions to the problem. Because some of these types of studies may be costly, manufacturers may assess the benefit of performing some or all types of studies and may resist performing the most expensive ones. It will, therefore, be necessary to decide about the relative benefit of performing each type of study. It may be assumed that performance of these types of studies will reveal problems, but in fact these studies may act to exonerate a product or manufacturer if the results can distinguish among, for instance, whether it was use error, factors associated with the illness to be treated, or faculty instructions for use, and not the product per se, that was responsible for injuries. Of course, it is also possible that there are problems with the product. In other instances, the FDA will decide to perform its own epidemiological studies to answer specific and important postmarket questions, although this is not routine.

Summary There are many mechanisms available to manufacturers to address issues that arise with marketed products (Table 10.4). In the case where the product represents a significant therapeutic advance, is a diagnostic tool, or is cheaper, faster, easier or safer to operate than alternatives currently available, postmarket requirements could be imposed as a condition of approval in order to speed the product onto the market. Other measures taken by the manufacturer could include less thorough methods than a study, such as

REFERENCES

157

uncovering a change in the severity, frequency, or number of adverse events by examining complaint files or by searching the MDR database for similar experience with other products. There may be spontaneous reports from patients, professionals, caregivers, and even reports in the news media, that can guide the company in assessing PM issues. The manufacturer and regulatory agencies should work together to develop the most comprehensive, least burdensome method of accomplishing the ongoing task of protecting the public health after a product is marketed. Knowledge about how their product performs is a great asset to the sponsor. The sponsor may improve their product’s market share and protect the company from litigation by developing an intimate knowledge of the strengths and weaknesses of their product. This double-edged sword may also bring to light problems that, hopefully, may be addressed in a way that will enhance the sponsor’s product without prohibitive costs.

References 1. McCurry J. Growing from within. Graying America figures to keep medical device industry operating. Site Selection magazine, September 2004: http://www.siteselction.com/features/ 2004/sep/medical/ 2. Hendrickson D. Biomed rounds: medical device industry displays immunity to economic slowdown. Mass High Tech, July 22 2002: http://www.masshightech.com [accessed August 8 2005]. 3. Weber A. Medical device makers push new frontiers. Assembly Magazine 2004: http:// www.assemblymag.com [accessed September 30 2005]. 4. Brown SL, Bright RA, Tavris DR. Medical device epidemiology and surveillance: patient safety is the bottom line. Expert Rev Med Devices 2004; 1: 1–2. 5. Gross T. Human factors and postmarket surveillance at FDA: http://www.fda.gov/cdrh/humfac/ hufactpg.html [accessed September 30 2005]. 6. Guidance for industry, review staff, and the clinical community: guidance on criteria and approaches for postmarket surveillance: http://www.fda.gov/cdrh/modact/critappr.pdf [accessed September 30 2005]. 7. Medical device reporting: http://www.fda.gov/cdrh/mdr/ [accessed September 30 2005].

11 Perspective from an academic on postmarket surveillance Lazar J. Greenfield Department of Surgery, University of Michigan, Ann Arbor, MI, USA

Introduction At first glance, a federal medical device regulatory program would be an unlikely academic resource and, indeed, academic functions of education and training are not a formal part of the Center for Devices and Radiological Health (CDRH) of the Food and Drug Administration (FDA). However, to succeed in its mission to improve the performance and safety of medical devices requires scientific experience and new knowledge. These are derived from scientific methods, continuing education, and research, each of which requires an academic approach. The practical importance of research is well recognized within the agency, particularly as it relates to device performance and changing technology. Other areas would benefit from change, and some of the processes are reviewed here in terms of weaknesses observed by the author. Many of these are acknowledged by the hard-working professionals within CDRH. Within CDRH, there are many opportunities to improve existing internal processes, to facilitate research by professional staff without administrative interference, to promote fellowships and scholarly activity that will aid recruitment, and to expand information resources to include existing valuable databases in outside agencies and registries. It is unreasonable to expect the FDA to meet the expanding needs of this technology-driven system without additional legislative support and significant increases in resources and funding. Once developed, tested, and approved, medical devices are used by healthcare professionals, and the facility of that interaction is an essential metric of their performance, as is Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

160

CH11

PERSPECTIVE FROM AN ACADEMIC ON POSTMARKET SURVEILLANCE

the reporting of adverse events related to their use. When the device fails to perform as expected, or when there is an adverse event connected to it, accurate data and meticulous analysis are necessary to support any conclusion regarding changes that need to be made. The results of such analyses should be widely communicated to the public, to health professionals, and to device users and manufacturers. The results of studies should also be used to continue to modify and improve the design and performance of the particular device, and to promote progress in device development. Information on device-related human performance should become incorporated in the education and training of current and future healthcare providers. Such information is also a rich resource for academic investigation. Investigative efforts by staff and outside scientists should be encouraged within the boundaries of patient and industry needs for confidentiality. As technology and knowledge continue to expand, the academic foundation of those charged with the use, review, and approval of medical devices must be expanded by continuing education. This chapter is a mixture of my observations and analyses that I made during the time I was a visiting scholar at CDRH and in using the adverse event reporting database. I also discuss some opportunities for academic research in medical device surveillance and epidemiology and educating health care professionals about medical device issues. My perspective is that of an outsider who had an opportunity to examine the use of the FDA adverse event reporting system at the FDA. As such, my impressions do not represent those of the FDA, but are my own thoughts about the potential for improvement of existing adverse event reporting systems and for encouraging a scholarly climate at the FDA. The conduct of scholarly research on medical devices epidemiology and surveillance presents opportunities for FDA scientific staff and for academicians outside of FDA.

Adverse event reporting Historically, our understanding of an adverse outcome of medical care began with the results of an autopsy. More information was gained from microscopic and bacteriologic advances, which clarified what clinicians could learn about the causes of death. For the past 50 years, most hospital surgical services have utilized a weekly review of complications and deaths to improve patient care, and to assure patient safety by review of the institutional systems involved in these events. These conferences to review operative morbidity and mortality are powerful teaching tools and the mainstay of surgical training programs. They have progressed from a focus on blaming to an academic review of the available literature on the subject and analysis of what could be done differently to prevent the problem from recurring. When a medical device is involved in an adverse event, there may or not be pertinent information from a bioengineer or risk manager to explain what is known about the role of the device. If the device is felt to be at fault, the expectation is that the healthcare facility will report the problem to the manufacturer, as required by the Safe Medical Device Act of 1991. The legal basis for the FDA requirements for such reporting has been reviewed in Chapter 2. But here we find a latent weakness in the postmarket surveillance

ADVERSE EVENT REPORTING

161

system, as the process becomes dependent on the diligence of reporting and how the manufacturer handles the information. There are separate regulations that include requirements for timely review and evaluation of complaints and maintenance of records, but manufacturers and distributors are not required to report all complaints or information received. Instead they are expected to decide whether the information constitutes a reportable event. The company is also given latitude to categorize the report as user error in its periodic report to the FDA. This shifts the blame from the device and may shield the company from any obligation to make changes in the device or to accept responsibility for the problem. Companies have even been known to cover up serious adverse events by not reporting them to their own regulatory staff [1]. Since the FDA is more interested in the detection of unexpected serious problems than reports of more common and expected occurrences, such as shearing of central line catheters or bone implants failing to osseointegrate, an alternative reporting method, called summary reporting, is allowed. For FY 2004, there were approximately 99 000 line item reports from 58 manufacturers for about 50 different types of devices, using a summary report [2]. In fact, summary reports have far exceeded individual reports since 2000 and now represent two-thirds of the total number of reports. In summary, although reports to CDRH of adverse events associated with medical devices are a useful mechanism for identifying serious problems and leading to the improved safety and effectiveness of medical devices, there currently exist significant weaknesses in this system. If the key to better management of the risks associated with use of medical devices is accurate, timely, and comprehensive reporting of significant near-misses and adverse events, it is not likely to come from industry itself, which has obvious conflict of interest. So how can this be achieved in the present regulatory environment?

Sources of information for adverse event reporting Effective analysis of medical device-associated adverse events depends on reliable and comprehensive information on the circumstances of the event, as well as the device itself. There is no better evidence of the failure of a ‘top-down’ legislative approach to the problem of adverse event reporting than the response to the Safe Medical Devices Act (SMDA) of 1990. The SMDA required mandatory reporting of device-related adverse events by a broad spectrum of user facilities. The user facilities included nursing homes, hospitals, ambulatory surgical facilities, and outpatient diagnostic and treatment facilities. Despite the FDA’s efforts to train healthcare workers by establishing a network of trainers at the district offices of medical facilities, few reports were received. In 1998, 277 reports of deaths from user facilities were received, in contrast to 980 device-related deaths reported from manufacturers. A similar study by the General Accounting Office found that less than 5% of all adverse events related to medical devices were reported to the FDA [3]. In light of this serious underreporting, Congress allowed the FDA to utilize a ‘subset of user facilities that constitutes a representative profile of user reports’ in an effort to

162

CH11

PERSPECTIVE FROM AN ACADEMIC ON POSTMARKET SURVEILLANCE

improve the quality and quantity of reporting of medical device-associated adverse events. The legislation was called the FDA Modernization Act of 1997 [4]. After a successful pilot study, the resulting new Medical Product Surveillance Network (MedSun) was expanded to its current size of more than 300 facilities. By shifting the focus from the broad universe of device users to the experience of targeted individual institutions, it became possible for the first time to gain some idea of the actual frequency of adverse events in relationship to the total usage of the devices in question (i.e. ‘denominator data’). Many corrective regulatory and non-regulatory initiatives have resulted from reports analyzed in the MedSun program [2]. Additional important lessons learned from this experience include the need for trained risk managers at each participating institution, for resolution of liability and patient privacy concerns, for timely gathering of all the information needed to analyze the event, and for feedback to the institution. But what should be the role of the healthcare provider?

Adverse event reporting by healthcare providers It seems intuitive that the nurse, physician, pharmacist, or other provider would top the list of responsibility for reporting an adverse event, since that individual must usually deal directly with correcting the problem. Although the ethical responsibility is clear, there is no legal requirement for such a report, which remains entirely voluntary. Often, because of the pressures of time and workload, another device is substituted or a workaround is created, and the event goes unreported. Some form of report may be filed and/or a request may be made for the clerk to notify the FDA if the problem is considered serious. Here, the latent problem of information reliability surfaces as a form is filled out with incomplete information by someone who was not directly involved. When an effort is made by FDA reviewers to contact the reporter and improve on the quality of the information, needed additional facts are often unavailable. Despite the effort to facilitate the involvement of health professionals in voluntary reporting in 1993 by the establishment of the MedWatch program [5], voluntary reporting (i.e. reports from medical device users) has accounted for a pitifully small percentage of total reports. In my own examination of the sources of reports in the Manufacturer and User Facility Device Experience (MAUDE) database from 2000–2004, 90% of the total of 242 000 reports were from the manufacturers and distributors, who were legally obligated to report. Only 13 600 reports (5.6%) were voluntary, and of these 11 000 (4.5%) were from user facilities. Only 1500 (0.6%) were from physicians. This situation is not likely to improve in the current climate of peer pressures, time constraints, and malpractice litigation. Therefore, the most effective way to facilitate adverse event reporting and improve patient safety is for the expansion of the MedSun concept to all healthcare institutions, including adding the requirement for appointment of a risk manager to the standards of the Joint Commission on Accreditation of Healthcare Organizations (JCAHO). Such a position would not only improve the risk management and liability status of the institution, but would also improve appropriate data collection and reporting of device-related

POSTMARKET CONDITION OF APPROVAL STUDIES AND SECTION 522 STUDIES

163

adverse events. This individual would also be the logical connection for handling product recall notices, which often are lost in hospital bureaucracies. MedSun should support this expansion by providing the training for risk managers and the development of databases to handle the increase in the number and length of reports received. This raises the question of what to do with the reports generated.

Adverse event report data management At the present time, the first CDRH responders to adverse medical device events consist of an interdisciplinary staff (mostly nurse analysts and biomedical engineers) who must assess the quality and significance of the report. This is a daunting responsibility, since more than 190 000 reports are received annually by CDRH. Fortunately, the majority of the reviewers have the clinical experience and determination to pursue the situations that seem the most important, and spend many hours trying to obtain additional information in order to confirm the gravity of the event. The information on a particular device is collected, and the reports ultimately are made available to the public on the MAUDE website [6]. The most serious limitation to this process is the lack of information on total usage of the device, as mentioned previously, thus precluding the calculation of adverse event rates. An additional problem for the analysts is that much of the expertise on new technology resides in the premarket Office of Device Evaluation. This expertise is focused on the process of approval of new devices, with little involvement in postmarket surveillance or analysis of adverse events. The vertical command and control in separate geographical areas is typical of the FDA’s dated organizational approach. By creation of such isolated, silo-type structures, this arrangement is the antithesis of current state-of-the-art management practices. What is needed is a more flexible, horizontal system that promotes connection and collaboration to bring expertise in a particular technology to a more comprehensive assessment of medical device performance over time.

Postmarket condition of approval studies and Section 522 studies One initiative to address this problem that has been under way for several years is the Total Product Life Cycle (TPLC) program. A pilot aspect of this program formally connects some premarket review processes with the Epidemiology Branch (EB) of the Office of Surveillance and Biometrics (OSB). This collaboration promotes EB involvement in potential postmarket investigations, including initiation and evaluation of product-specific postmarket studies. In FY 2004, there were four PreMarket Approval application (PMAs) that had epidemiologist participation in postapproval study design with the Division of Reproductive, Abdominal and Radiological Devices (DRARD), as part of a pilot project [2]. Epidemiologists were also involved in the analysis of adverse events in previously established postmarket studies, for inclusion in the final follow-up

164

CH11

PERSPECTIVE FROM AN ACADEMIC ON POSTMARKET SURVEILLANCE

reports. Since these examples represent only a small fraction of the total product reviews processed, starting in 2005 the DRARD pilot project was expanded to include epidemiologist involvement in all PMAs which are characterized by a potential need for associated condition of approval (COA) studies. Thus, in the current strategic plan for TPLC, the emphasis is on improved management practices to facilitate the approval process and ‘Risk management throughout the Total Product Life Cycle using the least burdensome means for industry’ [7]. Interestingly, there is little effort to establish or evaluate the actual life cycle of a product. Since implanted medical devices are removed when they fail or at autopsy, one might assume there would be interest in the reason for failure of the device or death of the patient. Such information is not available, neither is there any requirement for a manufacturer to share such information when it becomes available to them. That may fall into the category of ‘burdensome’. One approach to this problem would be to include permission from the patient to obtain the device at the time that consent is granted for its insertion, should the device fail or the patient die. Given the current lack of such critical information, there is good reason to include the requirement for disclosure of device-failure information by the manufacturer at the time of device approval. The inability to predict issues of durability or biocompatibility have long been recognized and can be used to require ‘condition of approval’ (CoA) studies. These postapproval requirements have been specified as ‘continuing evaluation and periodic reporting on the safety, effectiveness, and reliability of the device for its intended use’ [21 CFR 814.82 (a) (2)]. During 1998–2001, there were 127 PMAs approved, with 45 CoA attachments. The serious limitation in this process was the lack of procedures to track the progress or results of these studies. As a consequence, a study conducted in 2003 found that no results were received for 22% of them and two studies had not even been initiated [8]. This source of latent error is being addressed by the transfer of the responsibility for tracking CoAs to the Office of Surveillance and Biometrics (OSB). An automated system was developed to track and confirm the submission of the necessary reports. Further communication of the results of such studies remains a problem, since often not even the Advisory Panel involved in the original approval is made aware of the outcomes. The results are even less likely to become known to healthcare practitioners and patients, since they are not often published in the scientific literature. Utilizing a network of institutional risk managers would be a more effective means of communicating such information. Another postmarket surveillance tool available to CDRH is Section 522 of the Federal Food, Drug and Cosmetics Act, which allows the FDA to require manufacturers to perform specified postmarket surveillance studies for Class II or III products [3]. This is utilized for devices where failure would likely cause serious health consequences, where it is intended to be implanted for more than 1 year, or if it is a life-sustaining device used outside the user facility. Such studies are usually focused on only one or two aspects of performance, rather than the overall long-term risks and benefits associated with the device. Another significant limitation of this tool is the duration of the study, which is limited to 3 years. This time limitation may be too short for some safety or effectiveness

INDUSTRY USE OF INFORMATION FROM ADVERSE EVENT REPORTS

165

problems to be revealed, and is an impediment to the understanding of how devices function in children during growth and development [9]. There also appears to be reluctance to use this mechanism on the part of the FDA, since only two such studies have been requested in recent years. If a safety problem is found, the FDA can require the manufacturer to issue a safety alert or, if the device is felt to be unsafe, can issue a product recall. Again, communication of the results of these studies is problematic, with no readily available format for contacting all institutions and healthcare providers. The need for improved communication is well demonstrated by the experience with cardiovascular devices, where product recalls occur frequently and affect large numbers of patients. During 1990–2004 there were more than 130 product recalls involving more than 900 000 cardiovascular devices [10]. In addition, there were 16 safety alerts affecting almost 900 000 additional patients. Other efforts have been made to improve surveillance techniques for medical devicerelated hazards and actual harm by conducting research that uses computer-based tools [11]. Common device-related events were set up to be screened in the electronic medical record of a single large tertiary care institution. The resulting flagged events were reviewed by research nurses and compared with voluntary incident reports, telemetry problem checklists, ICD-9 discharge-coded events, clinical engineering work logs and patient surveys. The results showed little overlap among the methods, and no real advantage to distanced computer-based surveillance techniques [12]. Therefore, the chronic problem of underreporting of significant events was not improved by the addition of the computer-based technique. The importance of event recovery data to understand how medical devices are adapted by healthcare workers to working conditions requires improvement in voluntary reporting tools, more safety education, and case-based training in device-related issues. This reinforces the need for trained risk managers who can deal with personnel and product turnovers. Additional valuable information can be obtained from patients themselves as they assume more responsibility for decisions and disease management [12]. Their experience with medical devices on an outpatient basis has not often been considered, but can add another dimension to the evaluation and improvement of medical devices.

Industry use of information from adverse event reports In an ideal world, reports of adverse events or near-misses would be used by industry to improve their products. But such changes are expensive, and more likely to result from competition with other manufacturers than from problems uncovered by regulatory review. In fact, one company continued to sell a flawed product even after the problem was corrected and an improved device was available. This situation occurred because the device, an implanted defibrillator, was still considered a ‘highly reliable life-saving product’, according to the company [13]. The correction to improve insulation within the device and prevent the short-circuiting which had caused death in some patients had been

166

CH11

PERSPECTIVE FROM AN ACADEMIC ON POSTMARKET SURVEILLANCE

made in 2002, but not reported to the FDA until 2003. At the end of the following month, the company issued a recall for affected pacemakers and defibrillators, which were then announced as Class I or ‘life-threatening’ by the FDA. This episode not only reflects some of the limitations of the FDA’s current practices, but even more vividly the need for more accountability by industry for postmarket outcomes. FDA staff members usually become involved in device design after multiple adverse event reports, and then they recommend changes and testing procedures to mitigate risk. But it is always a difficult balancing act for the agency, which must preserve a good working relationship with industry to promote cooperation, while retaining its obligation to the interests of the public. Adding to the pressures on FDA staff is the proximity of Congress and the Administration, with predictable political pressures from those with close ties to industry. This can be demoralizing for the professional staff who already make significant sacrifices in salary and their professional careers to work within the confines of the FDA. When FDA staff use their experience with a particular type of device to write a scientific article, it is subject to peer review before publication and then usually published with a disclaimer from the FDA, as expressing the views of the authors and not the official position of the FDA. For example, a recent publication on technical problems with surgical staplers represented an important contribution to understanding the need for better adverse event reporting and the limits of that technology [14]. If investigators choose to report on a particular device made by one manufacturer, they enter the gray zone of potential violation of confidentiality of submitted information. This occurred in 2004, when a collaborative effort between a visiting scholar and OSB staff compared the risks of using an endovascular device to prevent rupture of abdominal aortic aneurysms with the published risk of an open operation. Based on evidence that patients treated with the endovascular device were continuing to die from ruptured aneurysms, the authors suggested that the open operation would likely provide better protection and lower mortality after an estimated period of 3 years. The article was peerreviewed and accepted for publication in the leading vascular surgery journal, which included prepublication of the abstract on the journal’s website. The company quickly filed legal protests with the journal and the FDA, claiming violation of confidentiality. In fact, the information was already public, since a Public Health Notice based on the same data had been published earlier by the FDA (FDA Public Health Notification: Updated 1 Data on Mortality Associated with Medtronic AVE AneuRx Stent Graft System: http:// www.fda.gov/cdrh/safety/aneurx.html). Consequent to this manufacturer’s protest, FDA administrators insisted on withdrawing the article, prompting a rebuke from the editors of the journal [15], a letter to editor by one of the authors [16], and a front-page report in the Wall Street Journal decrying the successful lobbying effort of the company [17]. Additional evidence of suppression of information by the FDA surfaced with the disclosure that same year that presentations and publications of drug-related studies had been altered or inhibited [18]. These examples of administrative censorship to favor industry can have a chilling effect on the research efforts of FDA professionals and damage their academic integrity. They reinforce the need for a more visible system of review and dissemination of FDA analyses and information.

ACADEMIC OPPORTUNITIES

167

Academic opportunities The opportunity to do original research and present the findings at scientific meetings is an important incentive for the professional staff at the FDA. Many important contributions have been made by them and published in the scientific literature. This also serves as a recruiting tool for individuals beginning their careers, along with the opportunity for additional training in the Fellowship program. Most of the recipients of the Fellowship have been engineering students and graduates. There is also a visiting scholar program, which usually attracts more senior physician academicians from outside the FDA [19]. Most of these programs involve collaboration with existing FDA staff, who also gain in the process from exposure to outside fields. Faced with the constant need to replace its professional staff, who are recruited elsewhere or retire, a special graduate education fellowship at the FDA for young physicians and nurses would be an excellent investment. For residents in a variety of medical training programs, the opportunity to become involved in the latest FDA technological areas and to learn more about the regulatory process would be very attractive. This could be offered in blocks of 6 or 12 months, since most training programs offer the option of an outside elective experience. Information about FDA resources online should be a part of the curriculum of healthcare professionals, who should also learn about the importance of adverse event reporting and human factors engineering as a part of patient safety. For members of FDA professional staff, availability of an academic sabbatical would be an excellent investment to offer the opportunity to spend a period of 6–12 months at a center of excellence for a particular emerging technology. These are granted in the academic environment on a cyclical schedule of approximately 7 years. Measurable gains to the FDAwould include not only increased knowledge and experience of staff, but also important insight into the future of the field, new medical device testing approaches, and personal connections with active investigators. Data mining has been studied intensively within the FDA, in an effort to compensate for the complexity of available data and their limitations [20]. As better information becomes available, through such sources as the safety reporting projects of the Agency for Healthcare Research and Quality, related projects of the National Institutes of Health, and databases with both inpatient and outpatient information, such as those from HMOs, Medicaid and Medicare, outcome registries of professional societies, and the database of the Emergency Care Research Institute (ECRI), an effort should be made to integrate the data on device outcomes or, at the very least, to provide a registry of registry-based studies. Further funding is needed for these efforts and for new epidemiologic studies of device safety, including the structured monitoring of device safety. Finally, the advantage of more independent review of safety issues within the FDAwas recognized for drugs, in February 2005, by the proposed establishment of the Drug Safety Oversight Board outside the Center for Drug Evaluation and Research [21,22]. Such an approach is needed for medical devices as well. The Drug Safety Oversight Board is supposed to include other expertise from DHHS and outside the agency, and both consumer and patient representatives. The intent is for it to have more visibility and

168

CH11

PERSPECTIVE FROM AN ACADEMIC ON POSTMARKET SURVEILLANCE

be able to disseminate emerging information more effectively and without interference from special interests. The need for morevisibility for device evaluation has also been recognized in the Senate, where a bill has been introduced (S. 470) to expand the clinical trials drug data bank to include ‘. . . all clinical trials conducted to test the safety or effectiveness (including comparative effectiveness) of any drug, biological product or device. . .’. This ‘Fair Access to Clinical Trials Act of 2005’, or FACT Act, has been referred to the Committee on Health, Education, Labor, and Pensions. In the Medical Device User Fee and Modernization Act of 2002, Congress authorized additional appropriations for postmarket surveillance activities, but never appropriated the funds [9]. Perhaps the public’s increased concern for patient safety will correct this unfortunate situation to allow the FDA to improve its technology and staffing to meet the needs of the twenty-first century.

Summary I have discussed adverse event reporting at the FDA as a medical academician who visited CDRH on a sabbatical. I describe what I saw as the strengths and weaknesses of the FDA adverse event reporting system and database which I used on while at FDA. I see the potential for improving the current system with the goal of improving patient safety. I also see a great potential for sharing between scientific staff at the FDA and academic institutions, as well as opportunities for academics to take a greater interest in medical device epidemiology and surveillance.

References 1. Bren L. Investigators’ report: company caught in cover-up of medical device malfunction. FDA Consumer Mag 2003; 37. 2. CDRH Annual Report, 2004: http://www.fda.gov/cdrh/annual/fy2004/ 3. Gross TP, Kessler LG. Medical device vigilance at FDA. Stud Health Technol Inform 1996; 28: 17–24. 4. FDA Modernization Act. Pub L No. 105-115, 111 Stat 2296, 1997. 5. Kessler DA. Introducing MEDWatch. A new approach to reporting medication and device adverse effects and product problems. J Am Med Assoc 1993; 269: 2765–2768. 6. Manufacturer and user device experience: http://www.fda.gov/cdrh/databases.html 7. CDRH Strategic Plan, 2000–2006: http://www.fda.gov/cdrh/strategic/ 8. OSB Report, 2005: http://www.fda.gov/oc/whitepapers/epi_rep.pdf 9. Committee on Postmarket Surveillance of Pediatric Medical Devices Board on Health Science Policy. Safe Medical Devices for Children, Field MJ, Tilson H (eds). Washington, DC: National Academies Press, 2005. 10. O’Shea JC, Kramer JM, Califf RM, Peterson ED. Sharing a commitment to improve cardiovascular devices. Part I: identifying holes in the safety net. Am Heart J 2004; 147: 977–984. 11. Samore MH, Evans RS, Lassen A, Gould P et al. Surveillance of medical devices-related hazards and adverse events in hospitalized patients. J Am Med Assoc 2004; 291: 325–334.

REFERENCES

169

12. Small SD. Medical devices-associated safety and risk. Surveillance and stratagems. J Am Med Assoc 2004: 291: 367–369. 13. Meier B. Heart device sold despite flaw, data shows. New York Times, June 2, 2005. 14. Brown SL, Woo EK. Surgical stapler-associated fatalities and adverse events reported to the Food and Drug Administration. J Am Col Surg 2004; 199: 374–381. 15. Cronenwett JL, SeegerJM. Withdrawal of article by the FDA after objection from medtronic. J Vasc Surg 2004; 40: 209–210. 16. Greenfield LJ. Regarding ‘withdrawal of article by the FDA after objection from Medtronic’. J Vasc Surg 2004; 40: 594. 17. Mathews AW, Burton TM. After Medtronic lobbying push, the FDA had change of heart. Wall Street J July 9 2004; CCXLIV(6). 18. Okie S. What ails the FDA? N Engl J Med 2005; 352: 1063–1066. 19. Medical Device Fellowship Program: http://www.fda.gov/cdrh/mdfp/ 20. Dumouchel WD. Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system. Am Statistician 1999; 53: 177–202. 21. FDA’s New Drug Safety Initiative: http://www.fda.gov/cder/drugSafety.htm http://www.fda.gov/ cder/drug/DrugSafety/DSOBMembers.htm 22. Drug Safety Oversight Board: http://www.fda.gov/cder/drug/DrugSafety/DSOBMembers.htm

12 Perspective from a pharmacoepidemiologist Thomas K. Hazlet Pharmaceutical Outcomes Research and Policy Program, University of Washington, Seattle, WA, USA

Introduction The academic research appeal of medical device epidemiology appears to have lagged far behind that of pharmacoepidemiology. This lag appears to be related to several fundamental differences between drugs or biologics and medical devices, because of classification or identification schemes employed in the USA that impact ease of exposure discernment and cataloging of academic literature describing medical device epidemiology. This chapter focuses on these differences through a review of existing literature, contrasts with pharmacoepidemiology, and listing of some research challenges and opportunities, and concludes with recommendations for future research.

Review of literature The academic literature was evaluated using the search phrase ‘ ‘‘medical device*’’ AND ‘‘epidemiology’’ ’ , adjusted to the research dataset’s specific requirements – different datasets use truncation symbols other than ‘*’. Research datasets included PubMed [with and without Medical Subject Heading (MeSH) terminology], Current Contents, Web of Science, International Pharmaceutical Abstracts, and Lexis-Nexis Academic Universe

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

172

CH12

PERSPECTIVE FROM A PHARMACOEPIDEMIOLOGIST

Table 12.1 Literature dataset descriptions Dataset title Uniform resource locator (URL) (Free/fee-based)

Description – generally copied from dataset vendor information

PubMed

PubMed is a service of the US National Library of Medicine that includes over 16 million citations from MEDLINE and other life science journals for biomedical articles back to the 1950s.

www.ncbi.nlm.nih.gov/entrez/ Free

Medical Subject Headings (MeSH) is the controlled vocabulary used for indexing articles for MEDLINE/ PubMed. MeSH terminology provides a consistent way to retrieve information that may use different terminology for the same concepts

Current Contents

Current Contents Connect is a multidisciplinary current awareness Web resource providing access to complete bibliographic information from over 8000 of the world’s leading scholarly journals and more than 2000 books

http://portal.isiknowledge.com Fee-based Web of Science http://portal.isiknowledge.com Fee-based International Pharmaceutical Abstracts www.ashp.org/ipa/ Fee-based

Web of Science provides access to current and retrospective multidisciplinary information from approximately 8700 of the most prestigious, high-impact research journals in the world Produced in cooperation with the American Society of Health-System Pharmacists, IPA includes 30 years of in-depth indexed reference to the world pharmacy (in the broadest sense) literature. Also includes related health, medical, cosmetic, and state pharmacy journals, and abstracts of presentations at major pharmacy meetings

Lexis Nexis Academic Universe

This service provides full-text documents from over 5600 news, business, legal, medical, and reference https://web.lexis-nexis.com/universe publications with a variety of flexible search options Fee-based

(Medical Academic Search Forms, Medical Journals). The Uniform Resource Locators (URLs) for the websites and the descriptions of the datasets are provided in Table 12.1. The findings, summarized in Table 12.2, were sparse. In contrast to results for the term ‘pharmacoepidemiology’, where PubMed yielded 582 hits using MeSH terminology and 743 hits as a text search, the, ‘‘medical device*’’ AND ‘ ‘‘epidemiology’’ ’ phrase yielded 47 and 29 hits, respectively. The other research datasets similarly yielded few hits, with a very high proportion of identified articles authored by personnel from the Centers for Devices and Radiological Health, Food and Drug Administration. Of the device epidemiology articles identified above, relatively few met the definition articulated by Torrence: ‘Medical device epidemiology is the study of the prevalence and incidence of use, effectiveness, and adverse events associated with medical devices in a

173

CONTRASTS WITH PHARMACOPEIDEMIOLOGY

Table 12.2 Results of literature searches

Research dataset PubMed using MeSH headings PubMed using phrase Current Contents Web of Science International Pharmaceutical Abstracts Lexis Nexis Academic Universe

Citations retrieved

Research citations (English language) (FDA)

47 29 21 33 9 11

6 19 [14] 3 [3] 11 [6] 2 [1] 3 [3]

Years 1992–2006 1990–2004 1997–2005 1993–2005 1993–2005 1984–2005

population. Identification of data sources is needed for medical device epidemiology [1]’. Many focused on nosocomial infection issues, where the device’s use contributed to the episode. The remainder involved laparoscopic entry access injuries [2,3], atrial defibrillation devices [4], IUDs [5], and silicone breast implants [6–12]. Searching for a specified device produces similar results. Using the MeSH headings ‘ ‘‘Defibrillators, Implantable’’ [MeSH] AND ‘‘Epidemiology’’[MeSH]’ yielded no hits, while the headings ‘ ‘‘Epidemiology’’[MeSH] AND (‘‘Breast Implantation’’[MeSH] OR ‘‘Breast Implants’’MeSH])’ yielded three hits. Note, however, that the MeSH lexicon must be used carefully. Substituting ‘epidemiology’[subheading] for the MeSH term yields 163 hits for breast implants, although many of the hits were less precise. It appears that at the present time, classification systems in use in the major research databases do not accommodate the breadth of the term ‘medical device’. Indeed, there is no MeSH term for ‘medical device’ per se, and the MeSH Browser maps the term to ‘Equipment and Supplies’.

Contrasts with pharmacopeidemiology The 1962 amendments to the Food, Drug and Cosmetic Act first contemplated establishment registration and the Drug Listing Act of 1972 provided FDA with a mechanism to require drug manufacturing establishment registration – the National Drug Code (NDC). Described in Title 21 Code of Federal Regulations §207.35, the regulation established a three-segment number that identified the drug’s manufacturer (or last repackager or ‘labeler’; this number is assigned by the FDA) in the first segment, the firm’s active ingredient(s) and specific strength, dosage form, and formulation for a particular drug in the second segment, and package sizes and configuration in the third segment: labeler code – product code – package code [13]. While NDC listing provided FDA with a tool that facilitated various administrative and drug safety activities, it provided burgeoning health insurers offering a pharmaceutical benefit with an accurate method for pharmacy reimbursement [14], whether from pharmacy or health system records [15] or insurance claims data [16]. As displayed in Figure 12.1, two billing patterns dominate the healthcare industry. For pharmaceuticals dispensed through community pharmacies or mail order, the identity of

174 Pharmaceutical

Health Care Practitioner

Epidemiology Opportunities

Prescriber

Medical Device

CH12

PERSPECTIVE FROM A PHARMACOEPIDEMIOLOGIST

Pharmacy

Data Switch

Pharmacy Benefits Manager

Health Plan or Payer

Patient ID Provider ID Drug ID (NDC) that uniquely identifies drug, i.e., 00591-0424-05 for Watson Lab’s hydrochlorothiazide & triamterene 37.5/25 mg #500 Quantity Charge, etc.

Health Plan or Payer

Data Switch

Patient ID Provider ID Procedure Codes that may identify device type, i.e., C2619 for a pacemaker, dual chamber, non-rate responsive, implantable

Chart review often required

Some electronic capture – i.e., IMS survey Patient identity encrypted Linkage to medical records rare

Patient identity encrypted Linkage to medical records rare

Linkage to medical records rare

Linkage to medical transactions possible, but access to clinical data rare

Data Swith – operates in the electronic insurance billing pathway and accepts or rejects claims, optionally performs edits, and directs the claim to the appropriate clams processor.

Figure 12.1 Typical billing stream for pharmaceuticals vs. medical devices

the drug or biological is included in the billing stream through the embedded NDC, specified in the National Council for Prescription Drug Programs standard. The NDC uniquely identifies the drug or biological received by the pharmacy. What the patient receives may differ. In the example in Figure 12.1, the NDC specifies a container of 500 tablets, a quantity that would not typically be dispensed to a patient for a drug taken once daily. In this instance, the pharmacy would repackage and dispense a 30- or 90-day supply. Although unusual, some medical device manufacturers have obtained NDCs for devices sold through pharmacies to facilitate billing. In contrast, medical products (drugs, biologicals, and medical devices) that are used or administered in a hospital or practitioner’s office are typically bundled into a procedure code, and reimbursed at a preset rate. In consequence, the specific identities of medical products used in the procedure are not discernable from the billing stream. Some very expensive devices, for example, an external assembled lower limb prosthesis, may be billed separately. In some instances, there is only one marketed medical device associated with a procedure code, so exposure to that device might be imputed from billing records [17]. Thus, a side effect of billing procedures for medical products dispensed in pharmacies is exposure data that may be acquired at a relatively low cost for many drug products and some medical products. To the extent that high-cost medical products used or distributed through health devices care practitioners are billed separately, some product-specific exposure information may be available. Some insurers are seeking to unbundle billing

REFERENCES

175

for some costly medical products, for example, chemotherapy agents, to better reflect market prices, and product-specific exposure information may become more available in the future. A recent article by Samore et al. [18] and an associated editorial by Small [19] focused on surveillance systems in a hospital setting, including key word searches in electronic medical records, incident reports, medical engineering logs, review of prolonged telemetry data, and a posthospitalization survey. Even in this practically ideal setting, the authors found that they ‘were not able to derive a definitive measure of the incidence of device-related problems, nor would such an estimate necessarily be generalizable to other institutions’.

Conclusions Several conclusions are evident from an academic perspective of epidemiology and medical devices. The first relates to medical literature classification schemes. As noted, the National Library of Medicine’s present MeSH structure does not seem to easily accommodate medical literature searches for medical device-related issues, and other common literature retrieval databases are even less precise. The MeSH and other classification systems should be modified to better accommodate searches on medical devices and epidemiology. The second relates to discernment issues. In contrast to pharmacoepidemiology, where the enterprise of detecting exposure is facilitated by billing records generated through pharmacies – and hence is available for both institutional settings and community exposure – a parallel structure is not available for medical devices. Bundled billing practices and lack of a unique identifier for medical devices complicates the discernment issue. In consequence, the epidemiology opportunities (bottom row, Figure 12.1) are constrained by the high costs of manual chart review, absence of unique medical product identifiers for most medical devices, and the missed opportunity for exposure information retrieval through data switches [1]. A uniform identification code should be developed and become routinely used in health care records. As medical device epidemiology emerges from its infancy and some of these issues are resolved, major research opportunities are likely to present themselves.

References 1. Torrence ME. Data sources: use in the epidemiologic study of medical devices. Epidemiology 2002; 13(3): S10–S14. 2. Chandler JG, Corson SL, Way LW. Three spectra of laparoscopic entry access injuries. J Am Coll Surgeons 2001; 192(4): 478–490. 3. Bhoyrul S, Vierra MA, Nezhat CR, Krummel TM, Way LW. Trocar injuries in laparoscopic surgery. J Am Coll Surgeons 2001; 192(6): 677–683. 4. Chen L, Keane AT, Every NR. The food and drug administration and atrial defibrillation devices. Am J Managed Care 1999; 5(7): 899–909.

176

CH12

PERSPECTIVE FROM A PHARMACOEPIDEMIOLOGIST

5. Lifeng Zhou MH-WDMC. Use of the New Zealand Intensive Medicines Monitoring Programme to study the levonorgestrel-releasing intrauterine device (Mirena). Pharmacoepidemiol Drug Safety 2003; 12(5): 371–377. 6. Brent J. Silicone breast implants and human rheumatic disease – is there a connection? Int J Toxicol 1998; 17(4): 433–447. 7. Bright RA, Jeng LL, Moore RM. National survey of self-reported breast implants – 1988 estimates. J Long Term Effects Med Implants 1993; 3(1): 81–89. 8. Brown SL. Epidemiology of silicone-gel breast implants. Epidemiology 2002; 13(3): S34–S39. 9. Brown SL, Hefflin B, Woo EK, Parmentier CM. Infections related to breast implants reported to the Food and Drug Administration, 1977–1997. J Long Term Effects Med Implants 2001; 11(1–2): 1–12. 10. Brown SL, Parmentier CM, Woo EK, Vishnuvajjala RL, Headrick ML. Silicone gel breast implant adverse event reports to the Food and Drug Administration, 1984–1995. Publ Health Rep 1998; 113(6): 535–543. 11. Liang MH. Silicone breast implants and systemic rheumatic disease – some smoke but little fire to date. Scand J Rheumatol 1997; 26(6): 409–411. 12. Riley DM, Classen DC, Stevens LE, Burke JP. A large randomized clinical trial of a silverimpregnated urinary catheter – lack of efficacy and staphylococcal superinfection. Am J Med 1995; 98(4): 349–356. 13. Kaplan AH. Fifty years of drug amendments revisited: in easy-to-swallow capsule form. Food Drug Law J 1995; 50(special issue): 179–196. 14. Guo JJF, Diehl MC, Felkey BG, Gibson JT, Barker KN. Comparison and analysis of the national drug code systems among drug information databases. Drug Inf J 1998; 32(3): 769–775. 15. Choo PW, Goldberg JH, Platt R. Ranitidine-associated autoimmune hemolytic anemia in a health maintenance organization population. J Clin Epidemiol 1994; 47(10): 1175–1179. 16. Motheral B, Fairman KA. Effect of a three-tier prescription copay on pharmaceutical and other medical utilization. Med Care 2001; 39(12): 1293–1304. 17. Malenka DJ, Kaplan AV, Sharp SM, Wennberg JE. Postmarketing surveillance of medical devices using Medicare claims. Health Aff 2005; 24(4): 928–937. 18. Samore MH, Evans RS, Lassen A et al. Surveillance of medical device-related hazards and adverse events in hospitalized patients. J Am Med Assoc 2004; 291(3): 325–334. 19. Small SD. Medical device-associated safety and risk: surveillance and stratagems. J Am Med Assoc 2004; 291(3): 367–370.

13 Medical device regulation and surveillance: perspective from the EU Lennart Philipson Medical Products Agency, Uppsala, Sweden

Introduction A complete description of the European regulatory system in the limited space of this chapter is not a trivial, or even a possible, task. The snapshots delivered here may be used as a primer to, and overview of, the system. For a detailed understanding of the mechanisms, please refer to the different publications available from the European Commission; the publications can be found on the website in the reference at the end of this text [1]. The European definition of a ‘medical device’ encompasses a wide variety of products. Although there is no list or register covering all devices currently available on the European market, estimates point to a total of approximately half a million different devices. Before the existence of the European directives on medical devices, the safety and regulatory approach varied considerably among the countries that today are members of the European Union. Today’s medical devices in Europe exist in a market common to all Member States; there are no exclusive national markets. Safety levels, regulatory aspects, and rules for market entry are the same for all Member States.

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

178

CH13

MEDICAL DEVICE REGULATION AND SURVEILLANCE: PERSPECTIVE FROM THE EU

Medical devices: the European directives and definitions From a regulatory perspective there are three groups of medical devices in Europe, which are: active implantable medical devices (AIMDs), general medical devices (MDs) and in vitro diagnostic medical devices (IVDs). For each of these groups there is a European directive that involves multiple issues for each of the different device types. The directives, listed in chronological order, are: AIMD 90/385/EEC [2] MD 93/42/EEC [3] IVD 98/79/EC [4] All three medical device directives belong to a larger group of directives under the New Approach umbrella. The New Approach concept is described in a separate section below. The purpose of EU’s product directives is to ensure safety and performance of the devices and at the same time enable free trade. The medical device directives have been implemented in each Member State’s acts and regulations.

Definitions of medical devices For an understanding of the regulatory issues of medical devices, it is important to know the definitions of the three groups of devices. The following definitions of medical devices are literal quotes from the three directives listed previously. A ‘medical device’ [3] means any instrument, apparatus, appliance, material or other article, whether used alone or in combination, including the software necessary for its proper application, intended by the manufacturer to be used for human beings for the purpose of: Diagnosis, prevention, monitoring, treatment or alleviation of disease; Diagnosis, monitoring, treatment, alleviation or compensation for an injury or handicap; Investigation, replacement or modification of the anatomy or of a physiological process; Control of conception; and which does not achieve its principal intended action in or on the human body by pharmacological, immunological or metabolic means, but which may be assisted in its function by such means. An ‘active implantable medical device’ [2] means any active medical device which is intended to be totally or partially introduced, surgically or medically, into the human body or by medical intervention into a natural orifice, and which is intended to remain after the procedure.

NEW APPROACH

179

The concept ‘active’ introduced above is defined [2] as: An ‘active medical device’ means any medical device relying for its functioning on a source of electrical energy or any source of power other than that directly generated by the human body or gravity; An ‘in vitro diagnostic medical device’ [4] means any medical device which is a reagent, reagent product, calibrator, control material, kit, instrument, apparatus, equipment, or system, whether used alone or in combination, intended by the manufacturer to be used in vitro for the examination of specimens, including blood and tissue donations, derived from the human body, solely or principally for the purpose of providing information: Concerning a physiological or pathological state, or; Concerning a congenital abnormality, or; To determine the safety and compatibility with potential recipients, or; To monitor therapeutic measures.

Note that active implantable medical devices and in vitro diagnostic medical devices are special types of medical devices, hence these devices also fit under the definition of a general medical device in MD 93/42/EEC [3]. Any product that achieves its principal intended action in or on the human body by pharmacological, immunological or metabolic means is not a medical device. Such a product is typically a pharmaceutical product.

New Approach The concept of New Approach is defined and described in the publication Guide to the Implementation of the Directives Based on the New Approach and the Global Approach [5]. Products covered by the New Approach directives are not subject to premarket approval by public authorities. The safety and function of medical devices is always the responsibility of the manufacturer. For low-risk products, the manufacturer may carry out the entire verification of safety and function without any external formal involvement. For products in the higher-risk classes, the safety and function must be verified in cooperation between the manufacturer and a Notified Body (Notified Bodies are described below). Medical devices belong to different classes, depending on the potential risk involved when using the device. Each directive has its own technique of classifying the devices. The Notified Body shall issue a certificate stating that the device fulfills the essential requirements for the applicable risk class. As a signal to the consumer that the product is manufactured in conformity with the requirements of the appropriate directive, the manufacturer attaches a CE marking (see below) to the device before placing it on the market.

180

CH13

MEDICAL DEVICE REGULATION AND SURVEILLANCE: PERSPECTIVE FROM THE EU

When the conformity assessment procedure (conformity with the requirements of the directive) for a device has been carried out in one Member State, the device has access to the entire European market. In EU Commission terminology, the regulatory systems for all Member States are harmonized. The directives allow for some, minor, national adaptations in the Member States implementations of the directives in national law. Most Member States require that labeling and instructions for use (including controls of the device, display text etc.) are written in the language of the Member State. The New Approach is based on the following principles: Harmonization is limited to the essential requirements (defined in the directives and below). Only products fulfilling the essential requirements may be put on the market. The harmonized standards are presumed to conform to corresponding essential requirements. Application of harmonized standards is voluntary. Manufacturers may choose any solution that provides compliance with essential requirements. Manufacturers may choose whatever conformity assessment procedures are provided in the applicable directive.

Essential requirements An understanding of the term ‘essential requirements’ is central to the understanding of the New Approach concept. The entire Annex 1 of all three directives [2–4] is used for a description of essential requirements pertinent to the specific group of products covered by the specific directive. In summary, the essential requirements say that a product shall be safe and suited to its purpose when used according to the manufacturer’s intention. Safety and function shall be documented and possible side effects shall be minimized and described. Instructions for use and a list of known side effects shall be attached on the product or be printed in separate instructions for use.

Notified Bodies A Notified Body is an independent testing and certification organization which has competence to evaluate the safety, quality and performance of products. Most European Member States have one or more Notified Bodies. Notified Bodies are designated by the Member States. Following the designation, the Member State informs (notifies) the European Commission about the name and area of competence of the Notified Body.

NEW APPROACH

181

Information regarding Notified Bodies is published in the Official Journal of the European Communities and on the Commission’s Website. There is currently a total of more than 80 European Notified Bodies dealing with medical devices; this number is growing. Mutual Recognition Agreements (MRAs) exist between the European Union, Australia, Canada, Japan, New Zealand, Switzerland and the USA [6]. The objective of the MRAs is promoting trade of goods between the European Union and third countries by facilitating market access. Based on the MRAs, Notified Bodies (or corresponding assessment bodies) may be located in any of the participating countries and are, in MRA terminology, referred to as Conformity Assessment Bodies (CABs).

Declaration of conformity and the CE marking The CE marking is used for all products regulated under the New Approach directive; ‘medical device’ is only one of many product categories under this umbrella. The CE marking is the manufacturer’s indicator that the product fulfills all the requirements stated in the directive for the specific product. The CE marking is pictured in Figure 13.1. Before affixing the CE marking, the manufacturer must draw up, and keep, a written ‘EC declaration of conformity’ (with the requirements of the applicable directive), a full technical documentation, and a documentation of the risk analysis and verification processes. The amount, and type, of documentation required in the technical construction file depend to some extent on the risks associated with the device. A device associated with a higher risk has to be more extensively documented compared to a device associated with a lower risk. For many high-risk devices, documentation from a clinical evaluation is required. The manufacturer is required to keep this information for at least 5 years after the last production date and to produce the above documents upon request from the authorities.

Figure 13.1 CE marking

182

CH13

MEDICAL DEVICE REGULATION AND SURVEILLANCE: PERSPECTIVE FROM THE EU

Classification of medical devices in Europe For classification purposes, medical devices are sorted in one of a total of nine boxes/ classes. The potential danger of a malfunctioning device is the key parameter for the classification process. To determine the potential danger of a device, characteristics such as the degree of invasiveness, duration of use, and need for sterility are considered. A wheelchair, for example, is classified in a lower risk class than an artificial heart valve. The first step in the classification task is to decide if the device is a general medical device, an active implantable medical device, or an invitro diagnostic device. Definitions of these three categories are found in the respective directives. Active implantable medical devices (Directive AIMD 90/385/EEC [2]) belong to one single class. These are all high-risk devices. An often-used example of an active implantable medical device is the pacemaker. CE marking of active implantable medical devices always requires a Notified Body. Most medical devices belong to the general medical device directive (Directive MD 93/42/EEC [3]). Within this directive, there are four classes: class I, IIa, IIb and III. Class I is the lowest risk class. There are 18 rules to help the manufacturer decide to which of the four classes a product belongs. The device to be classified is fed through the set of rules, starting with number one, and will eventually be found to match one or more of the rules. The rule that assigns the highest risk class to the device is the rule that is valid for the actual device. Below are examples of classification of four different medical devices: Class I: Wheelchairs. Class IIa: Dental fillings. Class IIb: Respirators. Class III: Heart valves. Class I medical devices may be CE marked by the manufacturer without any external involvement. For products in Class IIa and higher, the manufacturer has to engage a Notified Body to a degree that increases with increasing risk class. The terminology for classifying in vitro diagnostic medical devices is different from other medical devices, in that the word ‘class’ is not used. IVD products fall in one of two groups: common IVDs and specified products. The group specified products is furthermore subdivided into three subgroups. Most IVD products (including instrumentation) belong to the group common IVDs, and may be CE marked by the manufacturer itself. All specified products must be CE marked in cooperation with a Notified Body. The subgroups under the specified products umbrella are devices for self-testing and reagents listed in list A and list B in Annex II of Directive IVD 98/79/EC [4]. Reagents are put into List A or

MARKET SURVEILLANCE

183

B (depending on the type of reagent) in meetings to which all Member States are invited. In case of disagreement between a manufacturer and a Competent Authority (CA, see below) regarding a classification, the CA has the preferential right of interpretation.

Market surveillance Member States shall establish or nominate authorities competent to monitor the compliance of products with the requirements of the applicable directives. They also arrange for such authorities to have and use the necessary powers to take the appropriate measures incumbent upon them under the directives. These authorities are referred to as Competent Authorities (CAs). The role of the Member States Competent Authorities within EU is market surveillance. This broad term includes activities such as: Assessment of clinical evaluations. Maintain a register of class I and IVD medical devices. Surveillance of manufacturers’ handling of incidents related to their devices (the vigilance system). Inspections of manufacturers. Inspections of the national Notified Bodies. Exercise the preferential right of interpretation in disputes regarding classification of medical devices. Cooperation with colleagues from other Member States on the EU level on regulatory issues. Standardization work. Distribution of information to manufacturers and users of medical devices. Market surveillance is financed by state funding in the Member States, not by fees from the manufacturers. The overall purpose of market surveillance is to ensure that users of medical devices have access to safe devices that are suited to the intended use. It is a postmarket surveillance system. The Competent Authorities have no product specific role prior to the point in time that the product is placed on the market.

184

CH13

MEDICAL DEVICE REGULATION AND SURVEILLANCE: PERSPECTIVE FROM THE EU

Traceability of medical devices in Europe Manufacturers are required to maintain a record of all the channels used to market a particular product. This is important as it enables the tracking of medical devices if a safety problem is discovered during the use of devices already on the market. Manufacturers are required to promptly present technical files and distribution data to Competent Authorities upon request. The national Competent Authorities also maintain a register of all products in class I and IVD products put on the market in that Member State. AIMD products and medical devices in Classes IIa, IIb and III can be traced by means of the four-digit code adjacent to the CE marking. This code is unique for the Notified Body that was used during the CE marking process. By way of the actual Notified Body, it is possible to find all the technical and sales information pertinent to the specific product. The European Commission has launched a database, EUDAMED, to assist Member States in their market surveillance. One of the functions of this database is a total register of all Class I and IVD medical devices on the European market. This register is kept up to date by all Member States. Under current discussion is the expansion of the EUDAMED register to include also the devices that today are listed with the Notified Bodies only. If a manufacturer of a medical device that is present on the European market does not have a registered place of business inside the European Union, the manufacturer must have an Authorized Representative with legal presence in Europe. The Authorized Representative is considered to be the legal manufacturer of the product and may be addressed by the authorities about all issues regarding the product (in Europe).

The vigilance system Every manufacturer is required to maintain a system enabling them to correct possible problems with their products on the market. When a manufacturer is aware of a problem with their device, he shall notify the relevant Competent Authority and describe the problem and the planned corrective action. The manufacturer shall notify the Competent Authority within 10 days (from the day the manufacturer becomes aware of the problem) for incidents and within 30 days for near-incidents. An ‘incident’ is defined as a malfunction that caused death or serious injury of a patient, user or other person. A ‘near-incident’ is a malfunction that might have caused the above. Manufacturer action can vary from no action at all to a complete recall of the product, depending on the severity of the problem. When the manufacturer has finished correcting the problem, a final report shall be filed with the CA. If the CA agrees with the corrective actions taken, then the file is closed; otherwise the case is reiterated until the CA is satisfied. When the manufacturer’s vigilance system is working as expected, the only task for the CA is to monitor the process. The vigilance system is extensively described in a guidance document from the European Commission entitled MEDDEV 2.12-1 rev 4 [7].

REFERENCES

185

The Safeguard Clause in the directives gives a national CA the power to demand the manufacturer to immediately withdraw the product from the national market of the CA. Actions from one CA shall be communicated to the European Commission and to the Competent Authorities in all other Member States for information. These communications are called Competent Authority Reports (CAR). As an addition to the Competent Authority Reports, every CA in Europe reports vigilance cases of common interest from their country to the EUDAMED database. The database is of help to Member States in following what is happening with a certain product in other countries within the European Union.

References 1. Website of the medical device sector of the European Commission: http://europa.eu.int/comm/ enterprise/medical_devices/index_en.htm 2. Council Directive 90/385/EEC of 20 June 1990 on the approximation of the laws of the Member States relating to active implantable medical devices. Official J L 20.7.1990; 189: 0017–0036. 3. Council Directive 93/42/EEC of 14 June 1993 concerning medical devices. Official J L 12.7.1993; 169: 1. 4. Directive 98/79/EC of the European Parliament and of the Council of 27 October 1998 on in vitro diagnostic medical devices. Official J L 7.12.1998; 331: 1. 5. European Commission. Guide to the Implementation of the Directives Based on the New Approach and the Global Approach. Luxembourg: Office for Official Publications of the European Communities, 2000; 112 pp; ISBN 92-828-7500-8. 6. Website describing the mutual recognition agreements: http://europa.eu.int/comm/enterprise/ international/index_en.htm 7. Website containing guidelines on a medical devices vigilance system: http://europa.eu.int/ comm/enterprise/medical_devices/meddev/2_12-1_04-2001.pdf

14 A consumer advocate’s perspective on medical device epidemiology and surveillance Diana M. Zuckerman, PhD National Research Center for Women & Families, Washington, DC, USA

Consumers show their confidence in the safety of medical devices when they spend the equivalent of more than $200 billion worldwide annually [1]. Nevertheless, consumers are not necessarily familiar with the term ‘medical device’ and might not be able to name any if asked. Virtually every consumer uses medical devices, and many have friends and family members with implanted medical devices. In recent years, the number of men, women, and even children with implanted medical devices has increased dramatically, as artificial knees, hips, heart valves, and shunts have become increasingly common [2,3]. Medical implants come in a very wide range of shapes, sizes, and substances, including the increasingly popular oils and gels that are injected into millions of faces every year to fill wrinkles and scars [4]. The use of implanted devices, either to replace aging body parts or to help people look younger, will certainly continue to increase as the baby boomers age. Lasers are also widely used medical devices, with more than 2 million eye laser surgeries performed in the USA in 2004, more than 1 million laser hair removal procedures, and numerous other laser procedures [5]. Medical devices received relatively little public attention throughout most of the twentieth century, with a few exceptions, such as: the excitement followed by disappointment about lives prolonged with an artificial heart; widespread media attention about infertility and deaths caused the Dalkon Shield in the mid-1970s; serious illness and death from toxic shock syndrome caused by tampons in 1980; and the growing popularity of replacement knees and hips in the last two decades. It was especially difficult to obtain Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

188

CH14

A CONSUMER ADVOCATE’S PERSPECTIVE ON MEDICAL DEVICE EPIDEMIOLOGY

useful safety data about devices prior to 1976, when the Food, Drug and Cosmetic Act was amended to give the US Food and Drug Administration (FDA) substantial authority to regulate medical devices. At that point, there were thousands of medical devices already on the market, most of which were ‘grandfathered’so that they could continue to be sold until the FDA determined whether studies of safety and effectiveness would be required. In the 1990s, very few consumer advocates or nonprofit organizations focused any attention on medical devices. Even so, the few groups that were concerned about medical devices generated considerable public attention regarding the questionable safety of specific devices, such as breast implants, jaw implants, and fetal monitors, especially in the USA, Canada, and the UK. Organizations such as the National Women’s Health Network, Canadian Women’s Health Network, Public Citizen Health Research Group, and the TMJ (temporamandibular joint) Association were vocal critics of specific implants, and they were joined by the National Research Center (NRC) for Women & Families when that research and advocacy organization was founded in 1999. In the last few years, increased attention has been given to the benefits of medical devices for lifesaving procedures as well as common age-defying cosmetic solutions, and the risks have also come under greater scrutiny. In 2001, while working with an informal coalition of approximately a dozen organizations called the Patient & Consumer Coalition, NRC for Women and Families and the National Women’s Health Network brought the issue of improving legislation regarding medical devices to the Coalition for the first time. As a result, broad-based consumer organizations in the USA, such as National Consumers League, Consumer Federation of America, Center for Medical Consumers, Gray Panthers, International Union of UAW, and The Title II Community AIDS National Network, have become knowledgeable about medical devices and started to raise questions about safety data and surveillance. Although consumer groups in or outside the government have been less vocal in other countries, medical devices have attracted the attention and concerns of organizations such as the Canadian Women’s Health Network, Health Canada’s Women and Health Protection, Women’s Implant Information Network New Zealand, and Silicone Support UK. Until 1993, medical device regulation within Europe was the responsibility of the health ministries of each country. Although few countries required clinical trials to determine safety, the regulatory process was nevertheless considered burdensome because different countries posed different requirements for inspecting and authorizing the sale of medical devices [6,7]. These variations made it difficult for European manufacturers to obtain approval to market their products in other countries. When the European Union (EU) established a harmonization program in 1993, however, the program mandated only general requirements for medical devices, and although the criteria include product safety and protection of health, clinical trials are not required to establish either [6]. Manufacturers need only demonstrate compliance in one EU country in order to sell a medical device throughout Europe. Conformity assessment bodies (CABs) are hired by the companies to determine and certify whether a company’s product meets the minimum technical requirements. Few consumers are knowledgeable about the specific safety requirements or approval process for medical devices in their 1

at which time it was called the National Center for Policy Research (CPR) for Women and Families; the name was changed in 2004.

EXAMPLES OF WIDELY PUBLICIZED PROBLEMS WITH SELECTED MEDICAL DEVICES

189

country. From the consumer perspective, the main concern is one’s health: do the benefits of a medical device outweigh the risks? Consumers want medical devices that can save their lives or improve their quality of life, and that means that the device should work, last as long as possible, and not have dangerous side effects. However, for many medical devices sold around the world, there is often limited clinical and epidemiological research to determine the risks and benefits. A brief history of the medical devices receiving the greatest public attention in recent years is instructive in illustrating the concerns expressed by consumers and the organizations that advocate on their behalf.

Examples of widely publicized problems with selected medical devices Dalkon Shield Medical devices were not systematically regulated in the USA and most other countries when the Dalkon Shield intrauterine contraceptive device (IUD) was studied in 1970 (see Figure 14.1). The inventor, Dr Hugh Davis, published a study claiming exceptional effectiveness with no serious risks [8]. The study’s shortcomings were overlooked: only 700 women participated in the study, they were followed for less than 6 months, the researchers did not report that the women used additional contraceptive foam for the first several months, and efficacy statistics were compiled within a week after the study was completed, so that later pregnancies were not reported. After the study was completed and publicized, Davis revised the IUD, adding copper and using a smaller size; the revised product was not tested for safety or efficacy. Approximately 2 million Dalkon Shields were inserted in women in the USA and Puerto Rico [8]. By 1974, pelvic inflammatory disease, ectopic pregnancies, septic abortions, sterility, and 12 deaths had been reported among women who used the Dalkon Shield, and the FDA requested that the manufacturer of the Dalkon Shield, A. H. Robins, remove the IUD from the US market. The company complied, but continued selling it in other countries. It was not until 1980 that the company advised doctors to remove the Dalkon Shield from women who still had them in their bodies, and the IUD was not recalled until 1985. By then, about 9500 cases had been litigated or settled, 6000 more cases were pending, and 16 new cases were being filed each day. Robins filed for Chapter 11 (bankruptcy) protection in 1985, and the settlement included a $2.5 billion trust fund for compensation of more than 100 000 women who sought damages [8]. In response to the Dalkon Shield disaster, and the increased recognition of the risks of medical products, Congress passed the 1976 Medical Device Amendments, which gave the FDA authority to systematically regulate all medical devices [9].

Tampons A few years later, in May 1980, investigators reported to the US Centers for Disease Control and Prevention (CDC; at that time the agency was called the Center for Disease

190

CH14

A CONSUMER ADVOCATE’S PERSPECTIVE ON MEDICAL DEVICE EPIDEMIOLOGY

Figure 14.1 Dalkon Shield poster from the 1970s. Courtesy of the FDA History Office

Control) 55 cases of toxic shock syndrome (TSS), a newly recognized illness characterized by high fever, sunburn-like rash, desquamation, hypotension, and abnormalities in multiple organ systems [10]. Fifty-two (95%) of the reported cases occurred in women, and the onset of illness occurred during menstruation in 38 (95%) of the 40 women from whom menstrual history was obtained. In June 1980, a follow-up report described three studies that found that women with toxic shock syndrome were more likely to have used tampons: case-control studies in Wisconsin and Utah and a national study by CDC.

EXAMPLES OF WIDELY PUBLICIZED PROBLEMS WITH SELECTED MEDICAL DEVICES

191

Subsequent studies established that toxic shock syndrome was more likely among women who used a new, highly absorbent tampon called RelyTM. By September, Rely tampons were voluntarily withdrawn from the market by the manufacturer. During 1980, 890 cases of toxic shock syndrome were reported, 91% of which were associated with menstruation; there were 28 deaths. In response, tampon makers reduced the absorbency of tampons and the FDA began to require that all tampon packages include package inserts explaining the risks of toxic shock syndrome.

Silicone gel breast implants In the late 1970s and early 1980s, following the addition of medical device regulation to the FDA’s responsibilities, the agency was overwhelmed with an enormous number of devices that had previously been on the market and now needed to be classified and possibly evaluated. Breast implants were among the many devices that were allowed to stay on the market until those reviews were completed. Silicone breast implants had been sold since the 1960s and remained on the market while decisions were made about what kind of safety and efficacy studies might be required. Meanwhile, numerous other silicone implants were considered under the law to be ‘substantially equivalent’ to breast implants, and therefore allowed to be sold without any clinical trials to prove safety Scientists and physicians started expressing strong concerns about the safety of silicone breast implants, and by the early 1980s, the suspected risks were officially described in the US government Federal Register [11]. However, it was not until 1988 that the FDA held a public meeting that focused on these risks, and an advisory committee recommended that the FDA establish a national registry of women who have breast implants. The US registry was never established. By 1990, approximately one million women in the USA and Canada, and unknown numbers in other countries, had breast implants, and a scientist at Health Canada had lost his job after publicly urging the agency to remove them from the market. Meanwhile, no government regulatory agencies had yet required the manufacturers to evaluate their safety and no empirical studies had been published regarding their effects on human health. In 1991, pressured by Congressional hearings and enormous news media attention in Canada and the USA regarding non-medical grade polyurethane coverings and reports of implant patients’ illnesses and complications, the FDA finally required the manufacturers to submit safety studies on silicone gel breast implants [11]. The company that made polyurethane-covered breast implants removed their product from the market amid studies indicating that the foam broke down to a known carcinogen, 2,4-toluene diamine (TDA), which was found in the breast milk of women with breast implants. Other implant companies, including Dow Corning, submitted safety data, which FDA scientists reviewed and found to be inadequate [11]. Despite being on the market for almost 30 years, the studies submitted by the breast implant makers were deficient in many respects: they included small sample size and all the women were studied for less than 1 year. However, silicone gel breast implants had been widely available for more than two decades and had become increasingly popular,

192

CH14

A CONSUMER ADVOCATE’S PERSPECTIVE ON MEDICAL DEVICE EPIDEMIOLOGY

providing considerable income for implant manufacturers and plastic surgeons; both groups lobbied heavily to keep them on the market. That pressure, however, was counterbalanced by implant company documents, made public in the course of several lawsuits, indicating that company scientists had expressed concern about the lack of safety data, and the leaking of silicone from intact silicone implants [11,12]. In 1992, as a compromise, silicone gel breast implants were allowed to remain available as a ‘public health need’, with the FDA limiting their availability to clinical trials, primarily for women who have mastectomies, breast deformities, or to replace a broken gel implant. A similar compromise was instituted in Canada. At the same time, other countries considered similar restrictions. In most countries, these restrictions were lifted several years ago and silicone gel breast implants are available to virtually any patients who want them, but the restrictions were not lifted in the USA and Canada until Fall 2006, and even then extensive long-term post-market safety studies were required. In 2001, after years of consumer advocacy pressure by Silicone Support UK, using information published in medical journals and compiled by consumer organizations, the European Commission adopted plans to improve informed consent for women considering breast implants, and urged member countries to establish minimum age limits and to establish registries in all 15 EU Member States [13]. The UK had previously established the first registry of breast implant patients in 1993. Several countries, including Australia and Denmark, have followed suit. All registries are voluntary, which limits the number of patients. Meanwhile, patients won multimillion dollar law suits against implant companies in the early 1990s, so the manufacturers entered into an international legal settlement with patients totaling more than $3 billion dollars, all the while claiming that their implants were safe and not responsible for the health problems that the settlement compensated.

TMJ Implants In 1992, Congressional hearings brought attention to even more obvious health problems caused by jaw implants used to treat temporamandibular joint (TMJ) disorder. Several companies sold TMJ implants made of silicone or other materials, and Dow Corning sold silicone sheeting that could be used for custom-made TMJ implants. Another company, Vitek, made TMJ implants with Teflon and proplast. Most adverse reactions that were reported to the FDAwere for implants made from silicone or Teflon; the friction of the joint caused the jaw implant to flake or break, and the body reacted to the particles with an immune reaction that could cause debilitating pain, bone loss, and in some cases with the Vitek implants, bone degeneration in the joint and skull [14,15]. Like breast implant patients, TMJ implant patients reported systemic autoimmune symptoms and reported that their physicians often assumed that the symptoms were unrelated to their implants. However, patients with TMJ implants reported many symptoms in the jaw joint area, so that at least some of the risks of the implants were identified relatively quickly.

CONSUMER CONCERNS

193

In response to law suits from patients with permanent jaw damage, Vitek declared bankruptcy, but their implants continued to be sold under the names Novamed, Inc., and Oral Surgery Marketing, Inc. When the FDA required them to stop selling their implants, the head of the companies, Dr Charles Homsey, left the USA and sold the TMJ implants in other countries [14]. In 1993, the FDA notified the World Health Organization of its concerns about the proplast implants, and in 1994 the FDA wrote to regulatory agencies in Japan, Italy, Switzerland, Canada, Mexico, Australia, and New Zealand, and the Director General of the European Union, to describe the serious and debilitating complications of proplast implants among TMJ patients in the USA [14]. The examples of the Dalkon Shield, tampons, silicone gel breast implants, and TMJ implants all indicate that there can be substantial risks for medical devices used within the body. The latter three examples also indicate how, even in a country that regulates medical devices, pre-1976 ‘grandfathered’ devices have been allowed to be sold that can have devastating effects on human health. Also, when regulators in one country demand that a product be removed from the market, companies can continue to sell their products in other countries with less stringent regulations for medical devices. Particularly in small countries, where the number of patients using the products is modest, or in products that work well at first but fail over time, the risks of a defective or poorly designed device may not be noticed for many years.

Consumer concerns Consumer concerns about device manufacturers and their research As these examples illustrate, consumers or their physicians were the first to complain about the adverse reactions to these medical devices, and in most cases the manufacturers defended their products and challenged consumers in court. In some examples, such as tampons, healthcare professionals were instrumental in bringing attention to the problem; in others, physicians tended to assume that the medical devices were safe and unrelated to the problems being reported. In the case of breast implants, it was only when it became clear that there were no safety data to back up company claims, and internal corporate documents indicated the possibility of a cover-up, that the products were withdrawn from the market or restricted, usually with belated pressure from the regulatory agencies of countries such as the USA and Canada. In recent years, the European Commission has applied pressure on EU countries to institute safeguards that implant manufacturers and plastic surgeons were not providing, such as informed consent that provides information about specific risks. All these examples have a fundamental scientific problem in common: the lack of meaningful short-term or long-term safety research. It was only when unexpected adverse reactions were reported – by the CDC, as part of law suits, or by physicians – that there was pressure on government regulatory agencies to require that research be conducted. In the case of breast implants, this was initiated by a Health Canada engineer who served as a whistleblower, generating media attention in Canada that spread to the USA [11].

194

CH14

A CONSUMER ADVOCATE’S PERSPECTIVE ON MEDICAL DEVICE EPIDEMIOLOGY

Manufacturers defend the lack of research for medical devices, stating that, unlike pharmaceutical companies, devices tend to be modified frequently in response to the requests and recommendations of physicians. For example, AdvaMed, the largest medical technology trade association in the world, claims that: ‘Medical device innovation development differs significantly from pharmaceutical innovation in that most devices on the market today result from a series of incremental improvements to preexisting devices. These improvements result from continued vigilance by the manufacturer and substantial input from the provider community. Although well-designed research plays a significant role, formal research projects cannot substitute for the one-to-one interaction between the researchers tasked with developing and improving a technology and the clinical personnel who use it in their therapeutic and diagnostic interactions with patients’ [16].

AdvaMed represents 800 companies selling more than half of the healthcare technology products purchased worldwide [16]. Based on their view, a study of an implant made 10 years ago or even 2 years ago might be irrelevant to the product being sold today. However, in some cases described in this chapter, there was evidence that research indicating risks was not published or made public.

Consumer concerns about regulatory safeguards for medical devices Concerns about insufficient regulatory safeguards for medical devices reflect the differences between these devices and prescription drugs. Historically, most devices were used outside the body (such as scalpels and band-aids), and there was a perception that ‘what you see is what you get’, making research seem less important. As implanted medical devices have become more common, long-term research has become more important, but the safeguards and resources for regulatory agencies, in the USA and other countries, has not kept up with the increased importance of those devices. In most countries, medical devices are routinely approved for marketing on the basis of short-term studies. This is also true in the USA, although manufacturers of high-risk devices are often required to do longer-term postmarket studies as a condition of device approval. Postmarket studies that are required because of concerns that arise after product approval, rather than as a condition of approval, are limited by FDA regulations. For example, 3 years is the maximum time that the FDA can impose for postmarket research requirements on medical devices ordered after approval without the agreement of the manufacturer; that is not sufficient to examine long-term safety [17]. Moreover, recent reports by the Institute of Medicine and the FDA indicate that postmarket studies, imposed as a condition of approval, have been inadequately monitored, and that the studies were often not performed or finished [18,19]. Add to that the corporate rationale that devices are constantly being improved and therefore regulatory flexibility is necessary, and there is a clear conflict between consumer demand that products be proven safe and corporate demands that products be approved quickly and be allowed to change without the need for new approval applications. These issues are raised in countries all over the world, and there is not one country that has insisted on or

CONSUMER CONCERNS

195

consistently enforced long-term postmarket surveillance of medical devices, not even of those implanted for very long-term use. FDA regulations differ in the safety criteria for medical devices compared to new drug approvals, and these differences are similar in other countries as well. Drugs must be safe for the uses recommended in labeling, which is interpreted as meaning that the benefits outweigh the risks. In contrast, a medical device must have a ‘reasonable assurance of safety’, which is more ambiguous; the law requires that the ‘probable benefits to health’ should outweigh ‘any probable risks’ (21CFR860.7). This has been interpreted as a less stringent criterion for safety and effectiveness, where scientific proof that the benefits outweigh the risks is not necessarily required. In racially and ethnically diverse countries such as the USA, the potential for racial and ethnic differences in responses to implanted medical devices has become an issue of concern among consumer groups. The NRC for Women & Families, the National Medical Association, and the Congressional Black Caucus of the US Congress have all expressed their concern that implant makers rarely include racial and ethnic minorities in their studies. Since individuals of African or Asian ancestry are more likely to have keloid scarring, and since individuals of African ancestry are more susceptible to autoimmune diseases, medical implants may be more risky for those groups. However, it is impossible to know whether this is the case if no studies have been done. Consumer groups have the opportunity to influence regulatory decisions in countries using independent advisory panels, such as is the case in the USA and Canada. Consumers are represented on the advisory panels and also have the opportunity to speak during the open public comment periods. However, whatever the roles consumers play, there is reason to be concerned that advisory panels tend to be a rubber stamp for approval. In a study released in 2006, NRC for Women & Families compared recommendations from FDA advisory panels for medical devices with advisory panels for prescription drugs. Votes between 1998 and 2005 were compared for five randomly selected device advisory panels and six randomly selected drug advisory panels. During those 8 years, the advisory panels recommended approval for 82% of medical devices that they reviewed, compared to 76% of prescription drugs under review. Some panel members always voted for approval for any product during their entire tenure on the advisory panel. NRC for Women & Families concluded that the less stringent criteria for approval for medical devices created an expectation that most medical devices were ‘reasonably safe’ and therefore suitable for FDA approval. Although panel members often expressed concern about the lack of safety information, they apparently assuaged those concerns by recommending postmarket studies and other conditions of approval. Unfortunately, as discussed later in this chapter, postmarket studies and surveillance are often not enforced [18].

Consumer concerns about long-term safety of implants Of all the concerns that consumers have about medical devices, the long-term safety of implanted devices has attracted the most attention. There is widespread agreement

196

CH14

A CONSUMER ADVOCATE’S PERSPECTIVE ON MEDICAL DEVICE EPIDEMIOLOGY

among consumer advocates that in most countries the current statutes regulating medical devices are inadequate for ensuring adequate safety studies, especially for life-saving and implanted devices. In the USA, the Patient and Consumer Coalition has participated in meetings with individual FDA officials, FDA forums, meetings with Members of Congress and their staff, and Congressional briefings to urge policy makers to require better research, including long-term safety studies, and better postmarket surveillance to improve the safeguards for implanted medical devices. These concerns are similar to those expressed by consumer advocates in Canada, the UK, and other countries. The lack of long-term safety studies is a particular problem for implanted devices. Medical devices are allowed to be sold without proof of long-term safety. Solutions that have been suggested by consumer groups include the following: Government regulatory agencies must devote more resources to postmarket surveillance that focuses on long-term efficacy, reliability, and safety. Government regulatory agencies should be required to closely monitor, document, and audit all medical device Phase IV trials. The studies should be monitored for the adequacy of informed consent and human subject protection, the quality of study design, and the accuracy of results. Registries should be used more often to keep track of adverse reactions to devices and as a mechanism to inform patients of recalls or other problems. Adverse event reporting must be improved for medical devices, especially implanted devices. All hospitals, Health Maintenance Organizations (HMOs), nursing homes, and other healthcare providers should be required to immediately submit all adverse event reports to government regulatory agencies, and this should be stringently enforced. Information technology must be employed to facilitate the submission of adverse event reports. Government regulatory agencies or health agencies should be required to write and distribute consumer guides that provide unbiased, clearly-worded research-based information about the risks and benefits of medical devices used by consumers, such as implanted medical devices. This need is exemplified by the fact that, although US health experts have focused increasing efforts to provide understandable materials for consumers, there is little effort to develop consumer-oriented written materials for medical devices, since devices are often ‘used’ by medical professionals (sometimes by surgically implanting them in patients) rather than by patients. In the rare instances when postmarket studies or surveillance are required (which is more likely in the USA than in other countries) consumer advocates are concerned that such studies and surveillance are not monitored adequately to ensure that they are conducted appropriately, or to ensure that companies or physicians provide information relevant to adverse reaction reports. Whether this is due to inadequate resources or inadequate

REGULATORY MECHANISM RECOMMENDATIONS

197

regulatory authority, it has become increasingly obvious in recent years that patients can not be assured they have the information they need to avoid medical disasters resulting from insufficiently safe medical devices, particularly those implanted in their bodies. For example, a study by the FDA of all 127 premarket agreements (PMAs) approved during 1998–2000 found that 45 required postapproval studies. Although the law requires manufacturers to include information about these studies in their annual reports, only 19 of the 45 legally required studies (42%) were mentioned in annual reports. For the 11 PMAs where the results were due, final results had not been submitted in six (54%) cases [19]. This would make it impossible for consumers to obtain the information they need about the long-term safety of these devices. In 2006, the FDA announced initiatives to improve the enforcement of postmarket requirements; future reports will evaluate the results of those FDA initiatives [20]. Given the failure of device companies to submit required data, government regulatory agencies would benefit from subpoena power to compel manufacturers and healthcare providers to deliver documents relevant to all mandated regulatory functions regarding medical devices. Since premarket studies for medical devices are often small and of short duration, these postmarket studies take on even greater significance. That manufacturers agree to these studies as a condition of approval for their medical device and then do not finish them should be adequate reason for withdrawal of the product from the market, until such a time as adequate studies are completed.

Regulatory mechanism recommendations Recalls Several well-publicized recalls have brought attention to shortcomings in removing defective products from the market. For example, in 2002, a defect in a bronchoscope manufactured by Olympus led to persistent bacterial contamination of the instruments (see Chapter 17, ‘Medical device-related outbreaks’). A recall was delayed for two months, and problems continued even after the recall. As reported in newspapers across the country, the recall notice to Johns Hopkins Hospital was sent by Olympus to a loading dock instead of the department using the bronchoscope. As a result, Johns Hopkins continued using the defective instruments for several months after the recall was initiated [21]. Apparently, other medical centers also were unaware of the recall, which was not widely publicized and which we found was posted on neither the company’s nor the FDA’s websites. As a result of this and other examples, consumers and their advocates have become increasing concerned about medical device recalls. In the USA, an article in a mainstream women’s magazine, Good Housekeeping, in March 2004, explained that neither device companies nor physicians are required to send patients a notification that a medical device implanted in their body has been recalled. The magazine encouraged readers to respond, resulting in more than 10 000 consumers joining a campaign to

198

CH14

A CONSUMER ADVOCATE’S PERSPECTIVE ON MEDICAL DEVICE EPIDEMIOLOGY

change these policies – the largest response the magazine has ever received [22]. Similarly, consumer organizations agree that government regulatory agencies need to have a more active role in the oversight of medical device recalls, for example: Government regulatory agencies need a clear and explicit legal mandate to assume primary responsibility for the supervision, monitoring, and enforcement of all medical device recalls, rather than requesting that the companies provide recall information to the public. The agencies should be required to quickly and efficiently disseminate accurate and pertinent information regarding the recall of medical devices to patients and healthcare providers. An office or agency independent of the health regulatory agency is needed to investigate the circumstances surrounding the withdrawal of any approved medical device from the market.

Legislative and regulatory changes requested As a result of the EU’s streamlined process, which rarely requires clinical trials to examine the safety of medical devices, the USA came under pressure to ease their approval process for medical devices. In 2002, the US Congress passed the Medical Device User Fee and Modernization Act (MDUFMA), and this law was amended in 2004 [23]. Consumer groups opposed many aspects of these bills, which weakened rather than strengthened FDA’s regulatory muscle. In their opposition, consumer groups pointed out three general concerns: The law favors rapid medical device approval over medical device safety. The law ignores the need for improved postmarket surveillance. The law sets time limits on reviews of medical device safety, which could divert resources from other important FDA functions. Consumer groups were especially concerned that the bill supports privatization of several essential regulatory functions of the FDA, by allowing for third-party reviews and inspections. The bill extended a previous law implementing 510(k) review by third parties of most Class II devices. In addition, the law initiated/expanded the use of nonFDA ‘accredited persons’ to conduct inspections of medical device facilities, including Class II and Class III devices that are permanently implantable, life-sustaining, or lifesupporting. The third parties must be selected from a list of accredited persons compiled by the FDA; however, the specific accredited third party can be chosen by medical device manufacturers. Compensation for accredited persons is determined by medical device manufacturers in agreement with the third parties, and is paid by the manufacturer. Consumer groups point out that this arrangement creates a clear financial conflict of

CONSUMER GROUP ACCOMPLISHMENTS: MIXED RESULTS

199

interest; if a company wants to be hired for these tasks, it is in their financial interest to please their customers. Consumer concerns about the need for legislative and regulatory changes have increased during the early years of the twenty-first century, in response to several well-publicized failures of medical devices. Although consumer groups have been concerned about how recalls are handled, recalls also bring attention to issues with the approval process, not just the recall process. For example, recent recalls of heart valves and defibrillators have brought public attention to life-threatening problems that can result from defective medical devices [24–28]. Thus far, however, consumer groups have been unsuccessful in their efforts to strengthen the regulation of medical devices. On the contrary, medical device companies can get new products approved in an expedited process that does not necessarily require clinical trials if the new product is considered ‘substantially equivalent’ to another product on the market. The definition of ‘substantial equivalence’ is very vague, and has included products made of different materials and/or for very different intended uses – differences that could substantially affect safety and effectiveness.

Consumer group accomplishments: mixed results As a reflection of the growing clout of consumer organizations, in September 2005, Health Canada held its first-ever public meeting of an advisory panel, for the review of a controversial medical device: silicone gel breast implants. The public meeting was in response to consumer complaints about a secret meeting that took place in March 2005, with an ‘independent’ advisory panel. The controversy arose when it became known that the ‘expert advisors’ who participated in the panel meeting included two men who were paid consultants to one of the breast implant manufacturers, Inamed, whose products were being reviewed by Health Canada. In fact, both ‘expert advisors’ had testified on behalf of the safety of Inamed silicone gel breast implants at an FDA advisory panel in April of the same year. When consumer groups pointed out that paid consultants were unlikely to make unbiased judgments about the product, Members of Parliament joined them in demanding a more open, balanced process. As a result, Health Canada officials held a public advisory panel meeting, modeled after the FDA public meetings. However, consumer groups were shocked to learn that the same industry-paid consultants who were on the panel for the secret meeting would remain on the panel for the public meeting, as well as other consultants to one or both implant manufacturers. Only one consumer advocate was on the panel, and patients and advocates were given only 3 minutes each to testify during the public comment period. The expense and inconvenience of traveling to Ottawa, and concerns that the panel vote for approval was preordained, apparently outweighed patients’desire to publicly testify for 3 minutes, and few consumer representatives or patients testified. Recent consumer efforts in the USA indicate similarly mixed results. In response to consumer and Congressional pressure about inadequate postmarket surveillance of medical devices, the FDA announced its intention to address shortcomings in 2006,

200

CH14

A CONSUMER ADVOCATE’S PERSPECTIVE ON MEDICAL DEVICE EPIDEMIOLOGY

and held the first in a series of ‘workshops’ on this topic in early February 2006. The first workshop was described afterwards in the FDA’s press release as a meeting ‘between the FDA and AdvaMed’, the organization that represents device manufacturers. Consumer groups were not included in the planning or agenda of the workshop, and were not notified that it was held until after the meeting was over. Moreover, the meeting was organized by AdvaMed rather than the FDA, and participants were charged several hundred dollars each to attend, which likely reduced the participation of government employees and representatives of nonprofit organizations. Presumably consumer groups will be invited to a later workshop organized by the FDA, but the question arises as to why they were not included as an integral part of all FDA meetings on the topic. These examples indicate that consumer organizations are actively pushing for better regulatory processes and safeguards for medical devices, but are meeting with limited success to even ‘be at the table’ and have their voices heard. As device manufacturers savor their victories in streamlining the approval process, consumer advocates complain that they are tilting at windmills in the face of regulatory agencies that seem oblivious to conflicts of interest, unconcerned about long-term safety data, and indifferent to the shortcomings of postmarket surveillance. Nevertheless, consumer advocates continue to make their voices heard, and partly as a result of those efforts the news media are focusing more attention on dangerous defects and numerous recalls of specific medical devices. In the USA, Canada, and the UK, legislators have joined consumer advocates in their demands for greater safeguards, and the combined forces of consumer groups, Members of Congress, Members of Parliament, and the attention of the news media may eventually influence government regulatory agencies and device manufacturers, resulting in improved safety studies and postmarket surveillance.

References 1. Medical device regulations: global overview and guiding principles. World Health Organization, 2003: http://www.who.int/medical_devices/publications/en/MD_Regulations.pdf 2. Kojic EM, Darouiche RO. Candida infections of medical devices. Clin Microbiol Rev 2004; 17: 255–267. 3. Garver D, Kaczmarek RG, Silverman BG, Gross TP, Hamilton PM. The epidemiology of prosthetic heart valves in the United States. Texas Heart Inst J 1995; 22: 86–91. 4. American Society of Plastic Surgeons. Procedural statistics: http://www.plasticsurgery.org/ public_education/2004Statistics.cfm [accessed October 10 2005]. 5. American Society for Aesthetic Plastic Surgery. Beam me up Scotty: plastic surgeons find better results and fewer complications with new laser treatments. October 6 2005 press release: http:// surgery.org/public/news-release.php?iid¼414andsection¼ 6. The US–EU Recognition Agreement: its implications for the US medical device industry: http:// www.devicelink.com/mddi/archive/01/05/003.html 7. Medical device reporting. Improvements needed for FDA’s system for monitoring problems with approved devices. GAO/HEHS-96-65, 1997: http://www.gao.gov/archive/1997/he97021.pdf [accessed October 13 2005]. 8. Sobol RB. Bending the Law: The Story of the Dalkon Shield Bankruptcy. Chicago, IL: University of Chicago Press, 1991; 1–6.

REFERENCES

201

9. FDA milestones in women’s health: http://www.fda.gov/womens/milesbro.html [accessed February 15 2006]. 10. Historical perspectives reduced incidence of menstrual toxic-shock syndrome – United States, 1980–1990: http://www.cdc.gov/mmwr/preview/mmwrhtml/00001651.htm [accessed October 10 2005]. 11. The FDA’s Regulation of Silicone Breast Implants. Staff report prepared by the Human Resources and Intergovernmental Relations Subcommittee of the House Government Operations Committee, December 1992. 12. Gladwell M. Documents tell risks of implants. Washington Post February 11 1992; 1. 13. BBC News. Victory for breast implant victims, November 15 2001: http://news.bbc.co.uk/1/hi/ health/1657856.stm 14. TMJ implants. A consumer information update, April 2001: http://www.fda.gov/cdrh/consumer/ tmjupdate.pdf 15. Congressional Hearing. Are FDA and NIH ignoring the dangers of TMJ (jaw) implants? Hearing before the Human Resources and Intergovernmental Relations Subcommittee of the Committee on Government Operations, House of Representatives, June 4 1992. 16. AdvaMed: testimony to the House Energy and Commerce Health Subcommittee on ‘Assessing HIPAA: how federal medical record privacy regulations can be improved’, March 22 2001, available at: http://www.advaMed.org/publicdocs/hipaatestimony3-22-01.html 17. Postmarket Surveillance Studies: www.fda.gov/cdrh/devadvice/352.html [accessed October 14 2005]. 18. Zuckerman, DM. FDA Advisory Committees: Does Approval Mean Safety? National Research Center for Women & Families, 2006, available at http://www.center4research.org/pdf/FDA_ Report_9-2006.pdf. 19. Brown S, Bezabeh S, Duggirala H. Center for Devices and Radiological Health condition of approval studies as a postmarket tool for PMA approved cohort 1998-2000. Rockville, MD: Food and Drug Administration: http://www.fda.gov/oc/whitepapers/epi_rep.pdf [accessed October 10 2005]. 20. Field MJ, Tilson H (eds). Committee on Postmarket Surveillance of Pediatric Medical Devices, Institute of Medicine. Safe Medical Devices for Children, 2005: http://www.iom.edu/ report.asp?id¼28277 [accessed October 14 2005]. 21. Medical Device Postmarket Transformation Initative: http://www.fda.gov/cdrh/postmarket/ mdpi.html [accessed February 15 2006]. 22. Patterson P. Bronchoscope recall falls through cracks. OR Manager 2002; 18: 20. 23. Good Housekeeping. Dangerous devices. Article and campaign described on: http://magazines. ivillage.com/goodhousekeeping/hb/health/articles/0,,284594_651884-14,00.html [the characterization as the magazine’s largest response was made by editor Toni Hope in a personal communication]. 24. Medical Device User Fee and Modernization Act (MDUFMA) of 2002: http://www.fda.gov/ cdrh/mdufma/ [accessed October 10 2005]. 25. Medtronic announces a nationwide, voluntarily recall of small subset of two implantable cardioverter-defibrillator models: http://www.fda.gov/oc/po/firmrecalls/medtronic04_04.html 26. Medtronic announces additional devices affected in voluntary recall of certain monophasic LifePak1 500 automated external defibrillators: http://www.fda.gov/oc/po/firmrecalls/lifepak04_25.html 27. FDA updates consumers on Guidant Corporation’s implantable defibrillators: http://www. fda.gov/bbs/topics/NEWS/2005/NEW01198.html 28. FDA issues nationwide notification of recall of certain Guidant implantable defibrillators and cardiac resynchronization therapy defibrillators: http://www.fda.gov/bbs/topics/NEWS/2005/ NEW01185.html

15 Pediatric Medical Device Use Judith U. Cope and Thomas P. Gross US Food and Drug Administration, Rockville, MD, USA

Introduction Few clinical studies focus on the use of medical devices in children; most are conducted in an exclusively adult population. As a result, there is often inadequate information to assess use of particular devices in the pediatric population. In addition, as with other regulated products, ‘off-label’ use (i.e. outside of approved indications) is common in both adult and pediatric patients. Part of what is driving this use is unmet clinical needs, arising primarily from the relative lack of availability of pediatric-specific devices. The reasons for this state of affairs and the response to unmet needs are addressed. In addition, to better understand pediatric device use, examples of some common pediatric conditions and special devices issues are discussed. Furthermore, unique considerations with regard to device safety apply to the pediatric population; illustrative cases are presented. As is true for adults, medical devices used in the pediatric population can be categorized in several different ways. A regulatory framework for their categorization, used by the FDA, is outlined. Historically, legislation has shaped this framework, initially with the passage of the 1938 Federal Food, Drug, and Cosmetic Act, but much more substantially with the 1976 Medical Device Amendments to that Act. The Medical Device User Fee and Modernization Act (MDUFMA) of 2002 is no exception. Several provisions of MDUFMA, as they apply to the pediatric population, are addressed. Lastly, thoughts on the future of pediatric device use, surveillance, and epidemiology are presented and discussed. Combined efforts in this arena will augur well for pediatric device development and device safety.

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

204

CH15

PEDIATRIC MEDICAL DEVICE USE

Medical device use in children Pediatric diseases and conditions that lead to device use, when compared to those in adulthood, differ by their etiology, consequences, and rates of occurrence. Heart failure during childhood usually results from congenital heart defects, not myocardial infarction, and the rapid onset of life-threatening infections, asthma, and accidents are common causes for pediatric hospitalization, with high rates of morbidity and mortality. It is therefore important to focus on the intended age group in which a device will be used, whether it is intended to diagnose, manage a temporary disease state, treat an injury, or cosmetically correct a defect or other condition. Also, within the pediatric population, the incidence of medical conditions varies with the age of the child, and while certain devices may be designed or modified for use in a specific targeted age group, they must often be safe, effective and adaptable for use in children of other ages as well. Key age-dependent factors of body size, anatomy, physiology, immunity, and hormonal status of the child should be understood, as well as cognitive and emotional growth. Recognizing the importance of age, the FDA recently specified pediatric age categories of neonate, infancy, childhood, and adolescence to be considered for the use of medical devices [1]. As briefly noted above, few clinical studies have been conducted to study the risk:benefit of various devices used in children and insufficient data has been collected to support pediatric indications for devices that may already be used in adults. Reasons for this include: lack of market awareness by device manufacturers and researchers (i.e. need to identify, characterize, and prioritize pediatric device needs); lack of market stimulus (i.e. the relatively small market share and lack of broad patent protections); statutory and regulatory hurdles (i.e. current mechanisms may be perceived as overly burdensome for small device firms, which account for the majority of firms); technical challenges (e.g. in selection of biomaterials or in device design for long-term use, or in conducting animal and clinical studies); insufficient numbers of children who volunteer to participate in trials; the extended time of study to evaluate longer-term safety endpoints (e.g. growth and neurodevelopment); complex ethical factors; and the not uncommon practice of adapting a device for immediate use without concern for conducting burdensome studies. Currently, a systematic needs assessment is under way to determine the scope of unmet device needs in the pediatric population. This needs assessment is a collaborative effort between FDA and other major stakeholders (e.g. the American Academy of Pediatrics). In the interim, while striving to avoid suboptimal treatments and/or serious complications, the healthcare community needs to appreciate the unique characteristics of children in both health and disease and consider not only the short-term risks, but those that may last a lifetime. To better understand some of the common pediatric conditions and special device issues for these various age groups, the following device examples will be discussed: (a) thermometers to detect fever in the neonate and young infant; (b) pressure-equalizing (PE) tubes to treat fluid build-up in the middle ear in older infants and toddlers; (c) orthopedic devices to stabilize and allow healing of fractures in older children; and

MEDICAL DEVICE USE IN CHILDREN

205

(d) cosmetic device interventions to further correct deformities and defects or enhance appearance in adolescents.

The thermometer A seemingly simple device, the thermometer is critically important for determining whether or not a young infant has fever. During the first months of life, infants have underdeveloped immune responses that make them particularly susceptible to infections. An elevated temperature may be the only sign indicating that a young infant is critically ill. The incidence of bacteremia among infants with fever who present to the emergency room during the first few months of life varies somewhere in the range of 3–15% [2–5]. Sepsis is one of the leading causes of death during early infancy, with greater than 40 000 cases reported annually in the US [6] and the highest rates of severe sepsis occurring in the neonatal period. Detection of fever in this young infant population may warrant a complete sepsis work-up, with diagnostic blood tests, spinal tap and a urinary catheterization or bladder needle aspiration for urinalysis and culture. Thus, if a thermometer incorrectly indicates an elevated temperature (a false positive reading), then invasive procedures may be performed, with additional costs of antibiotic therapy and unnecessary hospitalization. Alternatively, if it erroneously registers no fever (a false negative reading), especially in the absence of other signs or symptoms of sepsis, then the young infant may not receive necessary treatment and face serious, life-threatening disease. The thermometer must be accurate, reliable, and suitable for use in various different clinical settings, such as the hospital, doctor’s office, and home. The rectal thermometer has been generally accepted as the ‘gold standard’ reference against which other temperature-taking devices are tested [7], and in clinical practice it is the thermometer of choice during the first 6 months of life [8]. A variety of other thermometer instruments have been developed, including: pacifier thermometers (oral pacifiers with built in thermometer); plastic strip thermometers that indicate skin temperature by changing color (liquid crystal thermometers); temporal artery thermometers (handheld infrared heat sensor devices to scan the temple skin surface); and infrared tympanic thermometers (to measure the radiant heat from the ear drum surface). Although the pacifier, plastic strip, and temporal artery thermometers are easy to use and acceptable to parents, they have low specificity and have been found to be less reliable [9–12]. The infrared ear thermometer has shown more promise; however, there continues to be concern about its accuracy and other problems with this device [13–16]. Importantly, Yaron et al. [17] emphasize that while comparison studies may report that various temperature device instruments may correlate with each other, this does not mean they correctly agree on the final temperature readings. Temperature-taking devices for use in children should be easy to use and safe. Rapid readings and noninvasive instruments are important when used on young children who may be restless and uncooperative, especially when they are ill. The old standby mercury-in-glass thermometers had many problems, including: parents had difficulty

206

CH15

PEDIATRIC MEDICAL DEVICE USE

reading the thermometer scale; disinfection was required after each use; 5–6 minutes of insertion time was needed; and glass thermometers were found to be uncomfortable for children. Currently, glass thermometers are being phased out because of concerns regarding the hazards of mercury exposure [18]. Although not perfect [19], digital thermometers have addressed many of the earlier concerns of mercury thermometers. And while rectal thermometers may pose some problems of longer placement time [20] and longer lag time in temperature change [21], they continue to be the standard tools used to evaluate young infants for fever. As new thermometer technologies continue to be developed, considerations should be given to which age groups will use the device and their clinical applications. Future studies should be conducted to determine the reliability of the new thermometers and avoid many of the drawbacks and limitations of previous studies by ensuring that temperature measurements are carried out independently, that each child study participant has temperature readings by both tests, that there is no time delay between the comparison thermometer measurements, and that data should be recorded regarding thermometer calibration [20].

Pressure-equalizing tubes in infancy After the first few months of life and throughout early childhood, one of the most common pediatric illnesses is ear infection (otitis media). While otitis media usually resolves with or without antibiotics, many children have recurrent episodes and fluid accumulation in the middle ear. The fluid may persist without signs or symptoms of acute infection or inflammation (otitis media with effusion, OME) [22]. The peak age incidence of OME is during the first 2 years of life [22,23] and more than 90% of infants aged 2 months to 2 years will have one or more episodes of OME [24]. Prospective studies of 2 to 4-year-old children reveal that 50% or more of OMEs resolve within 3 months [25,26] and 95% resolve within 1 year [26]. However, there continues to be concern about significant hearing problems associated with OME and delay in speech and language acquisition. Ongoing controversy exists as to the best management of these young children with recurrent otitis and OME. Surgical treatment consists of myringotomy with insertion of a small tube through the tympanic membrane. It is the most common surgical procedure among infants and young children [27]. During the first 3 years of life the overall prevalence is 21 per 1000 children in the USA [28]. The primary goal of PE tube insertion is to remove middle ear effusion, prevent fluid accumulation and thus restore hearing, reduce recurrence of infections and prevent developmental delays in speech, language and cognition. The complications and sequelae of tympanostomy tube insertion have been examined by various studies, including one meta-analysis [22,29,30]. They include ear drum perforation, altered membrane appearance or tympanosclerosis (51%), otorrhea (risk of 13%), and cholesteatoma (rare, occurring 10 to 20 years later). While these complications may be noted in children with OME who do not undergo surgery, they are seen more frequently in

MEDICAL DEVICE USE IN CHILDREN

207

children who had tube insertion [29]. Most randomized clinical trials of OME provide only short-term follow-up (days to weeks). Insertion of tubes results in significant short-term improvement in hearing. However, the long-term sequelae are not fully understood. Outcomes of various therapies are evaluated using different measurements of OME resolution (otoscopic findings and tympanometry) and hearing assessments (audiometry). Few have adequately assessed the developmental outcomes and disabilities for these young children over longer periods of follow-up. There is no consistent, reliable scientific evidence that strongly links OME with these long-term sequelae. The decision to treat a child with OME will depend on the duration and severity of the problem as well as the age of the child (which may affect the likelihood of associated developmental problems) and comorbid conditions (e.g. a craniofacial defect, such as cleft palate) [22]. Research supports the use of tympanostomy tubes to treat high-risk children who have persistent OME with recurrent bouts of acute otitis media, bilateral hearing loss (hearing loss of 20 dB or more) with developmental delay and behavior problems, and more complicated disease (e.g. OME with severe tympanic membrane retraction pockets). It remains controversial when and whether to use tympanostomy tubes in healthy young children who have OME. One large randomized clinical trial (RCT) looked at early vs. late tube placement for persistent OME in young children (mean age less than 18 months) and found that the earlier treated group had significantly fewer problems with effusion immediately after surgery, but found no differences in long-term developmental outcome [31]. These authors extended their study to 6 yearold children and still found that tympanostomy tubes did not improve developmental outcomes for school-age children [32]. Another large RCT looked at the effect of early tube insertion vs. observation alone in children (mean age, three years) with outcome assessments of hearing loss and behavior. They reported some short-term, but no long-term, benefits [33]. Because of increasing evidence suggesting that there are no long-term gains for healthy children, the Agency for Healthcare Policy and Research (now known as the Agency for Healthcare Research and Quality) recommended tympanostomy tube insertion only after months of observation [34,35].

Orthopedic devices in the growing child During middle childhood, many orthopedic appliances are frequently used to treat skeletal trauma or to correct a variety of deformities. Device use in this age group should take into account the ongoing physical growth of these children, as well as their high energy level and active lifestyle. Long-bone fractures occur frequently and are different from those seen in adulthood. The healing and remodeling abilities of developing bones in children are greater, and care must be taken to avoid long-term growth problems. Problems with non-union are rare, and some non-displaced fractures may need only minimal intervention, either simple splinting or casting. However, there may be separation or damage to the epiphysis (growth plate) and these fractures require open surgical reduction with device fixation and pinning. Device technology has evolved and flexible

208

CH15

PEDIATRIC MEDICAL DEVICE USE

plastic nails are now commonly used to treat many fractures, but overly active and often non-compliant children may still be at risk for re-fracture of these mended bones. Adverse fracture sequelae may manifest as different types of growth disturbances, including: premature partial closure of the growth plate, leading to angular deformity; complete interruption of the growth plate, resulting in the affected leg being shorter than the other; or, conversely, as the fracture heals there may be increased blood supply and overgrowth of the affected bone. Femoral fractures pose a particular challenge when they occur in the school-aged child. Very different from young pre-school children who are treated with spica casts, school-aged children are not skeletally mature, need to be mobile, must attend school, and have higher risk for problems associated with prolonged bed rest. These factors have led to an increased use of internal and external fixation devices in this age group [36–38]. Rigid intramedullary nails inserted into growing young children may be used; however, according to the results of a 1998 survey of pediatric orthopedists, avascular necrosis of the femoral head may rarely occur, and intramedullary rods are not recommended in children initially treated with external fixation because of an increased risk of infection [39,40]. Elastic nails are increasingly being used to treat closed femoral fractures and have the advantages of being minimal invasive, resulting in good callus formation with little scarring, showing very low rates of re-fracture and infection, and allowing the child’s early return to activities [41]. Long-term follow-up success for most musculoskeletal repairs really should not be judged until after skeletal maturity and ‘injury to a growing bone is always prone to longterm complications that may influence the patient’s entire life’ [42]. Mechanisms of injury, healing, and complications are also different for childhood spinal injuries involving the thoracic and lumbar areas [43]. An increasing number of case series reports of late-onset infections in pediatric spinal surgeries is of concern, with varying incidence rates of 1–7% [44]. One of the largest case series (n ¼ 937) found a 5% incidence of lateonset infections up to 8 years following spinal surgeries to correct spinal deformities of congenital, neuromuscular, and idiopathic types of scoliosis [45]. The etiology of lateonset infections (greater than 1 year postoperative posterior fusion for idiopathic adolescent scoliosis and congenital scoliosis types) has been thought to relate to the bulk of the implant materials, reaction to metal components, and possible contamination at the time of surgery. These severe complications pose additional challenges for surgeons regarding reimplantation, so as not to lose the spinal curve correction. They highlight the need for longer-term follow-up studies. Recently there has been much interest in evaluating outcomes for children undergoing spinal and other musculoskeletal instrumentation for various orthopedic conditions of childhood, especially spinal disorders (congenital scoliosis, idiopathic scoliosis, and kyphosis). A special survey instrument, the Pediatric Outcomes Data Collection Instrument (PODCI) questionnaire, has recently been shown to be reliable and valid in the assessment of function, disability, and patient satisfaction for children aged 2–18 years with different musculoskeletal conditions, and is used to compare endpoints with normal same-age children [46]. This instrument includes interview information from both the child and parent, with assessment of efficacy and child satisfaction related to their medical conditions and orthopedic interventions with regards to comfort, pain, function including

MEDICAL DEVICE USE IN CHILDREN

209

participation in sports and other activities, and overall satisfaction. It importantly aims to better understand both functional and emotional outcomes for children [47]. Advances and innovations in device technology have led to the use of newer implants, including bio-absorbable fracture fixation devices and mechanical growth plates, but these have not been well studied and for the most part are off-label devices for children. Resorbable fixation devices are increasingly being used in the repair and construction of craniofacial defects and injuries, which pose unique challenges for restoration of growth and function. While enthusiasm for their use continues, with a recent large cohort study (1883 pediatric patients) reporting a lower rate of reoperation than for metal implants [48], an earlier retrospective study of 100 very young pediatric patients who received resorbable plates with metal screws for bony craniofacial abnormalities and defects emphasized the need for further study of potential problems and long-term complications, such as stability, possible foreign body reactions, and infection [49]. Certain spinal disorders unique to childhood, such as severe congenital scoliosis, have led to the development of devices, such as spinal growing rods and the Vertical Expandable Prosthetic Titanium Rib implant (VEPTR), that aim to address special problems of spine fusion procedures performed during childhood. Recent studies suggest that early spinal fusion results in loss of thoracic volume with restricted lung growth and sometimes may result in significant pulmonary morbidity [50]. Other growing rods and the use of anterior spinal stapling are currently being studied. Much of the rationale for newer orthopedic devices has emerged from anecdotal clinical experience, device improvisation, and expert opinion, rather than through an evidence-based approach and conduct of pediatric clinical trials. Further survey and cross-sectional studies of young children are desperately needed.

Devices used in corrective surgeries during adolescence The period of adolescence makes up most of the second decade of life. Guidance issued by the CDRH, defining the pediatric age range up to 21 years, supports the consideration that issues of physical and hormonal growth and emotional/behavioral development should be carefully addressed with device use in adolescent patients [63]. Human factors in device use should be considered as teenagers strive to become independent and take over their own medical care. One cannot always assume there will be parental supervision and good judgment in the use of medical devices. This is an age group known for its risk-taking behavior and noncompliance. Also, teens have heightened concerns about their bodily appearance and self-image. In order to optimize the clinical outcome of elective surgery during adolescence, timing is crucial. Adolescence is the second most dramatic period of body growth (second only to the fetal growth period). Girls and boys may grow 6 to 12 inches in height during their growth spurt and undergo dramatic sexual maturation changes. Many of the body organ systems are completing their final growth. While elective orthopedic surgeries will frequently be postponed until skeletal maturation has been attained, some surgeries, such as scoliosis repair, must be done before spinal growth

210

CH15

PEDIATRIC MEDICAL DEVICE USE

disturbances compromise lung function and negatively impact the postoperative cosmetic results. Umbilical and inguinal hernias that were undetected in earlier years may also be repaired when they are discovered in teenage years. In boys, large varicoceles in the scrotum and hypospadias (birth anomaly with misplaced urethral opening) repairs may be performed, with hopes to improve both aesthetics and function. Frequently, cosmetic surgeries, including rhinoplasty (nose reshaping), otoplasty (ear surgery), and various breast surgeries, including augmentation surgeries, are performed. According to American Society of Plastic Surgeons (ASPS) website and see that for 2005 there was a total of more than 333 000 cosmetic procedures performed on young patients under age 19. It is important to understand the patient motivation and expectation for having surgery [52,53]. An increasing demand for plastic surgical procedures may in part be related to increased media coverage about enhancing cosmetic procedures. Appropriate selection of patients and assessment of their physical and emotional maturation is crucial [54]. Few articles in the literature address teen attitudes regarding cosmetic surgery issues [53]. One small study of boys and girls examining their attitudes towards cosmetic surgery supports that teens should be screened for underlying depression and self-image disturbance to be sure they will not continue to seek body improvement surgeries. They emphasize that possible underlying psychological disturbance and underlying psychiatric problems should be appropriately identified, and addressed and treated before cosmetic surgery is performed. Recently there has been an increase in the number of teens undergoing breast implant surgery. Saline breast implants are FDA-approved in persons 18 years and older. However, being bombarded by the media images, small-breasted girls may feel inadequate and opt for a breast implant augmentation with unrealistic expectations. A large Danish cohort study of young women opting for enhancing breast augmentation with implants also supported the need to evaluate young women for underlying psychiatric problems [55]. The American Society of Plastic Surgeons website reports 3841 breast augmentations among women aged 18 years or younger. This ‘cup and gown’ procedure is sometimes paid for by parents as a high school graduation gift. Elective cosmetic surgery and other reconstructive technologies which are offered to teens may not always be appropriate and pose risks different from those seen in adults. Operating on tissues that have not fully developed may be problematic, and surgery might interfere with growth or cause problems years later. While most would agree that breast surgery should be postponed until the breast has completed its growth, the breast becomes mature during adolescence but is not fully complete until about age 25. Further maturation occurs with pregnancy [56]. In males, gynecomastia is a common condition that occurs during adolescence and usually regresses spontaneously within 1 year, but moderate to severe cases that do not resolve may undergo reduction mammoplasty surgery for cosmetic and psychological sequelae. While surgery in the past has used open excision techniques, newer liposuction technology has been used with the aim of avoiding large areas of scarring and nerve and vessel injury. Ultrasound-assisted liposuction (UAL) has been recently used to treat gynecomastia in both adolescent and adult males. UAL selectively destroys adipose

SPECIAL DEVICE RISKS AND SAFETY CONCERNS FOR CHILDREN

211

tissue and proponents report that it offers benefits of less swelling, bleeding and can be used in more fibrous breast tissue areas [57,58]. Disadvantages include increased operative time and the involved training. However, most studies involve case series by one surgeon and a wide age range of patients. Further investigations need to carefully examine patient satisfaction, cosmetic results, and incidence of complications for the adolescent age group. It is important that indications for device intervention during adolescence be conservative until appropriately designed clinical studies have demonstrated the safety and efficacy of these techniques when applied to teens. Studies should be conducted to better understand what adverse events are unique for teenagers and what preventive measures may be taken to ensure successful outcomes with device use.

Special device risks and safety concerns for children The pediatric population may be particularly vulnerable to various exposures associated with medical device use. Safety assessment for devices must consider the child’s age, size, body surface area, growth and development, physiology, and hormonal stage. Underlying disease and nutrition status are also important factors. Short-term, longterm, and cumulative exposures, as well as latent effects, must be considered, especially if the device is to be re-used or implanted for several years or even a lifetime (e.g. ventriculo-peritoneal shunts, orthopedic implants). Possible risks for carcinogenicity, reproductive toxicity, and longer-term effects of many medical devices are poorly understood, unknown, or difficult to assess. Two examples are illustrative of these concerns.

Di-2-Ethylhexylphthalate (DEHP) Di-2-Ethylhexylphthalate (DEHP) is a plasticizer that is commonly used to enhance flexibility in a wide variety of medical devices, such as intravenous catheters and tubing as well as various fluid- and blood-carrying containers and bags. It is not chemically bound within these devices and over the past few years there has been growing concern about the leakage or migration of the DEHP into body fluids and organs [59–61]. DEHP has been shown to produce different side effects in laboratory animals and is known to be an endocrine disrupter (causing adverse events by interfering with the endocrine system), resulting in antiandrogenic effects in male rats, with toxic effects on the development of the male reproductive system and production of sperm [62,63]. It also lowers 17ß-estradiol level in the blood of female rats [64]. While human studies indicate that it may be relatively safe, there continues to be concern of possible reproductive and developmental toxicity of DEHP, suggesting that extra precautions should be taken to limit DEHP exposure to especially high-risk patient groups, particularly during the fetal, neonatal, and peripubertal stages of development. CDRH recommends limited exposure and suggests alternatives for persons who may be at higher risk [65].

212

CH15

PEDIATRIC MEDICAL DEVICE USE

Radiation exposure in children The biological effects of ionizing radiation exposure from devices are important. It is generally accepted that there is no minimal threshold that can be said to be harmful. In practice, it is recommended that radiation levels be kept down to the lowest practicable level. One cannot say there is a safe dose. Children are considerably more sensitive to radiation than adults. With a longer life expectancy, children have a larger window of opportunity for radiation effects, namely carcinogenic risk [66]. One needs to assess lifetime risk and understand the genetic and carcinogenetic effects on children [67]. Of recent concern is the increasing use of pediatric computed tomography (CT), an invaluable tool for imaging children, especially those who are younger, sick, and less cooperative. CT has been useful in the evaluation of head trauma patients and delineating abnormal lesions, especially soft tissues areas. However, with increasing numbers of CT imaging being performed, the quantitative lifetime radiation risks for children undergoing CT need to be considered [68]. Attributable risk for cancer is not negligible and is higher for children than adults [68]. Research supports that pediatric CTexposures could be lowered in their milliampere doses to lower the radiation risk but still allow critical radiological assessment of children.

Regulatory framework Premarket submissions As noted in Chapter 2, there are several regulatory pathways to marketing devices. Devices marketed for pediatric use can be considered to fit into one of three broad categories, namely devices that: (a) may be used in both adult and pediatric patients; (b) require special sizes for pediatric patients; and (c) are only indicated for pediatric use. Examples of devices in the first category include syringes, wound dressings, and monitoring and imaging devices. Generally, there are no specific requirements for premarket testing of these products in pediatric patients. Examples of devices in the second category include heart valves, orthopedic implants, and ventricular assist devices. Premarket laboratory testing is required to assess the performance of these smaller devices; however, pediatric clinical testing may or may not be required. Factors that can influence the need for premarket pediatric clinical data include the nature of the device, the nature of the disease or condition, what is known about the performance of the device in adults, and the extent to which adult data can be extrapolated to pediatric use. Devices that are only indicated for pediatric use, the last category, include products such as newborn hearing screeners and occluders for patent ductus arteriosus. Studies of devices like these are conducted in children only when it is deemed that premarket clinical data are needed. In 2004, FDA issued a guidance entitled ‘Premarket Assessment of Pediatric Medical Devices,’ which, among other items, addresses the type of information necessary to provide reasonable assurance of safety and effectiveness for devices intended for use in pediatric populations [1].

REGULATORY FRAMEWORK

213

Devices in the first category are generally marketed via premarket notification [i.e. 510(k)], whereas devices in the latter two categories can also be marketed via premarket applications (i.e. PMAs), product development protocols (i.e. PDPs), or through humanitarian device exemptions (i.e. HDEs; see Chapter 2). A greater proportion of devices indicated for pediatric use, compared to adult use, are marketed as HDEs, since the target population is much smaller (in terms of numbers treated with the condition).

Legislation The Medical Device User Fee and Modernization Act (MDUFMA) addresses some issues that affect postmarket surveillance and epidemiology. For instance, MDUFMA requested that FDA enter into an agreement with the Institute of Medicine (IOM) for the latter to conduct a study for the purposes of determining whether FDA’s system for postmarket surveillance of medical devices provides adequate safeguards regarding the use of devices in pediatric populations. ‘Postmarket surveillance’ was meant to include postmarket activities, such as mandated studies, reports of adverse events and product problems related to device use, and registries. The IOM completed its work in 2005 and issued their report, Safe Medical Devices for Children [69]. The report has several recommendations, many of which extend beyond the pediatric population because the systemic shortcomings are not tied to any specific age group. Highlights of the IOM recommendations include: 1. FDA establishing a system for monitoring and publicly reporting on the status of postmarket studies, as well as the methods, findings, and disposition of these studies. 2. Amending legislation to allow for ordering studies as a ‘condition of clearance’ for 510(k) products, analogous to that for PMA products. 3. FDA establishing a central point of responsibility for pediatric issues with its device Center. 4. The need for children’s hospitals and other user facilities to establish a focal point for medical device safety. 5. FDA focusing more attention on adverse device events, e.g. through reporting system linkages and timely device labeling updates. 6. FDA promoting the development and adoption of common device coding and other related standards for linking use and outcome data, as well as FDAworking with other agencies and outside experts to strengthen methods and tools for epidemiologic research on device safety.

214

CH15

PEDIATRIC MEDICAL DEVICE USE

7. FDA collaborating with other research funding agencies and interested parties to define a research agenda and priorities for further evaluation of device safety and effectiveness in children. Having understood and anticipated many of these issues, FDA, by mid-2005, had efforts in place to address many of these issues. For instance, a tracking system is in place to monitor the status of postmarket studies and the Center has a Pediatric Steering Committee, working in conjunction with the Office of Pediatric Therapeutics, to evaluate the Center’s use of pediatric expertise and attention to pediatric device issues.

The future for pediatric medical device surveillance and epidemiology The promise of surveillance and epidemiology of pediatric medical devices resides in progress made on several fronts noted in this and other chapters in this book. The importance of developing a healthcare research infrastructure amenable to devicerelated investigations is paramount. First and foremost, that would include the ability to uniquely name and identify products to the model level and efficiently record and extract such information from healthcare-related databases. As noted in the chapter on device nomenclature (Chapter 7), collaborative efforts are under way to achieve this end, but much is yet to be done. The IOM report [69] clearly calls for such efforts. Equally important would be the refinement and/or development of databases to incorporate such information, whether it is in national datasets such as Medicare or in device-specific registries overseen by professional societies. FDA is increasingly collaborating with outside entities to explore and utilize these databases. In conjunction with these efforts is the need to greatly enhance documentation of adverse device-related outcomes in healthcare records. FDA, in collaboration with major stakeholders, is exploring means to begin to address this issue. Linking accurate and specific use data with accurate and specific outcome data in accessible databases is essential for robust and effective epidemiology, and device epidemiology is no exception. The IOM report makes clear the need for such developments by calling for strengthening methods and tools for epidemiologic research on device safety. In addition, as noted previously in the list of IOM recommendations, is the need for FDA to collaborate with other research funding agencies, such as the National Institutes of Health and the Center for Medicare and Medicaid Services, as well as other interested parties, to define a research agenda and priorities for further evaluation of device safety and effectiveness in children. The recent selection of a center of excellence devoted to medical device issues within the Centers for Education and Research on Therapeutics (CERTs) program, overseen collaboratively by the Agency for Healthcare Research and Quality and FDA, represents a viable avenue to begin to explore these issues. More immediate efforts are also under way to enhance the field. As previously noted in the chapter on the Medical Product Safety Network (MedSun; Chapter 5), nearly two

REFERENCES

215

dozen pediatric hospitals have been recruited as of the end of 2005. Examples of pediatricspecific issues that have been noted to date include: an extracorporeal membraneous oxygenation system (a portable cardiopulmonary bypass machine) alarm that is too faint for intensive care units; failure of a neonatal ventilator secondary to corrosion of components that do not meet current material standards; and an infant with a perforated colon secondary to air-reduction of an intussusception via a Foley catheter. The MedSun network has also been pilot-tested to conduct active survey-based surveillance. The initial efforts focused on adult cardiovascular devices, e.g. drug-eluting coronary stents and thrombosis, but future efforts will involve pediatric devices and focus on the subset of pediatric hospitals. MedSun is well-positioned to strengthen device-related surveillance and safety activities within pediatric hospitals, as called for in the IOM report. Registries are another arena of fertile development in device epidemiology. More emphasis has recently been placed by the FDA on registries as a mechanism to collect useful postmarket device performance information as a condition of approval of new products, e.g. carotid stents and urethral bulking agents. In addition, the National Institutes of Health will award a grant in 2005 for oversight of a national registry to collect detailed information on patients (including pediatric) implanted with left ventricular-assist devices. Reimbursement by the Center for Medicare and Medicaid Services will be contingent upon entering data into the registry. Similar efforts are being explored for other high-profile devices. In addition, professional societies are increasingly sponsoring the collection of short-term data in national registries, such as the National Cardiovascular Data Registry, sponsored by the American College of Cardiology, and the National Cardiothoracic Surgery Database, sponsored by the Society of Thoracic Surgeons. FDA has collaborated in studies using these and other registries [70,71]. Although no studies to date have been pediatric-based, the systems and methods in place will allow for such endeavors. These efforts combined augur well for a robust and promising future for the continued development of the epidemiology of pediatric medical devices.

References 1. Guidance for Industry and FDA staff: premarket assessment of pediatric medical devices, issued May 14 2004: http://www.fda.gov/cdrh/mdufma/guidance/1220.pdf 2. Baskin M, O’Rourke EJ, Fleisher GR. Outpatient treatment of febrile infants 28–89 days of age with intramuscular administration of ceftriaxone. J Pediatr 1992; 120: 22–27. 3. Bonadio WA. Evaluation and management of serious bacterial infections in the febrile young infant. Pediatr Infect Dis J 1990; 9(12): 905–912. 4. Herr SM, Wald ER, Pitetti RD et al. Enhanced urinalysis improves identification of febrile infants ages 60 days and younger at low risk for serious bacterial illness. Pediatrics 2001; 108(4): 866–871. 5. Slater M, Krug SE. Evaluation of the infant with fever without source: an evidence-based approach. Emerg Med Clin North Am 1999; 17: 97–126. 6. Watson RS, Carcillo JA. Scope and epidemiology of pediatric sepsis. Pediatr Crit Care Med 2005; 6(3 suppl): S3–5.

216

CH15

PEDIATRIC MEDICAL DEVICE USE

7. Robinson JL, Seal RF, Spady DW, Michel MR. Comparison of esophageal, rectal, axillary, bladder, tympanic, and pulmonary artery temperatures in children. J Pediatr 1998; 133(4): 553– 556. 8. Hay WW, Hayward A, Levin MJ, Sondheimer JM. Current Pediatric Diagnosis and Treatment, 15th edn. New York, NY: McGraw-Hill, 2001. 9. Press S, Quinn BJ. The pacifier thermometer. Comparison of supralingual with rectal temperatures in infants and young children. Arch Pediatr Adolesc Med 1997; 151(6): 551. 10. Callanan D. Detecting fever in young infants: reliability of perceived, pacifier, and temporal artery temperatures in infants younger than 3 months of age. Pediatr Emer Care 2003; 19(4): 240–243. 11. Greenes DS, Fleisher GR. Accuracy of a noninvasive temporal artery thermometer for use in infants. Arch Pediatr Adolesc Med 2001; 155(3): 376–381. 12. Barton SJ, Gaffney R, Chase T et al. Pediatric temperature measurement and child/parent/nurse preference using three temperature measurement instruments. J Pediatr Nurs 2003; 18(5): 314–320. 13. Drake-Lee A, Mantella I, Bridle A. Infrared ear thermometers vs. rectal thermometers. Lancet 2002; 360(9348): 1883. 14. Kelly B, Alexander D. Effect of otitis media on infrared tympanic thermometry. Clin Pediatr 1991; 30(suppl): 46–49. 15. Saxena AK, Topp SS, Heinecke A, Willital GH. Application criteria for infrared ear thermometers in pediatric surgery. Technol Health Care 2001; 9(3): 281–285. 16. Porwancher R, Sheth A, Remphrey S et al. Epidemiological study of hospital-acquired infection with vancomycin-resistant Enterococcus faecium: possible transmission by electronic ear-probe thermometer. Infect Control Hosp Epidemiol 1997; 18: 771–773. 17. Yaron M, Lowenstein SR, Koxiol-McLain J. Measuring the accuracy of the infrared tympanic thermometer: correlation does not signify agreement. J Emerg Med 1995; 13(5): 617–621. 18. AAP Committee on Environmental Health. Mercury in the environment: implications for pediatricians. Pediatrics 2001; 108: 197–205. 19. Jones HL, Kleber CB, Eckert GJ, Mahon BE. Comparison of rectal temperature measured by digital vs. mercury glass thermometer in infants under two months old. Clin Pediatr 2003; 42(4): 357–359. 20. Craig JV, Lancaster GA, Taylor S et al. Infrared ear thermometry compared with rectal thermometry in children: a systematic review. Lancet 2002; 360(9333): 603–609. 21. Greenes DS, Fleisher GR. When body temperature changes, does rectal temperature lag? J Pediatr 2004; 144(6): 824–826. 22. Rovers MM, Schilder AG, Zielhuis GA et al. Otitis media [review]. Lancet 2004; 363(9407): 465–473. 23. Teele DW, Klein JO, Rosner B. Epidemiology of otitis media during the first seven years of life in children in greater Boston: a prospective, cohort study. J Infect Dis 1989; 160(1): 83–94. 24. Paradise JL, Rockette HE, Colborn DK et al. Otitis media in 2253 Pittsburgh-area infants: prevalence and risk factors during the first two years of life. Pediatrics 1997; 99(3): 318–333. 25. Fiellau-Nikolajsen M. Tympanometry in three-year old children. Prevalence and spontaneous course of MEE. Ann Otol Rhinol Laryngol Suppl 1980; 89(3, pt 2): 223–227. 26. Zeilhuis GA, Rach GH, van den Broek P. Screening for otitis media with effusion in preschool children. Lancet 1989; 1(8633): 311–314. 27. Pokras R, Kozak LJ, McCarthy E. Ambulatory and inpatient procedures in the United States, 1994. Vital Health Stat 13 1997; 132: 1–113. 28. Bright RA, Moore RM, Jeng LL et al. The prevalence of tympanostomy tubes in children in the United States, 1988. Am J Publ Health 1993; 83(7): 1026–1028. 29. Kay DF, Nelson M, Rosenfeld RM. Meta-analysis of tympanostomy tube sequelae. Otolaryngol Head Neck Surg 2001; 124(4): 374–380. 30. Maw AR. Development of tympanosclerosis in children with otitis media with effusion and ventilation tubes. J Laryngol Otol 1991; 105(2): 71–77.

REFERENCES

217

31. Paradise J, Feldman HM, Campbell TF et al. Effect of early or delayed insertion of tympanostomy tubes for persistent otitis media on developmental outcomes at the age of three years. N Engl J Med 2001; 344(16): 1179–1187. 32. Pardise JL, Campbell TF, Dollaghan CA et al. Developmental outcomes after early or delayed insertion of tympanostomy tubes. N Engl J Med 2005; 353(6): 576–586. 33. Wilks J, Maw R, Peters TJ et al. Randomised controlled trial of early surgery versus watchful waiting for glue ear: the effect on behavioral problems in pre-school children. Clin Otolaryngol Allied Sci 2000; 25(3): 209–214. 34. Stool SE, Berg AO, Berman S, Camey CJ, Cooley JR, Culpepper L, Eavey RD, Feagans LV, Finitzo T, Friedman EM. et al. Otitis Media with Effusion in Young Children. Clinical Practice Guideline, Number 12. AHCPR Publication No. 94-0622. Rockville, MD: Agency for Health Care Policy and Research, Public Health Service, U.S. Department of Health and Human Services. July 1994. 35. Roddey OF, Hoover HA. Otitis media with effusion in children: a pediatric office perspective. Pediatr Ann 2000; 29(10): 623–629. 36. Flynn JM, Skaggs DL, Sponseller PD et al. The surgical management of pediatric fractures of the lower extremity. Instr Course Lect 2002; 52: 647–659. 37. Flynn JM, Luedtke LM, Ganley TJ et al. Comparison of titanium elastic nails with traction and a spica cast to treat femoral fractures in children. J Bone Joint Surg Am 2004; 86A(4): 770–777. 38. Sponseller PD. Surgical management of pediatric femoral fractures. Instr Course Lect 2002; 51: 361–365. 39. Sanders JO, Browne RH, Mooney JF, Raney EM et al. Treatment of femoral fractures in children by pediatric orthopedists: results of a 1998 survey. J Pediatr Orthop 2001; 21(4): 436–441. 40. Letts M, Jarvis J, Lawton L, Davidson D. Complications of rigid intramedullary rodding of femoral shaft fractures in children. J Trauma 2002; 52(3): 504–516. 41. Hunter JB. Femoral shaft fractures in children. Injury 2005; 36(suppl 1A): 86–93. 42. Togrul E, Bayram H, Gulsen M et al. Fractures of the femoral neck in children: long-term followup in 62 hip fractures. Injury 2005; 36: 123–130. 43. Black BE, O’Brien E, Sponseller PD. Thoracic and lumbar spine injuries in children: different than in adults. Contemp Orthop 1994; 29(4): 253–260. 44. Soultanis K, Mantelos G, Pagiatakis A et al. Late infection in patients with scoliosis treated with spinal instrumentation. Clin Orthop Relat Res 2003; 411: 116–123. 45. Muschik M, Luck W, Schlenzka D. Implant removal for late-developing infection after instrumented posterior spinal fusion for scoliosis: reinstrumentation reduces loss of correction. A retrospective analysis of 45 cases. Eur Spine J 2004; 13(7): 645–651. 46. Daltroy LH, Liang MH, Fossel AH et al. Pediatric Outcomes Instrument Development Group. The POSNA pediatric musculoskeletal functional health questionnaire: report on reliability, validity, and sensitivity to change. J Pediatr Orthop 1998; 18: 561–571. 47. Lerman JA, Sullivan E, Haynes RJ. The Pediatric Outcomes Data Collection Instrument (PODCI) and functional assessment in patients with adolescent or juvenile idiopathic scoliosis and congenital scoliosis or kyphosis. Spine 2002; 27(18): 2052–2057. 48. Eppley BL, Morales L, Wood R et al. Resorbable PLLA-PGA plate and screw fixation in pediatric craniofacial surgery: clinical experience in 1883 patients. Plast Reconstr Surg 2004; 114(4): 850–856. 49. Rubin JP, Yaremchuk MF. Complications and toxicities of implantable biomaterials used in facial reconstructive and aesthetic surgery: a comprehensive review of the literature. Plastic Reconstr Surg 1997; 100: 1336. 50. Campbell RM Jr, Smith MD, Mayes TC et al. The effect of opening wedge thoracostomy on thoracic insufficiency syndrome associated with fused ribs and congenital scoliosis. J Bone Joint Surg Am 2004; 86A(8): 1659–1674.

218

CH15

PEDIATRIC MEDICAL DEVICE USE

51. American Society of Plastic Surgeons (National Clearinghouse of Plastic Surgery Statistics: Procedural Statistics Trends, 1992–2004) website: www.plasticsurgery.org [accessed November, 2006]. 52. McGrath MH, Schooler WG. Elective plastic surgical procedures in adolescence. Adolesc Med Clin 2004; 15(3): 487–502. 53. Pearl A, Weston J. Attitudes of adolescents about cosmetic surgery. Ann Plast Surg 2003; 50(6): 628–630. 54. McGrath MH, Mukerji S. Plastic surgery and the teenage patient. J Pediatr Adolesc Gyn 2000; 13: 105–118. 55. Jacobsen PH, Holmich LR, McLaughlin JK, Johansen C et al. Mortality and suicide among Danish women with cosmetic breast implants. Arch Intern Med 2004; 164(22): 2450–2455. 56. Neinstein LS. In Breast Disorders in Adolescent Health Care: A Practical Guide, 4th edn. Philadelphia, PA: Lippincott, Williams and Williams, 2002; 1063–1084. 57. Fruhstorfer BH, Malata CM. A systematic approach to the surgical treatment of gynecomastia. Br J Plast Surg 2003; 56(3): 237–246. 58. Graf R, Auersvald A, Damasio RC et al. Ultrasound-assisted liposuction: an analysis of 348 cases. Aesthetic Plast Surg 2003; 27(2): 146–153. 59. Allwod MC. The release of phthalate ester plasticizer from intravenous administration sets into fat emulsion. Int J Pharm 1986; 29: 233–236. 60. Loff S, Kabs F, Witt K et al. Plyvinylchloride infusion lines expose infants to large amounts of toxic plasticizers. J Pediatr Surg 2000; 35(12): 1775–1781. 61. Tickner JA, Schettler T, Guidotti T, McCally M et al. Health risks posed by the use of di-(2ethylhexyl) phthalate (DEHP) in PVC medical devices: a critical review. Am J Ind Med 2001; 39: 100–111. 62. Poon R, Lecavalier P, Mueller R et al. Subchronic oral toxicity of di-n-octyl phthalate and di(2ethylhexyl) phthalate in the rat. Food Chem Toxicol 1997; 35: 225–239. 63. Tyl RW, Price CJ, Marr MC, Kimmel CA. Developmental toxicity evaluation of dietary di-(2ethylhexyl) phthalate in Fischer 344 rats and CED-1 mice. Fundament Appl Toxicol 1988; 10: 395–412. 64. Davis BJ, Maronopot RR, Heindel JJ. D-(2-ethylhexyl) phthalate (DEHP) suppressed estradiol and ovulation in cycling rats. Toxicol Appl Pharmacol 1994; 128: 216–223. 65. Center for Devices and Radiological Health, US Food and Drug Administration, 2001. Safety assessment of di(2-ethylhexyl) phthalate (DEHP) released from PVC medical devices. Last updated web site July 12 2002: www.fda.gov/cdrh/safety/dehp.html [accessed November, 2006]. 66. National Cancer Institute website: Radiation Risks and Pediatric Computed Tomography (CT): A Guide for Health Care Providers: www.cancer.gov/cancertopics/causes/radiation-risks-pediatric-CT 67. Curry TS III, Dowdey JE, Murry RC Jr. Christensen’s Physics of Diagnostic Radiology, 4th edn. Philadelphia, PA: Lea & Febiger, 1990; 372–391. 68. Brenner D, Elliston C, Hall E, Berdon W. Estimated risks of radiation-induced fatal cancer from pediatric CT. Am J Roentgenol 2001; 176(2): 289–296. 69. Committee on Postmarket Surveillance of Pediatric Medical Devices, Board on Health Sciences Policy. Safe Medical Devices for Children, Marilyn J. Field MJ, Tilson H (eds[LAW3]). Institute of Medicine of the National Academies, Washington, DC, July, 2005. 70. Peterson ED, Kaul P, Kaczmarek RG et al. From controlled trials to clinical practice: monitoring transmyocardial revascularization use and outcomes. J Am Coll Cardiol 2003; 42: 1611–1616. 71. Tavris DR, Gallauresi B, Lin B et al. Risk of local adverse events following cardiac catheterization by hemostasis device use and gender. Pharmacoepidemiol Drug Safety 2003; 12(suppl 1): 144.

16 Selected medical devices used to manage diabetes mellitus Shewit Bezabeh MD, MPH and Joy H. Samuels-Reid MD FDA/CDRH/ODE/DAGID

Dale R. Tavris, MD, MPH FDA/CDRH/OSB/DPS

Introduction Diabetes mellitus (DM) is a major public health problem. In the United States (US), more than 18.2 million people are affected with the disease, and this number is expected to double by 2050 [1]. In addition, it is estimated that there are more than 5 million undiagnosed individuals [2]. The Center for Disease Control Diabetes Surveillance System showed that from 1980 to 2002, the number of Americans with diabetes more than doubled (from 5.8 to 13.3 Million) [3]. It is estimated that the direct and indirect cost of the disease was $132 billion in 2002 and that diabetes was the 5th leading cause of death [4]. The long- and short-term health consequences of poorly controlled diabetes are well known. Poorly controlled DM is a risk factor for cardiovascular disease, retinopathy, nephropathy, and neuropathy. Acutely, poorly regulated and uncorrected blood glucose can result in hyperglycemia, hypoglycemia and/or diabetic ketoacidosis (DKA), with resultant serious health hazards such as seizures, coma, or death. Before the discovery of insulin in 1921, strict dieting and starvation were the only therapeutic modality for DM, and the disease resulted in high mortality shortly after diagnosis. The discovery and use of insulin changed the disease from a terminal disease

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

220

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

to a chronic disease. Diet, exercise, various oral anti-diabetic medications and insulin remain the cornerstone in the management of the disease. It is estimated that there are over 6 million diabetics treated with insulin in the US [5]. The Diabetes Control and Complications Trial (DCCT) [6] published in 1993, and the United Kingdom Prospective Diabetes study clearly demonstrated that intensive blood glucose control significantly delayed the long-term complications of the disease [7]. The risk of microvascular events such as retinopathy, nephropathy, and neuropathy were significantly reduced when compared with conventional therapy. After the publication of these landmark studies, achieving tight blood glucose levels has become the standard of care in DM management. Most clinical guidelines reflect the current standard of care for management of DM and use the Hemoglobin A1c (HbA1C) test to assess the degree of blood glucose control. These guidelines recommend aiming for and maintaining an HbA1c level of 7% or less [8]. Intensive therapy requires multiple daily insulin (MDI) injections or a continuous subcutaneous insulin infusion (CSII) by external pump (see Figure 16.1). Intensive therapy also requires the patient to have frequent daily finger sticks to monitor blood glucose levels, since both MDI and CSII therapies are associated with increased risk of hypoglycemia compared to conventional management. In the DCCT, the rate of hypoglycemia was three times more in the treatment (MDI) group when compared to the control (conventional) group [6].

Figure 16.1 Portable insulin infusion pump, c 1980s. Reproduced with permission from the Science Museum/Science & Society Library

INSULIN DELIVERY DEVICES

221

Medical devices play an important role in managing and monitoring chronic diabetes. These devices are used for lancing, insulin delivery, and measuring blood glucose levels. The FDA regulates these medical devices, through a number of risk-based regulatory controls. Devices are placed into three different risk classes (see chapter 2 for full discussion of regulations). Briefly, Class I devices have well established safety and effectiveness and are subject to general controls where most are exempt from premarket review. Syringes and lancing devices belong to this class. Class II devices have an intermediate risk and require Special Controls (labeling, etc.) to ensure safe and effective use. Class II devices are cleared though the 510 (k) process, also known as premarket notification. Manufacturers must register to notify FDA, at least 90 days in advance, of their intent to market these devices. The FDA then determines whether the device is substantially equivalent to an already marketed Class I or II device. External insulin pumps, insulin pens, jet injectors, glucose meters and test strips are cleared through the 510(k) process. Only 10–15% of 510(k) submissions contain clinical data. Class III devices are those with highest risk or for which there are no adequate predicate devices. They are approved though the Pre-Market Approval (PMA) process in which pre-clinical and clinical data regarding their safety and effectiveness are reviewed prior to approval. Continuous blood glucose sensing devices are approved through the PMA regulatory process. Future stand-alone blood glucose sensing devices, implantable pumps, or artificial mechanical pancreases will most likely require approval through the PMA regulatory process. This chapter will review selected medical devices currently used in the management of DM, as examples of how epidemiology has been used to assess the postmarket performance of these devices. We will also briefly discuss emerging devices that may be used in managing the disease in the near future.

Insulin delivery devices Insulin delivery via pen devices Technological advances in insulin delivery have resulted in devices that offer more convenience, more accurate dosing, and less pain from smaller-gauged needles. The manufacturers of these devices have made claims of better patient compliance, improved quality of life, and better social acceptability. Insulin pen devices, first introduced in 1985, combine the insulin container and the syringe in a single unit. The compact size allows insulin to be administered in a discrete manner. Smaller and improved needles allow a more tolerable injection. There are pre-filled pens, which are single use, and in which the needle is inserted subcutaneously. Insulin is administered when the plunger is pushed down, usually by pressing the top of the pen. There are two types of pens currently in use: pre-filled and reusable [9,10]. Insulin pens meet the needs of patients with busy lives and provide convenience and comfort. Thus, in many parts of the world, insulin pens are the primary source (70% to

222

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

90%) of insulin delivery. Although popular in many parts of the world, use of the insulin pen in the U.S. is approximately 2% of the insulin user population [11]. Most patients seem to find the pens easier to use than needles and syringes which translate to better compliance; fewer missed injections, and consequently, improved outcomes. The pens administer insulin in increments of 0.5 unit, 1 unit, or 2 units. Dosing range varies from 0.5 units to 80 units. Some insulin requires thorough mixing before injection. There is a correlation between the size of the bore and the level of discomfort on injection: smaller bores seem to be less painful. Studies indicate that insulin pens have high dosing accuracy in a properly educated patient when performance and technique are monitored [10]. A study by Plevin and Sadur [10] found that adult patients showed high acceptance rates of insulin pens with 98% reporting that they were easy to use and convenient. The study also reported that 91% wanted to continue their use. Children also seem to accept insulin pens more readily and have more positive attitudes about them. Further, pen delivery appears to fit the lifestyles of children, particularly with respect to school, extra-curricular activities and sports. The pens are available in an array of colors making it easier to differentiate rapid-acting from intermediate-acting insulin devices. They are also less conspicuous, therefore drawing less attention to the user.

Jet injectors Pain is a problem with any traditional method of insulin delivery where needles are involved. Thus, jet injectors have been touted as needleless systems, and ‘‘no needles’’ is often interpreted as no pain. Jet injectors have been used extensively since 1947, and have provided a parenteral delivery system not dependent on a needles and syringes [12]. Jet injectors can deliver vaccines or drugs subcutaneously, intramuscularly, or intradermally. Some injectors employ high pressure and low volume stream of fluid, and others utilize small particles accelerated to a high velocity [12]. Most jet injectors are indicated for insulin administration. The recommended injection sites for the use of jet injectors include arms, legs, abdomen, hips and buttocks. These injectors facilitate the administration of insulin under high-pressure stream into the subcutaneous tissue without the use of a needle. Comparisons of insulin delivered by syringe versus jet injectors have concluded that absorption of insulin was more rapid with jet injectors than with a syringe. However, each insulin delivery system has its own clinical issues. Clinical concerns related to administration of insulin by jet injectors may include, but are not limited to, pain, tissue maceration, lipodystrophy, fat necrosis, infection and crepitation. In addition, dose accuracy will be dependent on depth of penetration, bioavailability, and the route of administration. Human factors such as ease of use, accuracy and reliability can influence the safety and effectiveness of these insulin delivery devices. The three main categories of jet injectors may be grouped as follows: personal use, low workload use (commercial), and high workload (mass inoculator). The safety issues

INSULIN DELIVERY DEVICES

223

concerning of high work load multi-use jet injectors mainly involve the nozzle, particularly nozzle contamination. There is the possibility of inter-patient transfer of blood and fluids. Residues from cleaning agents and drugs may also lead to nozzle contamination. Issues concerning efficacy of jet injectors arise mainly with respect to dosing accuracy. It is often difficult to ensure that the dose is delivered accurately to the target tissue and is bioequivalent to needle injection. It is also difficult to control whether the dose is concentrated or dispersed. Some studies have shown jet injectors to be well accepted in patients with diabetes and offer an alternative to insulin delivery by needles.

External insulin infusion pumps Continuous subcutaneous insulin infusion pumps (CSII) are small programmable devices, about the size of a pager, with a reservoir filled with insulin. Insulin is delivered through a catheter, inserted under the abdomen, to the subcutaneous tissue. The device allows programming of different pre-selected rates (basal rate) as well as bolus doses. CSII allows the user to program different pre-selected basal rates of insulin infusion based on patterns of perceived needs. The device also allows the diabetic user to select from several types of mealtime boluses. The earliest insulin infusion pumps received FDA approval in the late 1980’s. Since then a number of different pumps continue to be cleared through the 510(k) regulatory approval process as class II devices. To date, over 35 external insulin infusion pumps have been cleared [13]. Most recently, technological advancement has allowed the pump to incorporate communication software via an infrared interface allowing the user and the physician to transfer (download) data stored in the pump to a personal computer (PC) for display or print. The communication will also allow the user to program the pump via PC with selected pump settings for basal or bolus infusion. The number of pump users increased after the publication of the DCCT, which showed that achieving and maintaining tight blood glucose control prevented or slowed the progression of microvascular complications in type 1 diabetics. Based on devices sales information, it is estimated that there are about 450,000 pump users globally. In the US, about 250,000 diabetics currently use the device. It is also estimated that 5–10% of these pump users discontinue use for various reasons [14].

Advantage of Insulin Pumps Efficacy of Glycemic control Pump therapy is effective in achieving long-term blood glucose control in type 1 diabetic patients. The DCCT also showed that both CSII and MDI injection therapy are effective means of achieving normoglycemia in the insulinrequiring patient. Since the DCCT, a number of studies have compared the effectiveness of CSII with optimum MDI. A meta-analysis of 12 randomized control trials concluded that glycemic control was slightly better with CSII when compared with ‘‘optimized’’ injection therapy [15]. The same study concluded that the difference in control between

224

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

the two insulin delivery methods were small but significant. The analysis also showed that daily insulin usage is less with CSII. A number of studies enrolling smaller numbers of patients have also demonstrated that CSII is effective compared to MDI [16,17,18]. In 2000, FDA approved new long acting insulin Lantus (glargine) for once a day, bed time injection [19]. Lantus has a slower, more prolonged absorption and a relatively constant concentration/time profile over 24 hours without the pronounced peaks seen with other long acting insulin. Because of these constant serum concentration, Lantus is occasionally referred to as the ‘‘poor man’s pump’’[20]. However, there are no long term studies comparing Lantus with CSII. A short-term, 16-week study of pediatric patients comparing insulin analogs for MDI with CSII showed lower HbA1c were achieved with CSII than with glargine-based MDI treatment [21]. There are a number of factors other than effectiveness to consider when choosing a pump. Some of these factors will be discussed below. Hypoglycemia Initial studies raised concerns of severe hypoglycemia in patients using an insulin pump [22]. The incidence of hypoglycemia was noted to be about three times higher in the intensive therapy group (CSII and MDI) in the DCCT, and as a result, there was some reluctance to use CSII. However, a number of trials have suggested that when the pump is properly used the rate of hypoglycemic episodes are no more frequent, or are even less common, compared to MDI [23,24,25]. The decrease in hypoglycemic events is accompanied by an increase in self-reported warning symptoms of hypoglycemia, as well as by an increase in counter regulatory hormonal responses to hypoglycemia. Because of the reported lower incidence of hypoglycemia with pump therapy, some experts are now recommending hypoglycemia unawareness as an indication for pump use [26]. Pump therapy has also been recommended for certain conditions and for special subsets of the diabetic populations. CSII can help improve blood glucose control in patients with the ‘dawn phenomenon’, in which blood glucose is elevated before breakfast [27]. Pump users who experience the dawn phenomenon can program their pump to increase night time basal insulin infusion to counter pre-breakfast blood glucose elevation. Experts also have suggested that the dosing flexibility of CSII may aid in reducing gastroparesis. In addition, erratic blood glucose control, and optimization of glycemic control during pregnancy [28] are additional indications by some medical community practice for CSII use. For those who chose CSII, improved lifestyle flexibility due to better control during exercise, eating out, or travel is cited as one of the most important reasons for choosing the pump. The ability to change dosing from moment to moment, and living with different dosing flexibility may be another reason for the increased pump usage observed recently. The evidence on quality-of-life is limited to testimonies from those patients who have had a positive experience with CSII.

Disadvantages of CSII Short acting insulin used with CSII has no depot, as opposed to the subcutaneously used longer acting insulin. In addition, subcutaneous insulin delivered via injection is cleared

CONTINUOUS GLUCOSE MONITORING DEVICES

225

at a much slower rate compared to insulin delivered via CSII. Therefore, any interruption of insulin flow from CSII can result in rapid hyperglycemia and DKA. In addition, even though hypoglycemia is less common with CSII use, the proper use of an insulin pump requires a highly motivated patient who is able to monitor blood glucose frequently, operate the device, and work with a diabetes team. Infrequent blood glucose monitoring, poor cooperation with the patent’s treatment team, and inability to program the device properly can result in severe adverse events including hypoglycemia, hyperglycemia or DKA. Device-related safety issues may involve clogging of catheters, bleeding at the site, hematomas, and catheter-related infections at the insertion site. Catheter-related infection is the most commonly reported complication to the FDA through Medical Device Reports [29]. A number of manufacturing-related and device specific malfunctions may also occur. Some reports of water ingress resulting in over-perfusion, leakage of insulin due to cracks in the pump, and other malfunctions have also resulted in device failures and recall of the device [30]. For some patients, cost and having to wear a device may be a barrier to choosing the pump over MDI. In summary, CSII is as effective as or more effective than MDI in achieving and maintaining tight blood glucose levels. The key to successful use of the device depends on the appropriate selection of diabetics and pumps. A patient who is motivated to monitor daily blood glucose levels frequently, skilled at operating the device, and willing to work with a health team is required.

Continuous glucose monitoring devices The development of devices that can monitor glucose frequently or continuously through out the day and night presents the potential for improved management of diabetic patients, with consequent reductions in morbidity and mortality associated with diabetes. One of the advantages of these kinds of devices over self-monitoring blood glucose devices (SMBG) is that frequent measurements allow for a more precise understanding of daily glucose fluctuations. In addition, when approved as a stand alone device, continuous glucose sensors devices could eventually be a component of the mechanical artificial pancreas and eliminate the dreaded pain and fear of frequent finger lancing. For example, it has been demonstrated with one of these devices (MiniMed Continuous Glucose Monitoring System) that there are often substantial postprandial increases in plasma glucose levels, even in patients with satisfactory HBA1c levels who use pre-meal doses of rapid acting insulin analogs [31,32,33,34]. This device has also revealed frequent episodes of asymptomatic hypoglycemia (<60 mg/dL), especially at night [34,35], although these readings must be interpreted carefully because the system is less accurate at low glucose levels and is considered generally accurate only in the 40–400 mg/dL range [35,36]. As of the fall of 2004, two such devices had been approved by the FDA. One of these is the MiniMed Continuous Glucose Monitoring System (CGMS). This device was approved by the FDA for 3-day periods of ‘‘Holter-style’’ tracking in 1999 [37], as an

226

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

adjunct to routine fingerstick blood glucose monitoring to detect trends and track patterns in glucose level. Worn in the subcutaneous tissue of the abdomen, it measures glucose concentration in interstitial fluid, every five minutes, for up to 72 hours. The patient performs finger-stick glucose determinations 3 to 4 times daily and enters the data into the device for calibration. A computer calculates the glucose levels, which are then later available for retrospective patient (the Guardian RT, which is a modification of the original CGMS, has been recently approved for real-time display of results) analysis. The other approved device is the GlucoWatch biographer, approved by the FDA in 2001 [38]. This device, which is worn around the wrist like a watch, non-invasively measures glucose concentration in interstitial fluid by reverse ionotophoresis. Following a 3-hour equilibration period and calibration of the device by entering the glucose results from a traditional finger stick, the device can provide glucose measurements up to three times an hour for a 12 hour period as an adjunct to routine fingerstick blood glucose measurement, to detect trends and track patterns [39]. It also has an alarm function that can be individually programmed and is pre-set to detect low and high (the alarm levels are determined by the patient and his/her healthcare provider and can be set at any level) glucose levels, and rapid rates of decline. (The MiniMed device also now has hypo- and hyperglycemic alarm function)

Measures of accuracy Several studies that have looked at correlations between the glucose monitoring devices and corresponding venous or capillary measurements have found reasonably strong correlation coefficients.{ For the MiniMed CGMS these have ranged between 0.73 and 0.92 [37,38,39] with an average absolute differences between the two methods ranging between 12% and 19% [41,42,43,44]. One of these studies, which examined these correlations in actual clinical practice, found a correlation coefficient of 0.91 and a mean absolute difference of 18% [44]. If two point calibration was used instead of one point, to calibrate the system, the correlation coefficient was 0.95 and the mean absolute difference 8.2% [41]. The GlucoWatch Biographer has demonstrated similar degrees of correlation with blood or capillary glucose measurements. These studies have found correlation coefficients ranging between 0.85 and 0.90 in a clinical environment [45,46,47], and between 0.80 and 0.85 in a home environment [45,47], with an average absolute difference between the two methods of about 19.1 mg/dL [45]. The Clarke Error Grid demonstrates the clinical significance of the differences seen when comparing the true glucose value to the comparative device result. It identifies five { Note: Because manufacturers do not always publish the results of studies used to support their premarket application for marketing approval, the information available in the published literature does not always cover all information on the performance of a given device. Therefore, the information presented here for glucose monitoring devices may differ somewhat from the information on which these devices were approved.

CONTINUOUS GLUCOSE MONITORING DEVICES

227

zones of clinical significance, which are defined according to the presence and severity of treatment error which would result from an inaccurate reading, as compared with standard glucose measurement. Three studies [48,49,50] evaluating the accuracy of the MiniMed CGMS reported that 93% to 99% of measurements fell within zones A or B (which indicate an accurate enough reading to avoid a treatment error) of the Clarke Error Grid. Correlations have also been calculated between the MiniMed CGMS and HBA1c levels, with correlation coefficients of 0.53 [51] and 0.59 [52] reported. It makes sense that the correlations would be considerably stronger with blood glucose than with HBA1c, since the latter represents chronic blood glucose levels, whereas the glucose monitoring devices provide acute measurements.

Control of diabetes A major test of the usefulness, or clinical and public health impact of any diagnostic or therapeutic modality used in DM, is its ability for long term blood glucose control at desirable levels, as manifested by HBA1c levels. Some studies have shown statistically significant declines in HBA1c over periods ranging from 5 weeks to 6 months, using the MiniMed CGMS, but they have not used controls for the purpose of comparing MiniMed with conventional means of glucose monitoring [31,43,51,53]. Another study used a control group (which was monitored with at least 8 capillary measurements of glucose per day for three days) and showed a decline in HBA1c from 8.3% to 7.5% using the MiniMed CGMS (worn for three days on 3–5 separate occasions over a 2–3 month period) for tracking, compared to a decline from 8.0% to 7.5% in the control group, but the difference between the MiniMed and the control group was not statistically significant [54]. One study, which used a cross-over design, did find a statistically greater decline in the MiniMed group than in the control group [55]. This study demonstrated a decline in HBA1c from 7.70% to 7.31% over a six month period with the use of MiniMed CGMS (worn for three days every two weeks) for tracking, compared to a decline of 7.75% to 7.65% in the control group (difference between the two groups: p ¼ .01). The study consisted of 27 patients, who were randomized into the two groups and switched over at three months. Both groups were instructed to complete at least two self-monitoring blood glucose measures at different times each day, and a 7-point self-monitoring once a week. The use of GlucoWatch was found to have a similar effect on HBA1c, although only one study was found that assessed this issue: This study showed a decline in median HbA1c in 20 children randomized to a group that utilized GlucoWatch Biographer (4 times per week) plus conventional glucose monitoring (4 times per day) over a 3 month period from 8.9% to 8.4%, compared to a rise in median HbA1c among the children who were randomized to the control group (who received only the conventional glucose monitoring) from 8.6% to 9.0% over the same time period. Even though the GlucoWatch group began at a higher baseline value, their median value of HbA1c was statistically lower than the control group after three months (p < .05) [56].

228

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

Control of hypoglycemia Prevention of severe hypoglycemic episodes is another potential benefit of continuous glucose monitoring devices. One potential step towards prevention of these episodes is better identification of their occurrence. By enabling much more frequent measurement of glucose – especially at night – the continuous glucose monitoring devices could facilitate improved prevention of hypoglycemia. Several studies have demonstrated the ability of the MiniMed CGMS to detect frequent episodes of hypoglycemia that were asymptomatic and not detectable by conventional means [36,54,57]. One of these studies detected previously unrecognized hypoglycemia in 62.5% of type 1 and 46.6% of type 2 diabetes, mostly at night [57]. Similarly, one study demonstrated that GlucoWatch detected about 60% more cases of hypoglycemia than a control group which utilized conventional diabetes management [56]. Theoretically, by demonstrating previously undetected episodes of hypoglycemia, treatment can be tailored to avoid these occurrences, which would be expected to decrease the incidence of symptomatic hypoglycemia as well. However, no studies showing direct demonstration of this benefit for either device were found. Under current conditions, the continuous glucose monitoring devices are likely to be both costly and inconvenient: they require several finger sticks per day for calibration, medical professionals are required to interpret the computerized data, and approved for detecting trends and tracking patterns in glucose measurements, not as a replacement for routine fingerstick blood glucose monitoring. The current assessment provides some encouraging evidence for a potential positive public health impact of the continuous glucose monitoring systems, when used as an adjunct to conventional glucose monitoring. However, more studies are needed that use appropriate control groups to assess the ability to chronically control blood glucose while using these devices for blood sugar tracking, before more definitive statements can be made about their public health impact.

Future medical devices for the management of diabetes Implantable insulin pumps The long-term implantable insulin pump (LIIP) is a high risk, Class III device. It is currently limited to investigational use in the United States. This device has received approval for marketing in the European Community. The LIIP is implanted into the anterior abdominal cavity and a catheter implanted through the abdomen into the peritoneal wall. Like the external pump, the device is comprised of a pumping mechanism, drug reservoir, and a programmable microprocessor. As opposed to the external infusion pumps, the system requires the use of an external communicator which allows for remote setting of basal and bolus dosages.

FUTURE MEDICAL DEVICES FOR THE MANAGEMENT OF DIABETES

229

The LIIP differs from the external pump in that insulin is delivered directly into the peritoneal cavity resulting in more physiologic insulin delivery. Intraperitoneal insulin delivery results in lower peripheral insulin levels due to faster clearance when compared to that of subcutaneous delivery. This rapid clearance and lower peripheral insulin levels may have more metabolic benefit compared to subcutaneous (SC) insulin delivery.

Blood glucose control Both intraperitoneal and subcutaneous insulin delivery achieve and maintain similar levels of glycemic control over time [58]. Studies measuring mean HbA1c and mean blood glucose levels in both insulin dependent and non-insulin dependent diabetics showed that mean glucose levels are similar to that seen in the DCCT intensive group [59,60]. In addition, data from long-term studies have suggested that the route of insulin administration does not significantly affect HbA1c levels [61]. There are some studies that suggest that intraperitoneal insulin delivery may result in less hypoglycemic episodes than subcutaneous infusion [62,63]. The exact mechanism for this phenomenon is not known, however, more rapid clearance of peritoneal insulin delivery is suspected as one possible explanation. The rapid absorption of intraperitoneal insulin is thought to simulate physiologic insulin. Some short term studies have also suggested that the simulation of physiologic insulin delivery may also achieve better lipid control [64].

Blood glucose excursions Achieving near normal blood glucose and improvement in glucose fluctuations are necessary to minimize the long-term complications of diabetes mellitus. More uniform absorption of intraperitoneal insulin delivery may result in less glycemic fluctuations. Studies comparing LIIP to that of subcutaneous insulin delivery showed less blood glucose fluctuation as measured by a decrease in the standard deviation of daily blood glucose levels [63].

Safety and Effectiveness Currently, both the LIIP and the specially formulated insulin planned for use in the device are limited to investigational use in the United States. The safety and effectiveness profile of the device will be established during the Agency’s review of future approval applications for the device and drug. There are some published data dating back to over 15 years supporting the potential safety and effectiveness of the device. As with the external infusion pump, catheter site infections and seromas were observed commonly [65]. During the earlier period of the device investigation, catheter occlusion and short catheter life span were encountered, resulting in insulin under delivery with the resultant

230

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

hyperglycemia and DKA. In response to these problems, the current catheters used with the LIIP include a side access port. Direct access through the side port aids in diagnosing blockage as well as relieving the blockage by flushing the catheter [61]. Another safety issue that will require careful observation is the potential for development of anti-insulin antibodies. Anti-insulin antibody development and prevalence rate of increase of antibodies levels are similar during the initial therapy period for both intraperitoneal and subcutaneous insulin administration [66]. With continuous LIIP use, a small subset of diabetics may exhibit marked immunogenicity by developing elevated levels of these antibodies [67]. The clinical significance of this phenomenon is not clearly understood. However, markedly elevated antibody levels may result in a syndrome of either increased daily insulin requirement, or nocturnal hypoglycemia, or both [67]. Epidemiologic studies to assess risk factors for this may be useful. As mentioned above, the LIIP and the specially concentrated insulin (U400) are currently limited to investigational use only in the US. The safety and effectiveness of this product will be clearly delineated after completion of the review by the agency. If device is approved, it could provide another long-term option (up to 8 years) for delivery of insulin to the diabetic population. In addition, when a functioning long-term sensor becomes available, the implanted long-term pump will provide a key component for a long-term artificial mechanical pancreas.

Closed loop systems: artificial mechanical pancreas In the DCCT, the rate of hypoglycemia was three times more in the treatment (MDI) group when compared to the control (conventional) group [6]. Intensive and conventional insulin management require frequent daily finger sticks to monitor blood glucose levels. Diabetes management via needle injections results in poor quality of life, poor compliance, and increased incidence with disease transmission and problems with needle disposal. Though the benefit of tight blood glucose control in limiting the complications of the disease is well established, only 37% of diabetics achieve the level of control recommended by the American Diabetic Association [68]. Therefore, a safe and effective device to manage DM that eliminates the need for finger pricking, needle injections and their disposal is of a paramount public health importance. The creation of a closed-loop system (artificial mechanical pancreas) has been the ‘‘holy grail’’ in the management of DM. The closed loop system will consist of a pump, a sensor, and a control system to integrate both. A functioning closed-loop system will eliminate the use of needles and finger pricking, and will result in tight blood glucose control, thereby greatly improving the quality of the diabetic population. An artificial mechanical pancreas will require three different components working in harmony: a glucose sensor, an insulin pump device and a computerized control system to integrate the function of both devices. There are two possible system designs: external or implanted. In the case of an external artificial pancreas, the pump, the sensor, and the control system would be worn externally. In an implanted closed-loop system, all components of the system would be surgically implanted.

REFERENCES

231

Current state in the development of the artificial pancreas device The sensor technology lags behind the other components of the closed loop systems. To date, there is no FDA-approved, stand-alone glucose sensor suitable for use as a component in a closed loop system. However, a number of external and implantable sensors are at various investigational stages. There are two FDA-approved external sensors. The Cygnus GlucoWatch [38] and the MiniMed/Medtronic Continuous Glucose Monitoring System (CGMS) approved in 1999 (see above for more detail) [37]. Both sensors are approved as adjuncts to supplement finger stick monitoring, not as stand-alone replacements. A computer-controlled algorithm system should be able to sample, filter and interpret the sensor data and then correctly provide feedback with the appropriate amount of insulin dose from the pump to achieve near normal glucose levels. Integration of pump and sensor through the control algorithm have been investigated and described in the literature and some have even undergone clinical trials [69,70]. The technology of sensors is advancing, and a stand alone, continuous blood glucose monitoring device may be on the horizon for short term (days) use. The closed-loop system technology will probably progress slowly from short-term use in selected populations to widespread use in the larger diabetic population. Currently, there are data that show that short-term tight blood glucose control results in improved morbidity and mortality outcomes in the critically ill patient [71]. A number of experts advocate clinical practice standards for acute tight blood glucose control in the hospitalized diabetic patient [72]. As the closed-loop system technology evolves, a selected diabetic population will benefit from the device. The impact of the device will be to significantly improve outcomes of the critically ill patient who require blood glucose regulation. Some examples of these populations may include diabetics with acute myocardial infarction, diabetics in intensive care unit, post-operative patients, and gestational diabetics. In addition to reducing morbidity and mortality, such a system should result in great cost reduction. Ultimately, in addition to the benefits to the patient, automatic blood glucose reading and insulin delivery will reduce the work load of the health care worker involved in diabetic management.

References 1. U.S. Centers for Disease Control and Prevention. Diabetes: Disabling, Deadly, and on the Rise. http://www.cdc.gov/nccdphp: Accessed April 20, 2005. 2. U.S. Centers for Disease Control and Prevention. National Diabetes fact sheet. http://www. cdc.gov/diabets/pubs/: Accessed April 20, 2005. 3. CDC Diabetes program: http://www.cdc.gov/diabetes/statistics/prev/nationa/figpersons.htm: Accessed April 20, 2005. 4. American Diabetes Association: Economic costs of diabetes in the U.S. in 2002. Diabetes Care. 2003; 26: 917–932. 5. DeWitt DE, Hirsch IB. Outpatient insulin therapy in type I and type 2 diabetes mellitus. JAMA 2003; 289: 2254–2264.

232

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

6. Diabetes Control and Complications Research Group: The effects of intensive treatment on the development and progression of long-term complications in insulin dependent diabetes mellitus. N Eng J Med 329: 977–986, 1993. 7. UKPDS study group. Intensive blood glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetics. Lancet. 1998; 352: 837–853. 8. American Diabetes Association: Standards of Medical Care in Diabetes. Diabetes Care 27: S15– 35, 200. 9. Stewart KM, Wilson MF, Rider JM. Insulin Delivery Devices. Journal of Pharmacy Practice 2004. 17; 1:20–28. 10. Plevin S, Sadur C. Use of a prefilled insulin syringe (Novolin Prefilled) by patients with diabetes. Clin Ther 1993; 15 (2): 423–31. 11. Bohannon, N.J. Insulin delivery using pen devices. Postgrad Med 1999; 106 (5): 57–68. 12. Weller, C, Linder, M. Jet Injection of Insulin vs the syringe-and-needle method. JAMA, 1966, 195(10): 844–7. 13. http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm 14. Harsh I, Rudolph JW: An assessment of continuous subcutaneous insulin infusion therapy in an academic clinic (abstract). Diabetes 49 (Suppl. 1): 100A, 1998. 15. Pickup J, Mattock M, and Kerry S: Glycaemic control with continuous subcutaneous insulin infusion compared with intensive insulin injection with type I diabetes: Meta-analysis on randomized controlled trials. BMJ 2002; 324: 1–6. 16. Cersosimo E, Jornsay D, Arce C, Rieff M, Feldman E, Defronzo R. Improved Clinical Outcomes With Intensive Insulin Pump Therapy In Type 1 Diabetes. Diabetes 2002; 51(Suppl 2):515. 17. Hissa MN, Hissa ASR, Bruin VMS, Fredrickson LP. Comparison between Continuous Subcutaneous Insulin Infusion and Multiple Insulin Injection Therapy in Type 1 Diabetes Mellitus: 18-Month Follow-Up. Endocr Pract 2002; 8: 411–416. 18. Mathur R, Sosa M, Somma L, Peters AL. Comparison of Glycemic Outcomes in Patients with Type 1 Diabetes (T1DM) on Continuous Subcutaneous Insulin Infusion (CSII) or Multiple Daily Injections (MDI) Therapy. Diabetes 2002; 51(Suppl.2): 434. 19. http://www.fda.gov/cder/foi/label/2000/21081lbl.pdf 20. Ratner RE, Hirsh IB, Neifing JL, Garg SK, Mecca TE, Wilson CA: Less hypoglycemia with insulin glargine in intensive insulin therapy for type 1 diabetes: US Study Group of Insulin Glargine in Type 1 Diabetes. Diabetes Care. 2000; 23: 639–643. 21. Doyle (Boland) EA, Weinzimer SA, Steffen AT and et al: A randomized, Prospective Trail Comparing the Efficacy of Continuous Insulin with Multiple Daily Injections Using Insulin Glargine. Diabetes Care: 2004; 27: 155–8. 22. Locke DR, Rigg LA: Hypoglycemia coma associated with subcutaneous insulin infusion by portable pump. Diabetes Care 1981; 4: 389–391. 23. Bode BW, Steed RD, Davidson PC: Reduction in severe hypoglycemia with long-term continuous subcutaneous insulin infusion in type 1 diabetes. Diabetes care 1996; 19: 324–327. 24. Boland EA, Grey M, Oesterle A, and et al: Continuous subcutaneous insulin infusion. A new way to lower risk of severe hypoglycemia, improve metabolic control and enhance coping in adolescents with type 1 diabetes: Diabetes Care 1999; 22: 1784–1799. 25. Eichner HL, Selam JL, Holleman CB, Worcester BR, Turner DS, Charles MA. Reduction Of Severe Hypoglycemic Events In Type I (Insulin Dependent) Diabetic Patients Using Continuous Subcutaneous Insulin Infusion. Diabetes Res 1988; 2: 189–193. 26. Hirsch IB, Farkas-Hirsch, Cryer PE. Continuous Subcutaneous Insulin Infusion for the Treatment of Diabetic Patients with Hypoglycemic Unawareness. Diabetes, Nutrition & Metabolism 1991; 4: 41–43.

REFERENCES

233

27. Perriello G, Defeo P, Torlone E, Fanelli C, Santeusanio F, Brunetti P, Bolli GB. The Dawn Phenomenon in Type I (Insulin-Dependent) Diabetes Mellitus. Magnitude, Frequency, Variability, and Dependency on Glucose Counter regulation and Insulin Sensitivity. Diabetologia 1991; 34: 21–28. 28. Yogev Y, Chen R, Ben-Haroush A, Phillip M, Jovanovic L, M Hod. Continuous Glucose Monitoring for the Evaluation of Gravid Women with Type 1 Diabetes Mellitus. Obstetrics & Gynecology 2003; 4: 633–638. 29. http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfMAUDE/search.CFM: Accessed April 10, 2005. 30. http://www.fda.gov/cdrh/recalls/: Accessed April 10, 2005. 31. Bode BW, Gross TM, Thorton KR, Mastrototaro JJ. Continuous glucose monitoring used to adjust diabetes therapy improves glycosylated hemoglobin: a pilot study. Diabetes Res Clin Pract 1999, 46:183–190. 32. Chase PH, Kim LM, Owen SL, MacKenzie TA, Klingensmith GJ, Murtfeldt R, Garg SK. Continuous subcutaneous glucose monitoring in children with type 1 diabetes. Pediatrics 2001, 107: 222–226. 33. Kaufman FR. Role of continuous glucose monitoring in pediatric patients. Diabetes Technol Ther 2000, 2 (Suppl. 1): S49–S52. 34. Boland E, Monsod T, Delucia M, Brandt CA, Fernando S, Tamborlane WV. Limitations of conventional methods of self-monitoring of blood glucose. Diabetes Care 2001, 24:1858–1862. 35. Monsod TP, Flanagan DE, Rife F, Saenz R, Caprio S, Sherwin RS, Tamborlane WV. Do sensor glucose levels accurately predict plasma glucose concentrations during hypoglycemia and hyperinsulinemia? Diabetes Care 2002, 25: 889–893. 36. Mastrototaro JJ. The MiniMed continuous Glucose monitoring system. Diabetes Technology & Therapeutics 2000, 2 (Suppl. 1): S13–S18. 37. Summary of Safety and Effectiveness Data for the MiniMed Continuous Glucose Monitoring System (CGMS). Food and Drug Administration, 1999 (publ. no. PMA P980022). http:// www.fda.gov/cdrh/mda/docs/p990026S008.pdf (accessed 4/20/05) 38. http://www.fda.gov/cdrh/mda/docs/p990026S008.pdf (accessed 4/20/05) 39. ECRI. GlucoWatch Biographer: An automatic, noninvasive glucose monitor. Health Devices 31(8): 295–302, 2002. 40. Metzger M, Leibowitz G, Wainstein J, Glaser B, Raz I. Reproducibility of glucose measurements using the glucose sensor. Diabetes Care 2002, 25(6): 1185–1191. 41. Steil GM, Rebrin K, Mastrototaro J, Bernaba B, Saad MF. Determination of plasma glucose excursions with a subcutaneous glucose sensor. Diabetes Technology & Therapeutics 2003, 5(1): 27–31. 42. Gross TM, Ter Veer A. Continuous glucose monitoring in previously unstudied population subgroups. Diabetes Technology & Therapeutics 2000, 2(Suppl. 1): S27–S34. 43. Gross TM, Mastrototaro JJ. Efficacy and reliability of the continuous Glucose monitoring system. Diab Technology & Therapeutics 2000, 2(Suppl. 1): S19–S26. 44. Gross TM, Bode BW, Einhorn D, Kayne DM, Reed C, White NH, Mastrototaro JJ. Performance evaluation of the MiniMed Continuous Glucose Monitoring System during patient home use. Diabetes Technology & Therapeutics 2000, 2:49–56. 45. Tierney MJ, Tamada JA, Potts RO, Jovanovic L, Garg S, Cygnus Research Team. Clinical evaluation of the GlucoWatch biographer: a continual, non-invasive glucose monitor for patients with diabetes. Biosensors and Bioelectronics 16: 621–629, 2001. 46. Tamada JA, Garg S, Jovanovic L, Pitzer KR, Fermi S, Potts RO, the Cygnus Research Team. Noninvasive Glucose Monitoring: Comprehensive Clinical Results. JAMA 282:1839–1844, 1999.

234

CH16

SELECTED MEDICAL DEVICES USED TO MANAGE DIABETES MELLITUS

47. Garg SK, Potts RO, Ackerman NR, Fermi SJ, Tamada JA, Chase PH. Correlation of fingerstick blood glucose measurements with GlucoWatch Biographer glucose results in young subjects with type I diabetes. Diabetes Care 22: 1708–1714, 1999. 48. Guerci B, Floriot M, Bohme P, Durain D, Benichou M, Jellimann S, Drouin P. Clinical performance of CGMS in type 1 diabetic patients treated by continuous subcutaneous insulin infusion using insulin analogs. Diabetes Care 2003, 26 (3): 582–589. 49. Sachedina N, Pickup JC. Performance assessment of the Medtronic-MiniMed Continuous Glucose Monitoring System and its use for measurement of glycemic control in type 1 diabetic subjects. Diabetic Medicine 2003, 20: 1012–1015. 50. Djakoure-Platonoff C, Radermecker R, Reach G, Slama G, Selam JL. Accuracy of the continuous glucose monitoring system in inpatient and outpatient conditions. Diabetes & Metabolism 2003, 29: 159–162. 51. Salardi S, Zucchini S, Santoni R, Ragni L, Gualandi S, Cicognani A, Cacciari E. The glucose area under the profiles obtained with Continuous Glucose Monitoring System relationships with HBA1c in pediatric type 1 diabetic patients. Diabetes Care 2002, 25: 1840–1844. 52. Sharp P, Rainbow S. Continuous glucose monitoring and haemoglobin A1c. Ann Clin Biochem 2002, 39: 516–517. 53. Kaufman FR, Gibson LC, Halvorson M, Carpenter S, Fisher LK, Pitukcheewanont P. A pilot study of the Continuous Glucose Monitoring System – Clinical decisions and glycemic control after its use in pediatric type 1 diabetic subjects. Diabetes Care 2001, 24: 2030–2034. 54. Chico A, Vidal-Rios P, Subira M, Novials A. The continuous glucose monitoring system is useful for detecting unrecognized hypoglycemias in patients with type 1 and type 2 diabetes but is not better than frequent capillary glucose measurements for improving metabolic control. Diabetes Care 2003, 26: 1153–1157. 55. Ludvigsson J, Hanas R. Continuous subcutaneous glucose monitoring improved metabolic control in pediatric patients with type 1 diabetes: A controlled crossover study. Pediatrics 2003, 111: 933–938. 56. Chase HP, Roberts MD, Wightman C, Klingensmith G, Garg SK, Van Wyhe M, Desai S, Harper W, Lopatin M, Bartkowiak M, Tamada J, Eastman R. Use of the GlucoWatch biographer in children with type 1 diabetes. Pediatrics 111: 790–794, 2003. 57. Cheyne E, Kerr D. Making ‘sense’ of diabetes: using a continuous glucose sensor in clinical practice. Diabetes/Metabolism Research and Reviews 2002, 18(Suppl. 1): S43–S48. 58. Gin H, Renard E, Melki V, Boivin S, Schaepelynck-Belicar P, Guerci B, Selam JL, Brun JM, Riveline JP, Estour B, Catargi B; EVADIAC Study Group: Combined improvements in implantable pump technology and insulin stability allow safe and effective long term intraperitoneal insulin delivery in type 1 diabetic patients: the EVADIAC experience. Diabetes Metab. 2003 Dec; 29(6):602-7. 59. Nathan DM, Dunn FL, Burch J, McKitrick C, Larkin M, Haggan C, Lavin-Tompkins J, Norman D, Rogers D, Simon D: Post prandial insulin profiles with implantable pump therapy may explain decreased frequency of severe hypoglycemia, compared with intensive subcutaneous regimens, in insulin dependent diabetes mellitus patients. AM J Med 100: 412–417, 1996. 60. Dunn FL, Nathan DM, Scavini M, Selam JL, Wingrove TG.: Long-term therapy of IDDM with an implantable insulin pump. The Implantable Insulin Pump Trial Study Group. Diabetes Care. 1997 Jan; 20(1): 59–63. 61. Charles MA. Programmable implantable insulin infusion devices and diabetes care. Diabetes Technol Ther. 1999; 1(1): 89–96. 62. Dunn FL, Nathan DM, Scavini M, Selam JL, Wingrove TG.: Long-term therapy of IDDM with an implantable insulin pump. The Implantable Insulin Pump Trial Study Group. Diabetes Care. 1997 Jan; 20(1): 59–63.

REFERENCES

235

63. Gin H, Renard E, Melki V, Boivin S, Schaepelynck-Belicar P, Guerci B, Selam JL, Brun JM, Riveline JP, Estour B, Catargi B; EVADIAC Study Group.: Combined improvements in implantable pump technology and insulin stability allow safe and effective long term intraperitoneal insulin delivery in type 1 diabetic patients: the EVADIAC experience. Diabetes Metab. 2003 Dec; 29(6): 602–7. 64. Broussolle C, Jeandidier N, Hanaire-Broutin H, et al: French multicenter experience of implantable pump. Lancet 1994; 343: 514–515. 65. Renard E, Rostane T, carrier C, Marchandin H, et al: Implantable insulin pumps: infections most likely due to seeding from skin flora determine severe outcomes of pump-pocket seromas. Diabetes Metab. 2001 Feb; 27(1): 62–5. 66. Olsen CL, Chan E, Turner DS, et al: Insulin antibodies response after a long-term intraperitoneal insulin administration via implantable programmable insulin delivery system. Diabetes care 1994; 17:169–176. 67. Jeandidier N, Boivin S, Sapin R, Rosart-Ortega F, Uring-Lambert B, Reville P, Pinget M.: Immunogenicity of intraperitoneal insulin infusion using programmable implantable devices. Diabetologia. 1995 May; 38(5): 577–84. 68. Saydah S, Fradkin J, Cowie C. Poor control of risk factors for vascular disease among adults with previously diagnosed diabetes. JAMA. 2004; 291: 335–342. 69. E. Renald. Implantable closed-loop glucose sensing and insulin delivery: the future for insulin pump therapy. Curr. Opin. Pharmacol. 2 (2002) 708–716. 70. E. Renard, R. Shah. M. Miller, et al. Accuracy of real-time blood glucose measurement by longterm sensor allows automated insulin delivery in diabetic patients. Diabetes 51 (suppl. 2) (2002) 508-p (A126). 71. Van Den Berghe G, Wouters P. Weeker F, et al. Intensive insulin in critically ill patients. NEJM. 2000; 345: (19): 1359–1367. 72. American College of Endocrinology. Consensus Development Conference on Inpatient Diabetes and Metabolic Control. http://www.aace.com/pub/ICC/inpatientStatement.php (accessed 5/10/05).

17 Medical device-related outbreaks S. Lori Brown US Food and Drug Administration, Bothell, WA, USA

Hesha J. Duggirala and Dale R. Tavris US Food and Drug Administration, Rockville, MD, USA

Introduction An outbreak is the occurrence of more cases of disease than expected in a given area or among a specific population over a specific time period [1]. Epidemiologic investigations of outbreaks are an important public health measure [2]. Most scientists are familiar with the concept of the classic foodborne bacterial [3], viral [4], or prion-related [5] outbreaks in which contaminated food transmits disease, or the spectacular outbreaks of gruesome, lethal diseases, such as Ebola hemorrhagic fever [6]. Outbreaks are recognized when illness is noticed to be above the usual background level. An often cited example these days is smallpox. Since the eradication of smallpox in 1979, a single case of smallpox would be considered an outbreak – in fact, a public health emergency [1]. In practice, it is more difficult to recognize a disease outbreak when there is a constant background level of the disease in question. This is the case with influenza. As with foodborne or other outbreaks, confirming that there is a medical device-related outbreak in progress is the first step of the investigation. In some cases, the outbreak may not be noticed until it has ended: recognizing a medical device-related outbreak may take months or even years. As with outbreaks in general, it is often the keen observation of physicians, hospital epidemiologists, clinical laboratory personnel, nurses, or risk managers who notice unusual disease activity that demands closer scrutiny. Medical

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

238

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

device-related outbreaks could also be detected by surveillance if sentinel cases or rates above background levels are recognized and investigated. Defining the source of the outbreak is crucial for implementing public health measures to arrest the progress of a current outbreak. In some ongoing outbreaks, this may be achieved by removing contaminated devices from use or, when appropriate, by recalling the device in question. Designing plans and procedures to prevent future outbreaks based on information collected during an investigation is another important action. The causes of medical device-related outbreaks may be microbial, chemical, or other (Table 17.1). The device often plays the role of a ‘fomite’, that is, an inanimate object that may harbor pathogenic organisms and serve as an agent of transmission. As with foods, while some medical devices may act simply as a vehicle for transmission of microbial agents, some device materials promote microbial growth, adherence, or biofilm formation. When a medical device-related outbreak is suspected, all causes should be considered, since they are not limited to contamination with microbial agents but may also occur because of chemical or toxin contamination of the device or, in some cases, as the result of manufacturing problems. Examples of each of these will be discussed and we will describe the broad scope of products involved as well as the breadth of organisms and other causes of outbreaks. Medical devices from the simplest, such as tongue depressors or fingerstick devices, to the most sophisticated, such as fiberoptic bronchoscopes or ventilators, may be involved in outbreaks. We discuss medical device-related outbreaks in terms of multiple dimensions: medical device-related outbreaks due to a particular device type (endoscopes and bronchoscopes); outbreaks that occur during a particular common therapy (kidney dialysis); and outbreaks that occur in a particular healthcare setting (neonatal and pediatric intensive care units). We also discuss miscellaneous medical device-related outbreaks to emphasize the breadth of devices that may be involved, and some pseudooutbreaks, which are examples of colonization of devices and/or specimens. Pseudooutbreaks do not involve illness but are an important indicator of contamination Table 17.1 Possible causes of medical device-related outbreaks Cause Infectious agents Contamination during manufacturing or shipping Contamination during use Inadequate cleaning Inadequate sterilization Chemical agents Residues from sterilization process Residues from manufacturing process Breakdown due to environmental conditions or aging Inadequate rinsing after cleaning Other contributing factors Device design

ENDOSCOPES: BRONCHOSCOPES AND GASTROINTESTINAL

239

that might end in illness. It must be recognized that this is just a sampling of medical device-related outbreaks. Some devices are notorious for outbreaks of nosocomial infections so numerous that they alone could fill a volume.

Endoscopes: bronchoscopes and gastrointestinal (GI) endoscopes Bronchoscopes Bronchoscopes are long, rigid or flexible, narrow tubes with a light on the end that may be inserted via the throat to view the trachea and bronchi. This fiberoptic instrument allows direct observation of the airways with video images. The bronchoscope is used in the performance of biopsy, pulmonary lavage and suction, and other procedures. There are a large variety of bronchoscope models, attachments, and accessories from different manufacturers cleared by the FDA. Bronchoscopy procedures were reportedly performed as ambulatory or inpatient procedures in the USA nearly 500 000 times in 1996 [8]. This figure likely represents a minimum number of bronchoscopies performed per year. Numerous bronchoscope-related outbreaks are reported in the medical literature. The following is not an exhaustive review of all bronchoscope outbreaks but, rather, a sampling of published reports. Table 17.2 provides an overall summary of these selected outbreaks. In the late summer of 1999, five bronchial washing cultures were noted to be positive for Mycobacterium tuberculosis in a Wisconsin hospital that had only seen eight M. tuberculosis-positive cultures in the previous 4 years [9]. During the period in question, 19 bronchoscopic procedures were performed, using three different bronchoscopes. Restriction fragment length polymorphism (RFLP) analysis of isolated M. tuberculosis of nine patients after the identified index case revealed an RFLP pattern identical to the index case. In addition, one of the cases also had a subsequent sputum sample with a different RFLP pattern and was suspected to have been previously infected. Three of the culture-positive patients developed evidence of M. tuberculosis infection. A nurse who had assisted with the bronchoscopy in the index patient and the patient with the different RFLP pattern converted from a negative to positive skin test within a few months. Based on the bronchoscopy log, the same device was used in nine of the 10 affected patients (including the index case). This suggested that one case may have occurred due to laboratory cross-contamination of the sample from the one patient who reportedly had a different bronchoscope during his/her procedure. Another possible explanation for this was that this patient was exposed to the same bronchoscope as the others but that it was recorded incorrectly in the log. The three bronchoscopes were returned to the manufacturer for examination. Leak testing by the manufacturer indicated that the suspect bronchoscope had a small leak. The standard operating procedure for reprocessing bronchoscopes at this hospital included cleaning the lumen and port with an enzymatic detergent, leak testing, and a

9

8

22

8

—

Bronchoscope

Bronchoscope and endoscope

Bronchoscope

Bronchoscope

Number of cases

Bronchoscope

Device

M. tuberculosis

Pseudomonas aeruginosa

M. chelonae, Methylbacterium mesophilicum and others

M. tuberculosis (drug-resistant)

Mycobacterium tuberculosis

Agent

High frequency of positive cultures from bronchoalveolar lavage Positive bronchoalveolar lavage fluids from eight consecutive patients Cross-contamination of bronchoscopy specimens

Lethal TB and contamination of samples

TB or pseudoinfection

Adverse event

Faulty wash/disinfect switch on the automatic washer

A hole in the sheathof the bronchoscope caused incomplete cleaning and disinfection; direct delivery of M. tuberculosis from the index case into distal airways of subsequent cases Inadequate cleaning and disinfection of the bronchoscope led to transmission of drugresistant M. tuberculosis and contamination of patient samples leading to false positives Automated endoscope washers were contaminated with a biofilm that contaminated endoscopes and bronchoscopes while they were being washed Suction channel of two fiberoptic bronchoscopes were contaminated due to lapse in routine cleaning procedures

Explanation

Table 17.2 Endoscope-related outbreaks: bronchoscopes and GI endoscopes

(Continued)

Bryce et al., 1993 [13]

Kolmos et al., 1994 [12]

Kressel and Kidd, 2001

Agerton et al., 1997 [10]

Ramsey et al., 2002 [9]

Reference

16

39

20

10

Bronchoscope

Bronchoscope

GI endoscope

Number of cases

Bronchoscope

Device

Gluteraldehyde disinfectant

P. aeruginosa, Seratia marcescens

M. avium, M. tuberculosis, M. chelonae, M. gordonae, acid- fast bacilli P. aeruginosa and others

Agent

Bloody diarrhea, abdominal cramps, nausea, vomiting

Positive cultures, pneumonia

Respiratory or blood stream infections; three deaths

Positive specimens

Adverse event

Table 17.2 (Continued)

Toxicity from residual gluteraldehyde in endoscope

Loose biopsy-port cap in the instrument surmised to have sheltered organisms, rendering cleaning and disinfection procedures ineffective Loose biopsy-port cap allowed for formation of biofilm and ineffective cleaning and disinfection

Disinfecting implicated water tank to cleaning machine controlled the pseudoepidemic

Explanation

Durante et al., 1992 [29]

Kirschke et al., 2003 [18]

Srinvasan et al., 2003 [19]

Gubler et al., 1992 [15]

Reference

242

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

20–30 minute soaking in 7.5% hydrogen peroxide with 0.8% phosphoric acid disinfectant. However, the investigation indicated that leak testing was not being performed routinely as required. Also, there was no timer used to measure the duration of soaking the bronchoscopes in disinfectant. The investigators concluded that the bronchoscope could not be properly cleaned and disinfected because of the small hole found in the sheath during leak testing. M. tuberculosis sequestered in this hole evaded cleaning and disinfection. This led to direct transmission of M. tuberculosis from the index patient into the airways of other patients undergoing bronchoscopy. Prompt recognition of the outbreak and treatment with appropriate antibiotics could have prevented additional cases of TB from developing, underlining the importance of rapid recognition of medical device-related outbreaks. Investigation of a community outbreak of a multi-drug-resistant TB in South Carolina led to the conclusion that one immunocompromised patient was infected during bronchoscopy at a hospital following a bronchoscopy procedure on the index case from the community [10]. The immunocompromised patient died of multi-drug-resistant TB. Two additional cases of false-positive cultures had been obtained from patients whose specimens were contaminated by the tainted bronchoscope. In this case, lax cleaning procedures were identified as the reason for the contamination. Omission of vigorous mechanical cleaning, absence of leak-testing, inadequate disinfection, inappropriate drying procedures, and improper storage were all observed, despite written policies. In a Cincinnati hospital, an investigation was undertaken when the laboratory reported an unusual number of Mycobacterium chelonae and Methylobacterium mesosphilicum in fungal cultures grown from bronchoscopy procedures [11]. This coincided with the introduction of an automatic endoscope reprocessing (AER) system for bronchoscopes (AERs are also medical devices). The infection control department followed up with cultures of the environment and equipment. The authors reported that M. chelonae, pink bacteria, and Gram-negative rods were grown from used, but not unused, glutaraldehyde, and from the AER system. The AER units were sent to the Centers for Disease Control and Prevention (CDC) Biofilm Laboratory, where biofilm from tubing in the AER grew Methylobacterium. The original source of contamination could not be determined. Several studies conclude that lapses in following procedures for cleaning or disinfecting were the cause of an outbreak, whether due to lapses by personnel or to operating problems with the AER units [12–14] or the inadvertent use of contaminated water [15]. There have also been several AER recalls due to microbial contamination. In September 1999, the FDA and CDC jointly issued a Public Health Advisory after several incidents of infections in patients undergoing bronchoscopy [16]. The advisory noted that there were conflicting instructions for reprocessing (cleaning and disinfection) issued by endoscope manufacturers and AER system manufacturers, and that bronchoscopes were inadequately cleaned and disinfected and potentially damaged when inappropriate channel connecters were used with the AER – in other words, if it didn’t fit but was forced to fit [17]. The FDA and CDC Public Health Advisory counseled bronchoscope reprocessors to resolve any conflicting recommendation between the endoscope

ENDOSCOPES: BRONCHOSCOPES AND GASTROINTESTINAL

243

manufacturers’ and the AER manufacturers’ instructions, and to comply with the endoscope manufacturers’ instructions for cleaning and disinfecting the instrument. Several bronchoscope outbreaks were related to bronchoscopes that did not meet manufacturing specifications and resulted in a recall of 14 000 bronchoscopes worldwide. The manufacturer’s recall was reportedly initiated in response to cases reported to the manufacturer from a community hospital [18]. Johns Hopkins Hospital also observed an outbreak but was unaware of the bronchoscope recall [19–21]. Both outbreaks were reportedly due to a loose biopsy port that was not securely fastened and subsequently allowed microbial contamination and biofilm formation. In these two hospitals, the rate of isolation of Pseudomonas aeruginosa from bronchoscopy specimens was linked to the use of the recalled bronchoscopes, and the contaminated bronchoscopes were potentially linked to clinical outcomes – pneumonia and bloodstream infections. In both hospitals, cultures of the suspect bronchoscopes were positive for P. aeruginosa and in both hospitals the outbreak ended with the removal of the (recalled) bronchoscopes from service.

GI endoscopes Endoscopy of the small and large intestine, with or without biopsy, was reportedly performed nearly 1.5 million times in 1999 in US hospitals, according to hospital discharge data [22], and nearly 3.5 million times in 1996 ambulatory procedures [23]. Endoscopes used for gastrointestinal endoscopy have issues similar to those used for bronchoscopy. A study examining cleaning and disinfection at 22 hospitals and four ambulatory care centers indicated that high-level disinfection of gastrointestinal endoscopes was frequently not achieved by the methods employed [24]. Recent headlines have described notification of 177 endoscopy and colonoscopy patients in New York [25], 1331 endoscopy, colonoscopy, or sigmoidoscopy patients in California [26], 600 endoscopy patients in Washington [27], and 1300 patients in Northern Ireland [28] after recognizing lapses in procedures for cleaning or disinfecting endoscopes. Some patients were urged to undergo testing for hepatitis and HIV. One report of an endoscope-related outbreak of interest describes bloody diarrhea, abdominal cramps, nausea and vomiting in patients after endoscopy at a Maryland clinic [29]. No diarrheal pathogens could be identified from patients stool samples, neither was there growth from the port or tip of the endoscope. There was a significant association between development of diarrhea and one of the three clinical technicians charged with cleaning and disinfecting the sigmoidoscope. A health department investigation found 2 ml of fluid left in the endoscope after cleaning by this technician. Testing indicated that 2% gluteraldehyde, the disinfectant used, causes a necrotic, hemorrhagic colitis in rats [29]. This outbreak may have been related to residual gluteraldehyde left in the sigmoidoscope channel after cleaning and disinfecting by an inexperienced technician. In a review identifying 265 articles, 285 patient illnesses were attributed to GI endoscopy [30]. The authors concluded that the incidence of pathogen transmission by endoscopy was one per 1.8 million procedures. This is probably an underestimate of

244

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

pathogen transmission, since not all cases of illness will be reported in the medical literature, or even necessarily recognized.

Hemodialysis-related outbreaks In the USA, more than 340 000 patients are on dialysis or have a kidney transplant, with costs estimated at $17 billion per year [31]. The invasive nature of hemodialysis lends itself to outbreaks. Infection is the major cause of death in 15–30% of dialysis patients, with vascular access-related infections being a contributing cause [32]. There have been numerous disease outbreaks originating in hemodialysis centers, as described below (Table 17.3). In hemodialysis, a dialysis machine and a special filter called a dialyzer are used to reduce waste and fluids from the blood of patients with endstage renal disease. In 1992, six patients on hemodialysis acquired bloodstream infections with Klebsiella pneumoniae during an 11 day period at a hemodialysis unit in a tertiary care hospital [33]. Both a case-control study and a cohort study were conducted to identify risk factors for K. pneumoniae. The investigation narrowed the outbreak to a single shift with a patient with an arteriovenous fistula infected with K. pneumoniae. Investigators found that the dialyzer reprocessing technician failed to change gloves between patient contacts in the treatment area and reprocessing the infected patient’s dialyzers during that shift. The source of the outbreak was cross-contamination of dialyzers with K. pneumoniae and by inadequate dialyzer disinfection during reprocessing. Such an outbreak may have been prevented if gloves had been changed before and after patient care and before starting the dialyzer reprocessing procedure, and if the dialyzer compartments had been properly disinfected. A cohort study was carried out in Columbia, South America, to investigate 13 dialysis patients at one dialysis center who were found to be HIV-positive [34]. In 1993, testing of dialysis center patients confirmed a cluster of HIV seroconversions. A cohort study was conducted to determine risk factors for infection. The HIV seroconversion rate was found to be higher among patients dialyzed at the same time that a new HIV-positive patient was dialyzed. Transmission was probably due to poor infection control practices (most likely improperly processed access needles). The investigators suggested that Association for the Advancement of Medical Instrumentation (AAMI) and CDC infection control standards should be implemented to prevent transmission of infectious agents. In February of 1992 local public health authorities and the CDC were notified of pyrogenic reactions in 22 patients undergoing chronic hemodialysis at an outpatient dialysis center [35]. On a single day, patients in two shifts experienced intradialytic symptoms of chills, flushing, and/or fever. Patients in a third shift were given new unreprocessed hemodialyzers for their dialysis treatment and did not have symptoms or signs of a pyrogenic reaction. The investigation revealed that dialyzers reprocessed during the epidemic period contained elevated endotoxins in the blood compartment, and that the bioburden in water used to prepare dialysate was in excess of the AAMI

245

HEMODIALYSIS-RELATED OUTBREAKS

Table 17.3 Dialysis-related outbreaks Device

Number of cases Agent

Hemodialysis machine

6

Hemodialysis machine Hemodialysis machine

13

Hemodialysis machine

29

Hemodialysis machine

8

Hemodialysis machine

11

Hemodialysis machine

7

Hemodialysis machine

32

Hemodialysis machine

6

22

Klebsiella pneumoniae

Adverse event Bloodstream infections

Explanation

Dialyzer reprocessing technician failed to change gloves between patients contacts. Inadequate dialyzer disinfection during reprocessing HIV HIV Improperly processed seroconversion access needles Endotoxins Pyrogenic Elevated endotoxins in reactions blood compartment and bioburden in water in excess of AAMI standards Multiple Bloodstream Microbial overgrowth pathogens infections in waste handling option (WHO) drain ports Gram-negative Bloodstream Malfunctioning nonbacteria infections return valves of the WHO ports Gram-negative Bloodstream Intermittent bacteria infections contamination and inadequate disinfection of Orings in dialyzers Degradation Decrease vision Aged cellulose products and hearing, acetate membranes conjunctivitis, of dialyzers allowed headache and degradation products other neurologic to enter the symptoms bloodstream Hepatitis B Hepatitis B Handling of bloodvirus contaminated items infection without changing gloves or washing hands. Shared medications and supplies among infected and susceptible patients Hepatitis B Hepatitis B virus Supplies and multiinfection dose vials being shared

Reference Welbel et al. [33]

Velandia et al. [34] Rudnick et al. [35]

Arnow et al. [36]

Block et al. [37] Flaherty et al. [38]

Hutter et al. [39]

Morb Mortal Wkly Rep [40]

Hutin et al. [41]

246

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

standard for water. This outbreak emphasizes the need to use water that meets the AAMI bacteriologic and endotoxin standards for reprocessing hemodialyzers. An outbreak involving two chronic hemodialysis centers involved 29 patients in an 8 month period in 1996 [36]. Patients had bloodstream infections by 16 different pathogens. A case-control study revealed that cases were more likely to have received treatment on one highly contaminated dialysis machine. The investigation demonstrated that the outbreak of bloodstream infections resulted from microbial overgrowth in waste handling option (WHO) drain ports. The investigators emphasized that there is a need for staff education and well-designed equipment and practices involving the handling of dialysate waste. A similar outbreak to the one mentioned above occurred at a hemodialysis facility in Israel in 1997, in which eight cases of Gram-negative bloodstream infection occurred [37]. An investigation revealed the source of the infections to be the WHO systems in three of the 13 dialysis machines in use. The non-return valves of the WHO ports on the three implicated machines were shown to be malfunctioning. The investigators found that the manufacturer’s recommendations for daily testing and disinfection of the WHO system were not being followed, due to inadequacies in the instructions provided with the unit. The outbreak was halted following discontinuation of the WHO system. A university outpatient hemodialysis unit was the site of a 1988 outbreak of Gramnegative bacteria infection in 11 patients [38]. A case-control study and bacteriologic investigation of the hemodialysis unit revealed that the outbreak was caused by intermittent contamination and inadequate disinfection of O-rings in dialyzers. The investigators recommended that disinfection procedures should include measures to ensure adequate disinfections of O-rings. In 1996, seven dialysis patients at one hospital developed decreased vision and hearing, conjunctivitis, headache, and other neurologic symptoms shortly after hemodialysis [39]. A retrospective cohort study revealed that these adverse events were associated with exposure to aged cellulose acetate membranes of dialyzers, which allowed degradation products to enter the bloodstream. Prior to this incident, no expiration dating was required on dialyzers filters. As a result of this outbreak, FDA and CDC sent a letter to all US dialysis centers, advising rotation of dialyzers and elimination of old dialyzers. FDA now requests manufacturers to provide the date of manufacture on the labeling. The CDC investigated multi-state outbreaks of hepatitis B-infected hemodialysis patients in 1994 [40]. At a center in Texas, 14 patients were found to be infected with hepatitis B. An epidemiologic investigation indicated that staff frequently handled blood-contaminated items without changing gloves or washing hands. Two outbreaks in California involving seven cases of hepatitis B virus infection in at one center and 11 cases at another were both linked to shared medications and supplies among infected and susceptible patients. All of these outbreaks were associated with failure of the hemodialysis facilities to adhere to recommended infection-control practices for preventing the transmission of hepatitis B virus. These recommended practices emphasize the use of barriers and adherence to routine hand washing, appropriate disposal of needles and other sharp instruments, and disinfection and sterilization procedures.

NEONATAL AND PEDIATRIC INTENSIVE CARE UNITS

247

Another cluster of hepatitis B-infected hemodialysis patients was detected at two dialysis centers and one hospital in one county in late 1995 and early 1996 [41]. This outbreak involved six patients and was linked to supplies and multidose vials being shared routinely among patients. The outbreaks associated with hemodialysis centers emphasize the need to follow proper reprocessing and disinfection standards and guidelines. In addition, infection control standards should be implemented to prevent the transmission of infectious agents in this population.

Neonatal and pediatric intensive care units Device-related outbreaks are a major problem in neonatal intensive care units (NICU). The frequent use of medical devices, the invasive procedures, and the vulnerability of the patients make the NICU a place where attention to detail is required, and where medical device-related outbreaks frequently occur (see Figure 17.1, Table 17.4). Identifying the source(s) of an outbreak may be difficult. In some outbreaks, the source may not be identified even after a thorough investigation [42,43].

Figure 17.1 Premature infant in an incubator in the neonatal unit. Reproduced with permission from Science Museum/Science and Society Library

248

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

Table 17.4 Medical device-related outbreaks in pediatric and neonatal intensive care units Device Manual ventilation balloons

Number of cases 3

Ventilator 85 temperature probe

Enteral feeding

1

Breast pump

2

Tongue 4 depressors/ wooden spatulas

Agent

Adverse event

Bacillus cereus

Three neonates developed serious infections; one fatality

Explanation

Reference

Inadequate cleaning Van der and sterilization of Zwet interior of manual et al. [44] ventilation balloons; 35 additional neonates were colonized but without infection. Sterilization of the balloons ended the outbreak Sphingomonas Persistent 85 mechanically Lemaitre paucimobilis outbreak for ventilated neonates et al. [45] 20 month were infected. The period outbreak was terminated by introduction of sterilization procedures for ventilator temperature probes Klebsiella Fatal bacteremia Suspected nasogastric Berthelot oxytoca in newborn and tubes. Enforcement et al. [46] colonization of of universal 24 newborns’ precautions (gloving) digestive tracts ended outbreak Pseudomonas Twins became ill Environmental Brown et al., aeruginosa survey included 2005 [47] breast pump Rhizopus Cutaneous Cutaneous fungal Mitchell microsporus lesions at site infection at site of et al. [50] of splint use occlusive bandage with progression over intravenous to necrotizing cannulation site. cellulitis. Wooden tongue Three deaths depressors, used as splints, subsequently tested positive for Rhizopus microsporus

Ventilator or ventilator component-related outbreaks A cluster of serious cases of invasive infection with a genotypic strain of Bacillus cereus over a 3 month period, including a death, inspired an investigation of the NICU in an

NEONATAL AND PEDIATRIC INTENSIVE CARE UNITS

249

Amsterdam hospital [44]. The investigation included both a retrospective review of laboratory records and prospective surveillance of new neonates for colonization of umbilicus, armpits, anus, and tracheal aspirates. An intensive environmental survey and surveillance of NICU health care workers was undertaken. A case-control study to identify risk factors for the acquisition of B. cereus clearly identified exposure to a particular ventilation machine as a risk factor with an odds ratio of 9.8 (confidence interval 1.09–88.2). Prospective surveillance of newly admitted neonates also indicated that this was a potential source of colonization. Environmental sampling led to finding several contaminated balloons, also known as ambu-bags, that were used for manual ventilation of neonates prior to being connected to a mechanical ventilator. Review of cleaning procedures indicated that while the exterior of the balloons were regularly cleaned, the interior of the balloons were inadequately cleaned and sterilized, probably leaving viable B. cereus spores inside the balloons. The outbreak subsided after autoclaving manual ventilation balloons was introduced as part of the sterilization procedure. Another reported outbreak related to ventilators or ventilator components involved tracheal colonization of 85 neonates with Sphingomonas paucimobilis, an organism ultimately traced to a contaminated ventilator temperature probe [45]. An investigation was initiated after finding S. paucimobilis in a routinely cultured tracheal aspirate from a neonate. Subsequent prospective surveillance of neonates identified 85 colonized neonates over a 20 month period – in each case, the neonate had been ventilated. No cases of pneumonia or sepsis were observed, but concern was justified because of the potential for opportunistic infection in immunocompromised neonates. The cleaning and disinfection procedure for the temperature probes was found to be inadequate, possibly because of cracks in the aged probes. The practice of disinfecting the probe with an alcohol swipe was replaced with a 20 minute immersion in alcohol. The outbreak subsided after this change was instituted. Enteral feeding procedures are another potential source of infection in the neonatal unit. One report describes retrospective and prospective studies after the demise of a premature infant due to Klebsiella oxytoca bacteremia [46]. Prospective surveillance of stool and throat cultures from new neonates indicated that K. oxytoca was usually recovered from stool first and throat subsequently. Since premature infants were routinely fed with nasogastric tubes, this was considered a potential source of colonization. The investigation also indicated that nurses were compliant with hand washing prior to enteral feeding, but did not wear gloves during feeding. Meetings with nurses to emphasize the importance of wearing gloves indicated that nurses felt that gloving interfered with their relationship with the babies they cared for. After standard precautions were reintroduced, the outbreak subsided. Contamination during feeding may also occur from mother’s milk contaminated during breast pumping, either with the mother’s normal flora or with pathogenic contaminants. A surveillance report to the FDA described premature twins who became ‘ill’ [47]. Environmental sampling identified the mother’s breast pump as the source of infection for the infants. The use of breast pumps successively by multiple mothers may be particularly worrisome, since some breast pumps cannot be sanitized and disinfected easily [48,49].

250

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

An unusual outbreak was reported from Birmingham Women’s Hospital in the UK. Four cases of cutaneous mucormycosis infections, two fatal, in preterm neonates were due to Rhizopus microsporus [50]. In an attempt to find the source of these infections, environmental samples were cultured on Sabouraud’s medium. The source of the infection was found to be wooden tongue depressors that were used to construct splints for intravenous cannulation sites. Several wooden tongue depressors from an open box cultured positive. Samples from unopened boxes did not culture positive with the fungus. In a related pseudo-outbreak, Rhizopus sp. was isolated from stool, skin, and throat swabs during surveillance on neonatal and pediatric wards in another hospital [51]. Because the majority of positive samples were of gastrointestinal origin, possible contamination from enteral feeding, food or mouth care products were suspected. Environmental sampling indicated that the source of contamination was wooden spatulas (tongue depressors) used to transfer stool samples to specimen cups. In contrast to the case described previously, contaminated spatulas were found in several unopened boxes. The UK Medical Device Agency issued a Hazard Notice advising that the practice of using tongue depressors as splints be immediately discontinued [52].

Miscellaneous device-related outbreaks Re-use of hypodermic needles and syringes has been a common cause of medical devicerelated outbreaks. In fact, concern over transmission of HIV-AIDS via contaminated needles or syringes is largely responsible for the nearly universal use of single-use, disposable syringes in the USA. However, other devices that are used for mass inoculation may be responsible for outbreaks. CDC reported that during routine investigations of hepatitis B cases, an observant epidemiologist at the Long Beach (California) Department of Public Health noticed that three cases reported receiving injections at a weightreduction clinic [53,54]. Clients at the clinic received injections of human chorionic gonadotropin, usually by jet injector. These devices were developed for mass immunization and work without a needle by forcing medication through the skin under high pressure [55]. Thirty-one cases of clinical hepatitis B among weight loss patients were identified. Only two of the patients had other identified risk factors for acquiring hepatitis B. The results of CDC laboratory testing of the jet injectors were ambiguous. While intentional contamination of the nozzle tip resulted in contamination during subsequent injection into vials, use of the injector on a hepatitis B-infected chimpanzee did not result in nozzle tip contamination when used as directed. However, blood was observed after using the device as instructed by the manufacturer, raising the possibility that contamination might occur via backwash of minute amounts of blood onto the nozzle. A cluster of Plasmodium falciparum malaria cases among patients at a Taipei hospital initiated an epidemiologic investigation [56]. This cluster in 1995 was unusual because, since 1960, all malaria cases in Taiwan were travel-related. A case-control study quickly identified an association between malaria infection and computed tomographic (CT)

SUMMARY

251

examination in the hospital. The cases were all linked to one of four CT scanners. The index case was identified as a patient who had recently traveled to Nigeria. No patients examined prior to the index case had malaria and six of 10 patients examined after the index case were infected. Only those patients (six) receiving contrast agent were infected, whereas those patients (four) who underwent imaging without contrast agent were not infected with P. falciparum. The likely source of infection was contrast medium contamination from blood backflow into tubing and the multiuse bag of contrast medium, which was suspected to have occurred after a power failure during the index patient’s infusion. Four of the seven infected patients died with malaria as an underlying or contributory cause of death. Latex gloves are a ubiquitous medical device required for patient and provider protection. A non-infectious outbreak was reported after the introduction of a new brand of latex gloves [57]. Three cases of starch peritonitis characterized by abdominal pain, distension, nausea, vomiting, low-grade fever and foreign body granuloma were reported in patients after obstetric or gynecologic procedures. Because of these and other problems related to glove powder, the FDA has received requests to ban all glove powders and some gloves are now labeled as ‘powder free’ [58]. A recent outbreak was the cause of a recall of a sublingual pCO2 detector [59]. This device is a disposable probe used to measure the partial pressure of carbon dioxide beneath the tongue as a sign of tissue hypoperfusion. The probe is provided in a metal canister filled with saline and sealed in a foil package. The manufacturer tested inventory samples of the sensor after a hospital informed the company of a rise in Burkholderia cepacia infections in pediatric patients. Reports from the hospital indicated that the device was used on 11 of 13 pediatric patients infected with B. cepacia. The probe and saline both contained the bacteria B. cepacia and other opportunistic pathogens that could cause serious infections [60].

Summary The general model used to epidemiologically characterize disease outbreaks is that of agent–host–environment. Medical device-related outbreaks add another dimension to this model. Tables 17.2–17.5 demonstrate the ubiquity of medical device-related outbreaks. Factors that contribute to this ubiquity are the fact that medical devices are often used in highly vulnerable persons to address a serious medical problem, and the fact that many medical devices are used in a highly invasive manner, thus facilitating the introduction of pathogenic organisms or the perpetration of trauma to the patient. Some medical device-related outbreaks are due to faulty devices, such as the bronchoscope-related outbreaks noted above, resulting from a loose biopsy port, which allowed microbial contamination and biofilm formation. When these kinds of outbreaks are identified, recall of the device, with subsequent alteration of its design, is generally effective in halting the outbreak and preventing the occurrence of similar future outbreaks. Medical device-related outbreaks may be due to chemical contamination

Table 17.5 Device Jet injector

Number of cases 60

Miscellaneous medical device-related outbreaks

Agent

Adverse event

Hepatitis B

Evidence of hepatitis B infection

Contrast medium infusion device catheter

7 (including index case)

Plasmodium falciparum

Latex gloves

3

Starch

Sublingual pCO2 detector

11

Burkholderia cepacia

Liposuction cannulae

9

Mycobacterium spp.

Explanation

Fifty-seven of 279 (24%) clients receiving chorionic gonadotropin injections with a jet injector at a weight reduction clinic had evidence of acute infection with hepatitis B virus Malaria in Contrast medium seven cases infusion device used diagnosed by during CT scanning demonstrating examinations may presence of P. have become falciparum in contaminated during blood smears. a power outage Four patients during examination died with of index case. malaria Approximately 0.25 as underlying ml index patient or contriblood was withdrawn buting cause back into catheter of death when electricity was interrupted. Every patient infused with this device for the remainder of the day after the index case was infected Starch peritonitis Cases were related in patients after to use of starchabdominal powdered latex hysterectomy gloves Positive culture Probe and saline in in patients. packaging cultured Fever, positive for bacteremia, B.cepacia. The pneumonia. Two relationship to pediatric patients sublingual pCO2 died, one from probe not yet pneumonia and determined the other from heart failure Fever, local Rapidly growing inflammation, mycobacteria in microabscesses, liposuction patients purulent from eight hospitals drainage from in Caracas, liposuction Venezuela wound, fistulae

Reference Canter et al. [54]

Chen et al. [56]

Malinger et al. [57]

Recall announcements [59,60]

Torres et al. [61]

253

SUMMARY

Table 17.5 (Continued) Device

Number of cases

Agent

Adverse event

Explanation

Single-use 4 pressure domes introduced for urodynamic studies Re-usable 127 electronic ventilator temperature probes

Pseudomonas aeruginosa

Urinary tract infections, septicemia, meningitis

Pseudomonas cepacia

Respiratory tract infection or colonization among intensive care unit patients

Intra-aortic balloon pump pressure transducer

Serratia marcescens

Bacteremia in cardiac intensive care patients

During cystometry to assess bladder pressure, single-use pressure domes were re-used. These were not decontaminated between patients Patients on ventilation had positive sputum samples. P. cepacia was isolated from odor counteractant cleaning solution used on ventilator probes Case-control study identified intra-aortic pump pressure transducer as source of infection. Cultures of transducers indicated that contamination of the pressure-sensitive membrane in the transducers was the likely source of infection

7

Reference Yardy and Cox[62] and MDA Safety Notification 2000 (09) [63] Weems [64]

Villarino et al. [65]

of devices which may occur as part of the manufacturing process, aging of the device, or cleaning and sterilization procedures. Medical device-related outbreaks may be caused by use-related problems. Perhaps the most common use-related problem is failure to meticulously follow recommended instructions for the prevention of transmission of microorganisms. Outbreaks can also occur when devices are used for purposes other than their intended use (i.e. off-label use.) An example of this is the fatal Rhizopus microsporus infections of preterm neonates due to the use of wooden tongue depressors to construct splints for intravenous cannulation sites, noted above. As with other kinds of outbreaks, good surveillance programs can greatly facilitate the identification of medical device-related outbreaks. Only by identifying these outbreaks can steps be taken to stop them and to institute measures for the prevention of future outbreaks.

254

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

References 1. Gregg MB (ed.). Field Epidemiology. New York: Oxford University Press, 2002. 2. Reingold A. Outbreak investigation: a perspective. Epidemiol Bull 2000; 21: 1–7. 3. Todd EC. Epidemiology of foodborne disease: a worldwide review. World Health Stat Q 1997; 50: 30–50. 4. Koopman M, von Bonsdorff CH, Vinje J, de Medici D, Monroe S. Foodborne viruses. FEMS Microbiol Rev 2002; 26:187–205. 5. Irani DN, Johnson DT. Diagnosis and prevention of bovine spongiform encephalopathy and variant Creutzfeldt–Jakob disease. Annu Rev Med 2003; 54:305–319. 6. Muyembe-Tamfum JJ, Kipasa M, Kiyungu C, Colebunders R. Ebola outbreak in Kikwit, Democratic Republic of the Congo: discovery and control measures. J Infect Dis 1999; 179(suppl 1): S259–262. 7. von Eiff C, Jansen B, Kohnen W, Becker K. Infections associated with medical devices: pathogenesis, management, and propylaxis. Drugs 2005; 65: 179–214. 8. Owings MF, Kozak LJ. Ambulatory and inpatient procedures in the United States, 1996. National Center for Health Statistics. Vital Health Stat 1993; 13(139): http://www.cdc.gov/nchs/data/ series/sr_13/sr13_139.pdf [accessed June 3 2005]. 9. Ramsey AH, Oemig TV, Davis JP, Massey JP, Torok TJ. An outbreak of bronchoscopy-related Mycobacterium tuberculosis infections due to lack of bronchoscope leak testing. Chest 2002; 121: 976–981. 10. Agerton T, Valway S, Gore B, Pozsik C et al. Transmission of a highly drug-resistant strain (strain W1) of Mycobacterium tuberculosis. Community outbreak and nosocomial transmission via a contaminated bronchoscope. J Am Med Assoc 1997; 278: 1073–1077. 11. Kressel AB, Kidd F. Pseudo-outbreak of MycobaceriuM. chelonae and Methylobacterium mesophilicum caused by contamination of an automated endoscopy washer. Infect Control Hosp Epidemiol 2001; 22: 414–418. 12. Kolmos HJ, Lerche A, Kristoffersen K, Rosdahl VT. Pseudo-outbreak of Pseudomonas aeruginosa in HIV-infected patients undergoing fiberoptic bronchoscopy. Scand J Infect Dis 1994; 26: 653. 13. Bryce EA, Walker M, Bevan C, Smith JA. Contamination of bronchoscopes with Mycobacterium tuberculosis. Can J Infect Control 1993; 8: 35–36. 14. Fraser VJ, Jones M, Murray PR, Medoff G et al. Contamination of flexible fiberoptic bronchoscopes with Mycobacterium chelonae linked to an automated bronchoscope disinfection machine. Am Rev Resp Dis 1992; 145: 853–855. 15. Gubler, JG, Salfinger M, von Graevenitz A. Pseudoepidemic of nontuberculous mycobacteria due to a contaminated broncoscope cleaning machine. Report of an outbreak and review of the literature. Chest 1992; 101: 1245–1249. 16. Bronchoscopy-related infections and pseudoinfections – New York, 1996 and 1998. Morb Mortal Wkly Rep 1999; 48: 557–560. 17. FDA and CDC Public Health Advisory: Infections from endoscopes inadequately reprocessed by an automated endoscope reprocessing system: http://www.fda.gov/cdrh/safety/endoreprocess.html [accessed June 3 2005]. 18. Kirschke DL, Jones TF, Craig AS, Chu PS et al. Pseudomonas aeruginosa and Serratia marcescens contamination associated with a manufacturing defect in bronchoscopes. N Engl J Med 2003; 348: 214–220. 19. Srinivasan A, Wolfenden LL, Song X, Mackie K et al. An outbreak of Pseudomonas aeruginosa infections associated with flexible bronchoscopes. N Engl J Med 2003; 348: 221–227.

REFERENCES

255

20. Fiddleman R, Magee A, Patt-Corner R, Jewler L et al. Improving patient safety by eliminating problems in the product alert process: http://info.rasmas.mitretek.org/whitePapers.html [accessed June 28 2004]. 21. Patterson P. Bronchoscope recall falls through cracks. OR Manager 2002; 18: 20. 22. 1999 National Hospital Discharge Survey: annual summary with detailed diagnosis and procedure data. Vital Health Stat 2001: 13(151): 38. 23. Advance Data from Vital and Health Statistics; No. 300. August 12, 1998. Number of ambulatory surgery procedures by procedure category, sex and age: United States, 1996. http://www.cdc. gov/search.do?action¼search&queryText¼Advanceþdataþ300 [accessed November 24, 2006]. 24. Kaczmarek R, Moore RM Jr, McCrohan J, Goldman DA et al. Multi-state investigation of actual disinfection/sterilization of endoscopes in health care facilities. Am J Med 1992: 92: 257–261. 25. Mackeen D. Patients possibly exposed to hepatitis, HIV. Newsday, June 16 2004: http://www.newsday.com/news/health/ny-liendo163852129jun16,0,738070.story [accessed June 29 2004]. 26. The Kaiser Papers: http://www.kaiserpapers.org/scaro.html [accessed June 29 2004]. 27. King W. 600 University of Washington patients told of cleaning lapse. Seattle Times, March 2004: http://www.iconocast.com/H/Health3B_News20_04/Health7.htm [accessed June 29 2004]. 28. Doherty E, Mitchell E, Smyth B. Investigation of the decontamination arrangements for endoscopes in Northern Ireland. Eurosurveillance Wkly 2004; 8(27): http://www.eurosurveillance.org/ew/2004/040701.asp#4 [accessed July 1 2004]. 29. Durante L, Zulty JC, Israel E, Powers PJ et al. Investigation of an outbreak of bloody diarrhea: association with endoscsopic cleaning solution and demonstration of lesions in an animal model. Am J Med 1992; 92: 476–480. 30. Spach DH, Silverstein FE, Stamm WE. Transmission of infection by gastrointestinal endoscopy and bronchoscopy. Ann Intern Med 1993; 118: 117–128. 31. Research updates in kidney and urologic health: http://kidney.niddk.nih.gov/about/Research_Updates/sum02/1.htm [accessed June 3 2004]. 32. Morbidity and mortality of dialysis. NIH Consens Statement Online, November 1–3 1993; 11(2): 1–33: http://consensus.nih.gov/cons/093/093_statement.htm#1_Introdu [accessed June 3 2005]. 33. Welbel SF, Schoendorf K, Bland LA, Arduino MJ et al. An outbreak of Gram-negative bloodstream infections in chronic hemodialysis patients. Am J Nephrol 1995; 15: 1–4. 34. Velandia M, Fridkin SK, Cardenas V, Boshell J et al. Transmission of HIV in dialysis centre. Lancet 1995; 345: 1417–1422. 35. Rudnick JR, Arduino MJ, Bland LA, Cusick L et al. An outbreak of pyrogenic reactions in chronic hemodialysis patients associated with hemodialyzer re-use. Artif Organs 1995; 19: 289–294. 36. Arnow PM, Garcia-Houchins S, Neagle MB, Bova JL et al. An outbreak of bloodstream infections arising from hemodialysis equipment. J Infect Dis 1998; 178: 783–791. 37. Block C, Backenroth R, Gershon E, Israeli R et al. Outbreak of bloodstream infections associated with dialysis machine waste ports in a hemodialysis facility. Eur J Clin Microbiol Infect Dis 1999; 18: 723–725. 38. Flaherty JP, Garcia-Houchins S, Chudy R, Arnow PM. An outbreak of Gram-negative bacteremia traced to contaminated O-rings in reprocessed dialyzers. Ann Intern Med 1993; 119: 1072–1078. 39. Hutter JC, Kuehnert MJ, Wallis RR, Lucas AD et al. Acute onset of decreased vision and hearing traced to hemodialysis treatment with aged dialyzers. J Am Med Assoc 2000; 283: 2128–2134. 40. Hendricks K, Sehulster L, Bell RL, Cousins J et al. Outbreaks of hepatitis B virus infection among hemodialysis patients – California, Nebraska, and Texas. Morb Mortal Wkly Rep 1994; 45: 285–289.

256

CH17

MEDICAL DEVICE-RELATED OUTBREAKS

41. Hutin YJ, Goldstein ST, Varma JK, O’Dair JB et al. An outbreak of hospital-acquired hepatitis B virus infection among patients receiving chronic hemodialysis. Infect Control Hosp Epidemiol 1999; 20: 731–735. 42. Maguire H, Watson J, George R, Taylor L et al. A multidisciplinary investigation of a cluster of deaths on a paediatric intensive care unit. J Hosp Infect 1993; 24: 95–108. 43. Svenningsen NW, Bekassy AN, Christensen P, Kamme C. Nosocomial Klebsiella pneumoniae infection: clinical and hygienic measures in a neonatal intensive care unit. Scand J Infect Dis 1984; 16: 29–35. 44. Van der Zwet WC, Parleviliet GA, Savelkoul PH, Stoof J et al. Outbreak of Bacillus cereus infections in a neonatal intensive care unit traced to balloons used in manual ventilation. J Clin Microbiol 2000; 38: 4131–4136. 45. Lemaitre D, Elaichouni A, Hundhausen M, Claeys G et al. Tracheal colonization with Sphingomonas paucimobilis in mechanically ventilated neonates due to contaminated ventilator temperature probes. J Hosp Infect 1996; 32: 199–206. 46. Berthelot P, Grattard F, Patural H, Ros A et al. Nosocomial colonization of premature babies with Klebsiella oxytoca: probable role of enteral feeding procedure in transmission and control of the outbreak with the use of gloves. Infect Control Hosp Epidemiol 2001; 22: 148–151. 47. Brown SL, Bright RA, Dwyer DE, Foxman B. Breast pump adverse events reports to the Food and Drug Administration. J Hum Lact 2005; 21:169–174. 48. Boo N-Y, Nordiah AJ, Alfizah H, Nor-Rohaini AH, Lim VKE. Contamination of breast milk obtained by manual expression and breast pumps in mothers of very low birthweight infants. J Hosp Infect 2001; 49: 274–281. 49. Blenkharn JI. Infection risks from electrically operated breast pumps. J Hosp Infect 1989; 13: 27–31. 50. Mitchell SJ, Gray J, Morgan MEI, Hocking MD, Durbin GM. Nosocomial infection with Rhizopus microsporus in preterm infants: association with wooden tongue depressors. Lancet 1996; 348: 441–443. 51. Holzel H, Macqueen S, MacDonald A, Alexander S et al. Rhizopus microsporus in wooden tongue depressors: a major threat or minor inconvenience? J Hosp Infect 1998; 38: 113–118. 52. Hazard Notice MDA HN 9604 May 1996, Wooden tongue depressor: http://www.medical-devices. gov.uk/mda/mdawebsitev2.nsf/e8be0ee313c493aa80256bbb00307b2e/ d6850ebf02d43d8f80256c8b003aa295/$FILE/hn9604.pdf 53. Shah R, Mackey K, Wallace H, Yawata K et al. Epidemiologic notes and reports hepatitis B associated with jet gun injection – California. Morb Mortal Wkly Rep 1986; 35: 373–376. 54. Canter J, Mackey K, Good LS, Roberto RR et al. An outbreak of hepatitis B associated with jet injections in a weight reduction clinic. Arch Intern Med 1990; 150: 1923–1927. 55. Science and Society Picture Library: http://www.nmsi.ac.uk/piclib/imagerecord. asp?id¼ 10287525 [accessed June 3 2005]. 56. Chen K-T, Chen C-J, Chang P-Y, Morse DL. A nosocomial outbreak of malaria associated with contaminated catheters and contrast medium of a computed tomographic scanner. Infect Control Hosp Epidemiol 1999; 20: 22–25. 57. Malinger G, Ginath S, Zeidel L, Avinoach I et al. Starch peritonitis outbreak after introduction of a new brand of starch-powdered latex gloves. Acta Obstet Gynecol Scand 2000; 79: 610– 611. 58. Medical glove powder report: http://www.fda.gov/cdrh/glvpwd.html [accessed Jun 3 2005]. 59. Nellcor announces nationwide voluntary recall of CapnoProbeTM sensors: http://www.fda.gov/ oc/po/firmrecalls/nellcor08_04.html [accessed September 10 2004]. 60. FDA announces a nationwide recall of Nellcor Puritan Bennet probes device: http://www.fda.gov/bbs/topics/news/2004/NEW01108.html [accessed September 10 2004].

REFERENCES

257

61. Torres J, Murillo J, Bofill L, Raeos A et al. Rapidly growing Mycobacterial infection following liposuction and liposculpture – Caracas, Venezuela, 1996–1998. Morb Mortal Wkly Rep 1998; 47: 1065–1067. 62. Yardy GW, Cox RA. An outbreak of Pseudomonas aeruginosa infection associated with contaminated urodynamic equipment. J Hosp Infect 2001; 47: 60–63. 63. Pressure transducer cover (SensorNor Pressure Dome) for urodynamic system of bladder pressure measurement – risk of cross infection if cover is re-used: http://www.medicaldevices.gov.uk/mda/mdawebsitev2.nsf/0/0D57D77D23BB0BBC00256AA400523282?OPEN [accessed June 3 2005]. 64. Weems JJ Jr. Nosocomial outbreak of Pseudomonas cepacia associated with contamination of re-usable electronic ventilator temperature probes. Infect Control Hosp Epidemiol 1993; 14: 584–586. 65. Villarino ME, Jarvis WR, O’Hara C, Bresnahan J, Clark N. Epidemic of Serratia marcescens bacteremia in a cardiac intensive care unit. J Clinical Microbiol 1989; 27: 2433–2436.

18 Risk of transmission of prions with medical devices S. Lori Brown US Food and Drug Administration, Bothell, WA, USA

Azadeh Shoaibi US Food and Drug Administration, Rockville, MD, USA

Prions are pathogenic proteins which cause neurodegenerative disease in animals and humans. These invariably fatal diseases include scrapie in sheep, bovine spongiform encephalopathy (BSE) in cattle, kuru, Creutzfeldt–Jakob disease (CJD), and new variant Creutzfeldt–Jakob disease (vCJD) in humans [1–4]. Human prion diseases are categorized as sporadic, acquired, and genetic [5]. Sporadic cases occur when no factor can be identified as the source of prions and these cases are suspected to have arisen by conformation change of normal cellular prion protein (PrPc) to a pathogenic prion protein (PrPSc). Sporadic CJD cases occur with an annual incidence estimated at one per million [5]. Acquired disease is due to pathogenic prions transmitted between humans or between other animals and humans, as described below. This transmission may be iatrogenic or from consuming prion-infected tissues, as in the case of kuru. Transmission between species does occur, although species barriers seem to generally prohibit transmission between species (sheep and humans in the case of scrapie, for instance, but not cattle and humans as in the case of vCJD from BSE) [5]. Genetic disease is associated with coding mutations in the normal prion protein gene and accounts for 15% of prion disease [5]. Prion disease prevalence in the US population was estimated to be less than 1 per 100 000 population in 2000 [2].

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

260

CH18

RISK OF TRANSMISSION OF PRIONS WITH MEDICAL DEVICES

Human disease onset after exposure to prions is typically long, up to decades after exposure. As a result, patients may be infected for long periods of time and remain asymptomatic. vCJD is distinguished by the relatively young age of onset and its rapidly fulminating course, as well as the predominance of early psychiatric symptoms and subsequent evolution of neurologic symptoms [6]. The electrophoretic properties of prion protein associated with vCJD resemble those found in brains of cattle with BSE and differ from those found in cases of other forms of CJD [7,8]. The similarity between isotypes of prion protein found in BSE and vCJD plaques as well as their efficient transmission despite the species barrier [9,10] provides strong evidence that vCJD is caused by the same agent as BSE. Unlike other infectious agents, prions do not have nucleic acid but are proteinaceous. However, these proteins are uniquely resistant to protease degradation. Prions appear to ‘infect’ via the unusual mechanism of converting or modifying normal cellular prion proteins (PrPc) into an infectious isoform (PrPSc). Prions are notoriously difficult to degrade or disinfect, with reports that some prions may withstand incineration at up to 600 C [11]. They are also resistant to inactivation by UV and ionizing radiation [1]. Because of the difficulty of disinfecting prions, there are two distinct issues that affect medical device safety: 1. Iatrogenic transmission of prion disease with contaminated invasive neurological or surgical instruments. 2. Iatrogenic transmission of prion disease from medical devices manufactured with a bovine or human component.

Iatrogenic transmission of prion disease via neurological or surgical instruments A history of surgery is associated with sporadic CJD in some [12,13], but not all, casecontrol studies comparing CJD cases to controls [14]. To date, there are no reports of vCJD transmission via contaminated surgical instruments. There have been several reports of CJD transmission from the use of contaminated medical devices which occurred in the 1970s. Two patients undergoing stereotactic electroencephalographic depth recording to characterize their epileptic conditions developed CJD within 2 years of the procedure [15]. Both cases were confirmed by histopathology at necropsy and by transmission of a progressive neurological disease to primates inoculated with brain suspension from these two patients. In retrospect, the electrodes had been previously used in a patient diagnosed with familial presenile dementia and myoclonic jerks in an attempt to explore and treat her myoclonus. After use in this patient, the electrodes were cleaned with benzene and sterilized in 70% alcohol and formaldehyde vapor. The diagnosis of all three patients was subsequently confirmed by transmission of CJD to a chimpanzee by implanting the electrodes directly into its brain over 2 years after the electrodes had last been used [16]. Within 18 months, the chimpanzee developed

SURGICAL INSTRUMENTS USED TO TRANSPLANT TISSUES

261

symptoms of neurological disease and showed signs of spongiform encephalopathy at autopsy. The wire electrodes that were used only on the scalp of these patients did not transmit disease to primates. Another neurosurgical transmission of CJD was reported in which the primary case, a 59 year-old woman, previously in apparent good health, began to feel tired and stopped working [17]. Soon after, she fell down a staircase at home. She was admitted to the hospital and electroencephalogram results were suggestive of CJD. In the course of testing, she was referred to the neurosurgery department and underwent a right frontal cortical biopsy, which confirmed the diagnosis. Her condition rapidly degenerated and she died two months after the first recorded symptom. The secondary patient was admitted for a severe headache following head trauma, also from a fall down stairs. He underwent a left fronto-temporal trephination followed by biopsy in the same operating room as the primary case, 3 days later. There were three sets of surgical instruments in the operating room used for this procedure, one of which had been used on the primary CJD patient a few days earlier. After washing with soap, the instruments were sterilized with dry heat at 180 C for 96 minutes. It is likely that the same set of instruments was used on the secondary patient. The primary case was believed to be a spontaneous case of CJD, based on molecular genetics testing that did not identify any pathogenic mutations of the PRNP (prion protein) gene, which encodes the normal host protein. In their worldwide survey of iatrogenic CJD cases, Brown et al. [18] noted that there are four reported cases of CJD transmission related to neurosurgery in addition to the two cases reported associated with stereotactic EEG probes. This is a small component of all iatrogenic cases: 139 reportedly after human growth hormone (GH) therapy with GH extracted from human pituitary, and 114 cases apparently transmitted by dura mater grafts (a medical device discussed below). Despite the low number of cases identified as related to surgical instrument transmission, there is concern that surgical instruments and other medical devices may serve as a reservoir for prions. A potential model for experimental transmission of prions by surgical instruments is the use of stainless steel wire. In one such experiment, stainless steel wires were exposed to freshly prepared brain homogenates from scrapie-infected mice [19]. These were subsequently washed or washed and treated with formaldehyde. Formaldehyde treatment reduced but did not eliminate the infectivity of the wires to subsequent mice. This study also confirmed the tight binding of prions to stainless steel, since activity could not be eluted from the stainless steel wires. Another study demonstrated that exposure of stainless steel wire segments to scrapie-infected mouse brain for five minutes was adequate to transmit scrapie to subsequent mice [20].

Surgical instruments used to transplant tissues known to have transmitted prion disease (corneal transplant) Surgical instruments that are used to transplant or surgically implant (or explant) tissues that have been associated with CJD transmission are also of concern. For instance, there

262

CH18

RISK OF TRANSMISSION OF PRIONS WITH MEDICAL DEVICES

have been several cases of CJD transmitted via corneal transplant [18,21] and there is also a theoretical risk of contaminating surgical instruments used to harvest corneas from cadavers or transplant them to recipients. Cross-contamination of harvested corneas by using contaminated instruments is of concern for the eye-banking community. The estimated potential loss of donated corneas by screening and deferring potentially infected cadavers (for CJD) would have an arguably large impact on the eye-banking industry [22]. One report estimates that only 1.3 of 45 000 harvested corneas would be likely to have CJD (based on a study of age-specific annual death records in the USA). They therefore find that given a 90% sensitivity of screening of the highest risk rate age groups would result in the loss of 21 000 donors over an 18 year period in order to correctly exclude one donor. However, while there is concern by industry over the loss of potential corneas for transplant due to more stringent donor screening, there is also concern over the potential cost of contaminating surgical instruments used on a donor who is later found to have CJD (or vCJD). The risk of contaminating surgical instruments during cornea extraction was discussed during a Food and Drug Administration Transmissible Spongiform Encephalopathy Advisory Committee. A representative from the eye-banking community described the potential astronomical cost of destroying reprocessed surgical instruments used to harvest corneas from cadavers if any donor was subsequently found to have CJD or vCJD [23]. While the potential for transmission via contaminated instruments may be remote, it cannot be ruled out. In one case-control study, ophthalmic surgery was not associated with CJD or vCJD [24], but another casecontrol study found that the risk for CJD was increased six-fold in patients who had undergone cataract or eye surgery (not corneal transplant) [13].

Surgical instruments in contact with lower-risk peripheral tissues Surgical instruments which contact high-risk neurological or ophthalmic tissues are of concern, but peripheral tissues may also be infected with prions. For instance, prion protein has been found in lymphatic tissue, including tonsils [25–27] and a recent report described evidence of prions in three appendix samples from 12 674 UK surgical patients [28]. In addition, 19 of 20 appendixes removed from vCJD patients at autopsy tested positive for prion protein. Pathologic prion protein has also been detected in the neuroepithelium of the olfactory mucosa in individuals with sporadic CJD, indicating that the olfactory tract is involved in the disease [29]. It is possible that the prion responsible for vCJD is more infectious than the prion responsible for (sporadic or acquired) CJD [30]. Reported transmission of vCJD by blood transfusion has recently been reported [31] but the failure to identify cases of CJD from blood transfusion indicates that plasma pools from patients with CJD, while possibly infected with prion, are not known to transmit disease [32]. A possible explanation is higher prion concentrations in peripheral tissue and blood in vCJD than in CJD [33]. Screening recommendations to minimize risk of transmission of spongiform encaphalopathies by blood are in place [34].

DECONTAMINATING SURGICAL INSTRUMENTS TO REDUCE RISK OF TSE TRANSMISSION

263

Decontaminating surgical instruments to reduce risk of TSE transmission Washing surgical instruments under the best of circumstances may be inadequate to remove prions. As described by the US Centers for Disease Control and Prevention, scientists disagree on the methods needed to deactivate prions [35]. The World Health Organization (WHO) advocates mechanical cleaning to reduce the bio-load and decontamination using multiple methods when possible, including autoclaving in sodium hydroxide [36]. Studies in FDA laboratories indicate that, in addition to the issue of laboratory personnel safety, such harsh methods may have an impact on the instruments themselves, causing corrosion or discoloration [37] and potential damage to autoclaves [38]. The WHO also recognizes that some complex medical devices and instruments are difficult to decontaminate and suggests that: ‘to the extent possible, such instruments should be protected from surface contamination by wrapping or bagging with disposable materials. Those parts of the device that come into contact with internal tissues of patients should be subjected to the most effective decontaminating procedure that can be tolerated by the instrument . . .’

In fact, some medical devices are so complex that cleaning and disinfecting them thoroughly may be difficult even for infectious agents that are more easily decontaminated than prions (see e.g. section on Endoscopes [39] in Chapter 17). Another group of devices known to be difficult to clean are dental instruments. One study suggests that routine cleaning and disinfection of endodontic instruments with narrow channels or serrated edges may be inadequate to remove prion contamination [40]. However, in another study, dental tissue (dental pulp, gingiva, salivary gland, tongue) from vCJD patients at autopsy did not have detectable prion protein [27]. The procedures required for eliminating any chance of CJD or variant vCJD from iatrogenic transmission between patients via surgical instruments used in invasive procedures are prohibitively expensive and would require elimination of virtually all reprocessing and re-use of surgical instruments or devices. The UK Department of Health has recently recommended that all surgical instruments used in brain biopsy operations should be quarantined until a definitive diagnosis is made [41] and that all tonsillectomies be performed with disposable instruments, a recommendation they subsequently modified [42,43]. The fear associated with inadvertent use of medical devices potentially contaminated by contact with infected patients may pose a serious problem. For instance, in Australia, nine neurosurgical patients were advised that they may have been exposed to surgical equipment that was used on a patient with CJD without proper sterilization [44]. Likewise, a recent news report warned thousands of surgical patients in the UK of their possible exposure to instruments used on patients with vCJD [45]. Because of the potential for long incubation periods, it may be years or decades before patients show

264

CH18

RISK OF TRANSMISSION OF PRIONS WITH MEDICAL DEVICES

signs or symptoms of disease. This uncertainty is a great burden for surgical patients and their families.

Decontamination and sterilization of surgical instruments Validation of methods and techniques for decontaminating and sterilizing instruments (or environments) that come into contact with prion-infected tissues is important [46–48]. There is research in this area but a thorough evaluation of the claims, some with commercial interest, will be important. In addition, methods for decontaminating or sterilizing medical devices must not interfere with the properties or the operation of the device: a procedure which results in corrosion or weakening of the device would be unacceptable. In addition, the methods must be accessible for a typical hospital or even a doctor’s or dentist’s office [49].

Theoretical transmission of prions by medical devices with a tissue component Many medical devices are manufactured with tissue components from animals. In addition, there are medical devices consisting of human tissue or manufactured with a human tissue component. A well-known example of this is cadaveric human dura mater [18,50]. This processed, lyophilized product is a medical device and is intended to repair defects in the dura mater. It is also reportedly used for diverse purposes, including treatment of vesicovaginal fistulae [51], as a palatal mucosal graft for vestibuloplasty [52], or reconstruction of the pelvic floor [53]. There have been at least 114 cases of CJD reported world-wide related to dura mater use [18]. In Japan, where dura mater grafts were used ‘not as functioning organs but as merely medical dressings, like bandages’, the relative risk for CJD for patients with a dura mater transplantation was calculated to be at least 29.2 [54]. In the USA, a patient advocacy group petitioned the FDA to ban the sale of human cadaveric dura mater and recall any unimplanted human cadaveric dura mater from all sources [55]. The FDA did not concur with this because most cases were due to use of dura from outside the USAwhich had not been processed properly. FDA released guidance on the subject for industry and FDA staff, reasoning that special controls, such as donor qualification screening and appropriate tissue processing methods, would be adequate to provide reasonable assurance of safety and effectiveness of processed human dura mater [56]. Given the history of the iatrogenic transmission of CJD from human dura mater transplantation (a medical device) and pituitary-derived growth hormone [18], it is not implausible that devices with a bovine component could theoretically transmit vCJD to a susceptible individual. It should be recognized that some tissues are higher risk than others for prion transmission with the highest risk tissues from the central nervous system, including eyes. The WHO has provided a guidance based on the tissue type (e.g. central nervous system, muscle, tonsils, etc) and its relative infectivity [57].

DECONTAMINATING SURGICAL INSTRUMENTS TO REDUCE RISK OF TSE TRANSMISSION

265

Medical devices with bovine components Because of the theoretical potential for transmission of vCJD to patients exposed to bovine tissue-containing devices, we developed a list of products with a bovine component. A list of medical devices was compiled based on a Center for Devices and Radiological Health database (the Biocompendium) that listed products that have an animal tissue component (this database was designed by Stanley A. Brown at CDRH). This database had not been updated consistently and was an incomplete list which we supplemented with a list of devices containing a bovine component reported by division directors from the Office of Device Evaluation, the office that reviews and clears or approves medical devices, and was provided by Karen Warburton of the CDRH TSE Working Group (Table 18.1). The tissue origin for bovine materials (as far as was possible, based on sometimes limited information) was determined for all products on the list. There were 49 product categories that included 149 trade name products. For the most part, bovine-containing products had components that were rated by WHO as being of low or no detected infectivity (e.g. collagen and bone) [57]. However, there were some products in which a specification of the tissue type was vague (e.g. viscera) or that it contained components from higher-risk tissues (e.g. pituitary extracts). Assessing risk of prion transmission for these products would be dependent on a number of factors, including the intended use of the product, country of origin of the bovine material, age of animal at slaughter, amount of bovine material in product, processing details, and other factors which we could not assess. Information on these factors is not routinely available for medical devices, since there are no regulatory requirements that it shall be available. Since this information is not available for these products, an in-depth risk analysis for each product could not be performed. It is of interest that others have used a model rating surgical sites assessing potential TSE transmission, based on the operative sites using the WHO classification of tissue infectivity [58]. This may be of use since, for instance, implanting high-risk material into high-risk tissue (e.g. pituitary tissue implanted in brain) is likely to be more risky that implanting low-risk material into low-risk sites (e.g. collagen injected in skin). Our own analysis indicated that, for the most part, products of bovine origin were from lower-risk tissues and were typically in contact with only lower-risk human tissues. None of the devices we identified fell into the highest risk category of both high-risk bovine tissue of origin and high-risk human tissue destination. Several risk analyses have been published that delve into factors to consider in assessing potential risk from using bovine products in human devices and drugs [59– 61]. For the most part, these risk analyses take into consideration the factors mentioned above, such as country of origin, tissue of origin, intended dose, quantities of raw material used to manufacture the product, and tissue processing which presumably (hopefully) leads to lowering infectivity. These risk analyses do not take into consideration the risk:benefit ratio of the medical product (e.g. life-saving vs. cosmetic), the availability of products that serve the same purpose without bovine components, the potentially devastating effect on the medical device (or drug) industry if there is a single case of vCJD transmitted via a medical product, or the loss of public confidence if

266

CH18

RISK OF TRANSMISSION OF PRIONS WITH MEDICAL DEVICES

Table 18.1 List of medical devices that have a bovine tissue component

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

Medical device

Manufactured from bovine tissue type

Suture, absorbable, natural Device, bypass, ventricular assist Prosthesis, arterial (or vascular) graft, synthetic, 6 mm Catheter, canula and tubing, vascular, cardiopulmonary bypass Patch, pledget, and intracardiac Heart valve replacement Prosthesis, arterial (or vascular)graft, <6 mm diameter Catheter, intravascular, therapeutic, <30 days Dressing Mesh, surgical, polymeric Breast implant Prosthesis, arterial graft, bovine carotid artery Snare, surgical Suture, absorbable, natural Suture, nonabsorbable, steel, monofilament, sterile Implant, eye sphere Implant, orbital, extraocular Cuff, nerve Polymer, ENT natural collagen material (for vocal cord) Bandage, liquid, dressing Catheter, intravascular, therapeutic, >30 days Agent, absorbable hemostatic, collagen-based Implant, dermal (collagen) for aesthetic use Agent, injectable, bulking, for gastro-urology Grafts, biological, arterial Material, dressing, surgical, polylactic acid Defibrillator, automatic implantable cardioverter Bone grafting material Media, corneal, storage Plug, punctum Graft, vascular, biological Graft, vascular, synthetic/biologic composite Bone graft substitute Filler, bone void, non-osteoinduction Graft, vascular, hemodialysis access, synthetic/biological composite Barrier, absorbable, adhesion Patch, pericardial Device, hemostatis, vascular Dressing, wound and burn, occlusive Dressing, wound and burn, interactive Collagen corneal shield

Viscera Pericardium Collagen, pericardium, gelatin, not specified Pericardium Pericardium, not specified Pericardium Blood vessel Collagen Collagen, chondroitin, serum Pericardium Collagen Blood vessel Viscera Viscera Viscera Bone, collagen Pericardium Ligament, tendon Collagen Collagen Collagen Collagen Collagen Collagen Blood vessels Collagen Collagen Bone, collagen Not specified Collagen Blood vessel, Collagen, gelatin Growth hormone, collagen Collagen Not specified Not specified Pericardium Collagen Collagen Collagen, pituitary extract, serum Collagen (Continued )

FDA MEASURES TO MINIMIZE RISK OF TRANSMISSION

267

Table 18.1 (Continued )

42

43

44 45 46 47 48 49

Medical device

Manufactured from bovine tissue type

Filler, recombinant human bone morphogenic protein, collagen scaffold, osteoinduction Filler, bovine protein mixture, bone morphogenic proteins, collagen scaffold, osteoinduction Implant, resorbable, bovine collagen, meniscal repair Media, reproductive Wrap, implant, orbital Device, separation, sperm Barrier, animal source, dental Bone grafting material, animal source

Collagen, growth hormone, bone morphogenic proteins Collagen

Collagen Not specified Pericardium Not specified Not specified Not specified

FDA is perceived as being incapable of protecting the public from contaminated medical devices or drugs. Nonetheless, these published analyses do serve the useful purpose of emphasizing how much we have to learn about TSEs before we can definitively assess risk. In addition, risk assessment has been applied to specific products. One such product is bone grafts, and several risk assessments describe the risk of transmission of disease to humans from bone grafts derived from bovine bone based on varying assumptions [62,63]. They consider the intended use of the product (as application route). They conclude that the risk of transmission is low but not zero.

FDA measures to minimize risk of transmission of BSE by medical products FDA has taken measures to protect the public from bovine spongiform encephalopathy. FDA regulates animal feed and has restricted feeding of most mammalian proteins to cattle and other ruminant animals. Additional steps to protect consumers and patients are restrictions on the import of cattle into the USA and rapid response to any report of BSE in the USA. The FDA is also involved in the development of diagnostic tests to quickly identify contamination with the BSE agent. CDHR also provides guidance to industry, which recommends restricting the source of bovine material used in manufacturing medical devices to countries that are free of BSE, maintaining traceable records for each lot of bovine material used in the manufacture of medical devices, and identifying all bovine tissue used in the manufacture of devices [64,65]. If bovine material is used from countries that cannot be certified free of BSE, then CDRH recommends that the manufacturer should provide evidence to indicate that the BSE agent is inactivated during the manufacturing process. These steps, and others recommended by CDRH, should reduce the risk of transmitting BSE to humans via medical devices with bovine components.

268

CH18

RISK OF TRANSMISSION OF PRIONS WITH MEDICAL DEVICES

Summary There are two distinct theoretical risks for transmitting TSE prions with medical devices. The first is the accidental transmission from patient to patient with surgical devices contaminated by pathogenic prions, and the second is the potential manufacture of medical devices with a component from animal tissue carrying pathogenic prions, followed by transmission to humans through medical procedures. Of particular interest these days is the potential for BSE or vCJD transmission via these mechanisms. To date, there are no recognized cases of either of these events occurring, but there is concern that there is potential for either to occur because it has been documented for CJD. Risk assessments are useful for developing an understanding of potential risk and they emphasize our lack of information to assess the potential transmission of TSEs because of the data gaps on the topic.

References 1. Prusiner SB. Prions. Proc Natl Acad Sci USA 1998; 95: 13363–13383. 2. Prusiner SB. Shattuck lecture – neurodegenerative diseases and prions. N Engl J Med 2001; 344: 1516–1526. 3. Johnson RT. Prion diseases. Lancet Neurol 2005; 4: 635–642. 4. Belay ED, Schonberger LB. The public health impact of prion diseases. Annu Rev Publ Health 2005; 26: 191–212. 5. Collinge J. Variant Creutzfeldt–Jakob disease. Lancet 1999; 354: 317–323. 6. Spencer MD, Knight RSG, Will RG. First hundred cases of variant Creutzfeldt–Jakob disease: retrospective case note review of early psychiatric and neurological features. Br Med J 2002; 324:1479–1482. 7. Collinge J, Sidle KCL, Meads J, Ironside J, Hill AF. Molecular analysis of prion strain variation and the aetiology of new variant CJD. Nature 1996; 383:685–690. 8. Will RG. A new variant of Creutzfeldt–Jakob disease in the UK. Lancet 1996; 347: 921–925. 9. Hill AF, Desbruslais M, Joiner S, Sidle KCL et al. The same prion strain causes vCJD and BSE. Nature 1997; 389: 448–450. 10. Bruce ME, Will RG, Ironside JW, McConnell I et al. Transmissions to mice indicate that ‘new variant’ CJD is caused by the BSE agent. Nature 1997; 389: 498–501. 11. Brown P, Rau EH, Johnson BK, Bacote AE et al. New studies on the heat resistance of hamsteradapted scrapie agent: threshold survival after ashing at 600 C suggests an inorganic template of replication. Proc Natl Acad Sci USA 2000; 97: 3418–3421. 12. Ward HJT, Everington D, Croes EA, Alperovitch A et al. Sporadic Creutzfeldt–Jakob disease and surgery: a case-control study using community controls. Neurology 2002; 59: 543–548. 13. Collins S, Law MG, Fletcher A, Boyd A et al. Surgical treatment and risk of sporadic CJD disease: a case control study. Lancet 1999; 353: 693–697. 14. Wientjens DPWM, Davanipour Z, Hofman A, Kondo K et al. Risk factors for Creutzfeldt–Jakob disease: a reanalysis of case-control studies. Neurology 1996; 46: 1287–1291. 15. Bernoulli C, Siegfried J, Baumgartner G, Regli F et al. Danger of accidental person-to-person transmission of Creutzfeld–Jakob disease by surgery. Lancet 1977; i: 478–479.

REFERENCES

269

16. Gibbs CJ Jr, Asher DM, Kobrine A, Amyx HL et al. Transmission of Creutzfeldt–Jakob disease to a chimpanzee by electrodes contaminated during neurosurgery. J Neurol Neurosurg Psychiat 1994; 57: 757–758. 17. El Hachimi KH, Chaunu M-P, Cervenakova L, Brown P, Foncin J-F. Putative neurosurgical transmission of Creutzfeldt–Jakob disease with analysis of donor and recipient: agent strains. C. R. Acad Sci III 1997; 320: 319–328. 18. Brown P, Preece M, Brandel J-P, Sato T et al. Iatrogenic Creutzfeldt–Jakob disease at the millennium. Neurology 2000; 55: 1075–1081. 19. Zobeley E, Flechsig E, Cozzio A, Enari M, Weissmann C. Infectivity of scrapie prions bound to a stainless steel surface. Mol Med 1999; 5: 240–243. 20. Flechsig E, Hegyi I, Enari M, Schwarz P et al. Transmission of scrapie by steel-surface-bound prions. Mol Med 2001; 7: 679–684. 21. Will RG. Acquired prion disease: iatrogenic CJD, variant CJD, Kuru. Br Med Bull 2003; 66: 255–265. 22. Kennedy RH, Hogan RN, Brown P, Holland E et al. Eye banking and screening for Creutzfeldt– Jakob disease. Arch Ophthalmol 2001; 119: 721–726. 23. FDA Transmissible Spongiform Encephalopathy Advisory Committee Meeting, July 18 2003: http://www.fda.gov/ohrms/dockets/ac/03/briefing/3969B2_2a.pdf [accessed June 5 2005]. 24. Juan PS, Ward HJT, De Silva R, Knight RSG, Will RG. Ophthalmic surgery and Creutzfeldt– Jakob disease. Br J Opthalmol 2004; 88: 446–449. 25. The BSE Inquiry, vol 8. The emergence of variant CJD: http://www.bseinquiry.gov.uk/report/ volume8/chapteb6.htm [accessed June 5 2005]. 26. Hill AF, Butterworth RJ, Joiner S, Jackson G et al. Investigation of variant Creutzfeldt–Jakob disease and other prion diseases with tonsil biopsy samples. Lancet 1999; 353: 183–189. 27. Head MW, Ritchie D, McLoughlin V, Ironside JW. Investigation of PrPres in dental tissue in variant CJD. Br Dental J 2003; 195: 339–343. 28. Hilton DA, Ghani AC, Conyers L, Edwards P et al. Prevalence of lymphoreticular prion protein accumulation in UK tissue samples. J Pathol 2004; 202: 300–302. 29. Zanusso G, Casalone C, Acutis P, Bozzetta E et al. Detection of pathologic prion protein in the olfactory epithelium in sporadic Creutzfeldt–Jakob disease. N Engl J Med 2003; 348: 711–719. 30. Nart P. Variant CJD may be more infectious than sporadic CJD. ProMED posting, September 19 2001; Posting No. 20010924.2322: http://www.promedmail.org/pls/promed/f?p¼2400:1202: 14707927578837286057::NO::F2400_P1202_CHECK_DISPLAY,F2400_P1202_PUB_MAIL_ ID:X,14531 31. Llewelyn CA, Hewitt PE, Knight RSG, Amar K et al. Possible transmission of variant Creutzfeldt–Jakob disease by blood transfusion. Lancet 2004; 363: 417–421. 32. Brown P. Pathogenesis and transfusion risk of transmissible spongiform encephalopathies. Dev Biol 2005; 120: 27–33. 33. Collins SJ, Lawson VA, Masters CL. Transmissible spongiform encepahlopathies. Lancet 2004; 363: 51–61. 34. Revised preventive measures to reduce the possible risk of transmission of Creutzfeldt–Jakob disease (CJD) and variant Cretuzfeldt–Jakob disease (vCJD) by blood and blood products: http://www.fda.gov/cber/gdlns/cjdvcjd.pdf [accessed November 25 2005]. 35. Guidelines for environmental infection control in health-care facilities, 2003: http://www. cdc.gov/ncidod/hip/enviro/Enviro_guide_03.pdf [accessed June 5 2005]. 36. WHO Infection control guidelines for transmissible spongiform encephalopathies: http://www. who.int/emc-documents/tse/docs/whocdscsraph2003.pdf 37. Brown SA, Merritt K, Woods TO, Busick D. Effects on instruments of the World Health Organization – recommended protocols for decontamination after possible exposure to

270

38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55.

56. 57.

58.

59.

CH18

RISK OF TRANSMISSION OF PRIONS WITH MEDICAL DEVICES

transmissible spongiform encephalopathy-contaminated tissue. J Biomed Mater Res B Appl Biomater 2005; 72: 186–190. Brown SA, Merritt K. Use of containment pans and lids for autoclaving caustic solutions. Am J Infect Control 2003; 31: 257–260. Bramble MG, Ironside JW. Creutzfeldt–Jakob disease: implications for gastroenterology. Gut 2002; 50: 888–890. Smith A, Dickson M Aitken J Bagg J. Contaminated dental instruments. J Hosp Infect 2002; 51: 233–235. Mayor S. UK government advises tighter measures to reduce risk of CJD transmission during neurosurgery. Br Med J 2003; 326: 517. Single-use instruments for tonsil and adenoid surgery: http://www.dh.gov.uk/assetRoot/04/01/ 41/67/04014167.pdf Frosh A, Joyce R, Johnson A. Iatrogenic vCJD from surgical instruments. Br Med J 2001; 322: 1558–1559. Zinn C. Nine hospital patients may have been exposed to CJD. Br Med J 2000; 320: 1296. Leake J. Thousands of UK hospital patients exposed to vCJD/mad cow: http://www.rense.com/ general14/paat.htm [accessed November 2 2004]. Manuelidis L. Decontamination of Creutzfeldt–Jakob disease and other transmissible agents. J Neurovirol 1997; 3: 62–65. Race RE, Raymond GJ. Inactivation of transmissible spongiform encephalopathy (prion) agents by Environ LpH. J Virol 2004; 78: 2164–2165. Fichet G, Camoy E, Duval C, Antioga K et al. Novel methods for disinfection of prioncontaminated medical devices. Lancet 2004; 364: 521–526. Sehulster LM. Prion inactivation and medical instrument reprocessing: challenges facing healthcare facilities. Infect Control Hosp Epidemiol 2004; 25: 276–279. Hannah EL, Belay ED, Gambetti P, Krause G et al. Creutzfeldt–Jakob disease after receipt of a previously unimplicated brand of dura mater graft. Neurology 2001; 56: 1080–1083. Alagol B, Gozen AS, Kaya E, Inci O. The use of human dura mater as an interposition graft in the treatment of vesicovaginal fistula. Int Urol Nephrol 2004; 36: 35–40. Sezer B, Selcuk E, Erturk S, Gomel M. Comparison of autogenous mucosal grafts and collagenbased solvent-preserved allografts for vestibuloplasty. Quintessence Int 2004; 35: 234–239. Salom EM, Penalver MA. Pelvic exenteration and reconstruction. Cancer J 2003; 9: 415–424. Nakamura Y, Aso E, Yanagawa H. Relative risk of Creutzfeldt–Jakob disease with cadaveric dura transplantation in Japan. Neurology 1999; 53: 218–220. Petition to FDA asking them to ban the use of cadaveric dura mater because it risks exposing neurosurgery patients to Creutzfeldt–Jakob disease. HRG Publication No. 1587: http://www. citizen.org/publications/release.cfm?ID¼7050 [accessed November 2 2004]. Durfor C. Class II special controls guidance document: human dura mater. http://www.fda.gov/ cdrh/ode/guidance/054.html [accessed June 6 2005]. WHO Guidelines on transmissible spongiform encephalopathies in relation to biological and pharmaceutical products: http://www.who.int/biologicals/Meeting-Reports/Doc/whotse 2003.pdf Rabano A, de Pedro-Cuesta J, Molbak K, Siden A et al.; EUROSURGYCJD Research Group. Tissue classification for the epidemiological assessment of surgical transmission of sporadic Creutzfeldt–Jakob disease. A proposal on hypothetical risk levels. BMC Publ Health 2005; 24: 5–9. Rohwer RG. Analysis of risk to biomedical products developed from animal sources. Dev Biol Stand 1996; 88: 247–256.

REFERENCES

271

60. Bader F, Davis G, Dinowitz M, Garfinkle B et al. Pharmaceutical Research and manufacturers of America (PhRMA) BSE committee. Assessment of risk of bovine spongiform encephalopathy in pharmaceutical products. Biopharmacology 1998; 11: 18–31. 61. Committee for Proprietary Medicinal Products. Note for guidance on minimizing the risk of transmitting animal spongiform encephalopathy agents via medicinal products. European Agency for the Evaluation of Medicinal Products, Human Medicines Evaluation Unit, London, 2000: http://www.hsa.gov.sg/docs/GuidelinesforMinimisingtheRiskofTransmissibleSpongiformEncephalopathyTSEContaminationinHumanMedicinalProducts.pdf 62. Sogal A, Tofe AJ. Risk assessment of bovine spongiform encephalopathy transmission through bone graft material derived from bovine bone used for dental applications. J Periodontol 1999; 70: 1053–1063. 63. Wenz B, Oesch B, Horst M. Analysis of the risk of transmitting bovine spongiform encephalopathy through bone grafts derived from bovine bone. Biomaterials 2001; 22: 1599–1606. 64. CDRH current recommendations on measures to minimize risk of TSE agents in medical devices: http://www.fda.gov/ohrms/dockets/ac/04/briefing/4019B2_9.pdf [accessed November 29 2005]. 65. Guidance for FDA Reviewers and Industry: Medical devices containing materials derived from animal sources (except for in vitro diagnostics): http://www.fda.gov/cdrh/ode/88.html [accessed November 29 2005].

19 Surveillance and epidemiology as tools for evaluating the materials used in medical devices Roselie A. Bright US Food and Drug Administration, Rockville, MD, USA

Introduction The other chapters in this book exemplify, among other things, that effective medical device epidemiology involves many disciplines besides epidemiology. Two of those other disciplines are the sciences of materials and how the human body reacts to materials. When epidemiologists study the effects of medical devices, they should consider whether the material is the responsible attribute.

General considerations that affect the design of epidemiological studies of materials used in medical devices Materials The range of materials used in devices is enormous (see e.g. Figure 19.1). Historically, medical devices were made from natural materials, such as cotton, latex, ceramics, silver, and gold. As technology developed, so did the materials used in devices. Materials scientists have learned to refine and manipulate the properties of materials, such as natural rubber, and produce ever more sophisticated synthetics, such as plastics,

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

274

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

Figure 19.1 Artificial joints, ligaments, and blood vessels, ca. 1985; a selection of devices illustrating an array of synthetic materials used in replacement surgery. Clockwise from top left: stainless steel and ultra-high molecular weight polythene hip replacement; two bone fractures held together by plates from nylon and carbon fibre in epoxy resin; replacement knee ligaments from carbon fibre and polyester braid; replacement blood vessels consisting of Dacron polyester. Reproduced with permission from the Science Museum/Science and Society Library

hydroxyapatite, silicone gel, and various alloys. Many (most of the above) materials were first used for a single purpose, but then were used for more and more applications. A user selecting one type of device may have a choice of materials, each with its own set of properties, such as: Medical gloves – natural rubber or vinyl. Hip prostheses – various combinations of metal on metal, polyethylene on metal, or ceramic on metal (see Chapter 29). Contact lenses – soft hydrophilic polymer or rigid hydrophobic polymer. Hemodialysis membrane material – cellulose acetate, various polymers. Some devices, such as hip prostheses, use a combination of materials. Other examples of combinations include syringes made of both metal and plastic, bulking agents composed

GENERAL CONSIDERATIONS THAT AFFECT THE DESIGN OF EPIDEMIOLOGICAL STUDIES

275

of firm spheres suspended in a gel (the spheres and the gel may each contain more than one material), and a plastic, metal or carbon heart valve with a fabric mesh rim as a base for sutures. The primary goals for materials scientists are good device performance and safety. The continuing evolution of device materials suggests that these goals have not yet been satisfactorily achieved. Surveillance and epidemiology studies of marketed devices contribute to new knowledge of materials performance and safety in conditions of actual use.

Material failures The most obvious materials-related safety problem with devices is outright failure of the material. Table 19.1 shows several examples. The first two involved physical problems related to fracture of a component of the device. For the heart valves, fracture of the strut resulted in improper functioning of the valve[1]. For the lead retention wire, fracture resulted in migration and trauma to critical cardiovascular tissues [2,3]. In the third material failure example, the materials enclosing an implanted electrical part failed to maintain the seal that protects the electrical function from moisture (see Chapter 26) [4]. In the fourth, failure of the material resulted in the inability to effectively clean and sterilize the device, a bronchoscope, thus infecting later patients (see Chapter 17). Table 19.1 Reported adverse events related to device materials, by type of material failure, device involved, and specific description Material failure Fracture

Seal failure

Surface defect

Examples of devices involved Bjork–Shiley heart valve strut fracture

Retention wire fracture of the Accufix atrial ‘J’ lead Case for the implanted signal processor component of cochlear implant

Hole in the sheath of a bronchoscope

Description The valve included a strut that controlled the movement of the flap portion of the valve. A manufacturing deficiency resulted in valve failures [1] If the retention wire breaks, the J-shaped lead can perforate parts of the heart or blood vessels, which may have severe consequences [3], or the fractured retention wire itself can migrate [2] Seals failed, allowing moisture to enter many of the implanted cases and ruin the signal processor. The symptoms experienced by many patients included very loud noise, which may have damaged auditory nerves in the signal pathway [4] (see Chapter 26) Bacteria in the hole evaded the cleaning and disinfection process and infected patients (see Chapter 17)

276

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

Tissue reactions To some extent, both desirable and undesirable responses to materials depend on the use of the device and which body tissues are exposed. Skin is generally the tissue most able to defend against material toxicity, followed by mucosal tissue. Tissues that are generally reachable only by cutting or puncturing are usually more sensitive to foreign materials. Tissue reactivity also varies among individuals and, depending on the material, may be partially related to differences in age, co-morbidity, or race. The bulk quantity of material, or surface area contacting the tissue, may or may not also be a factor. The tissue reaction mechanisms may differ immediately after exposure compared to the reaction after longer exposure. Many types of reactions have been noted (see Table 19.2). These reactions may be mild or severe. Simple irritation has been noted for devices in direct contact with skin, for example, gloves and the external part of the cochlear implant system [6,7]. Different types of allergic reactions have also been noted; the examples in the table feature latex gloves [5,8–10], midline vascular catheters [11,12], and dialysis membranes [13,14]. An especially severe type of reaction that seems to be an allergic reaction but has a different mechanism is anaphylactoid reaction, which has been noted for various blood filters: polyacrylonitrile hemodialysis membranes [15–20], low-density lipoprotein columns [21–24], and bedside leukocyte filters [25]. Other devices, such as prosthetic heart valves, can cause blood clots [26]. Neurologic damage can result from the use of very old cellulose acetate hemodialysis membranes [27–29]. Granulomatous foreign body reactions can result from silicone implants (see Chapter 27) or wear particles from hip prostheses [30]. Mercury toxicity has been claimed to result from mercury amalgam tooth fillings [31,32]. Liver toxicity has been reportedly related to polyurethane breakdown products [33]; polyurethane has been used to cover implanted electric lead wires [34], breast implants [35], and as a potting material in hemodialyzers [33]. Corneal ulcers are mainly related to contact lens wear [36,37].

The roles of surveillance and epidemiology Effective surveillance of medical devices may include collecting adequate information on the materials involved. Other chapters in this book discuss in more detail the importance of data on the exact identity of devices used; this information can be cross-referenced to find the particular materials. Epidemiologic studies of devices can be designed to take advantage of knowledge about the device materials. For example, devices made of different materials can be compared against each other to study performance and safety characteristics. When such studies are designed, a common bias to consider is ‘confounding by indication’. For example, a cemented hip prosthesis offers quick recovery time but is susceptible to cracking and failure over time; it might be appropriate for a person with an expected remaining life of less than 10 years, but inappropriate for a heavy or active person [38];

GENERAL CONSIDERATIONS THAT AFFECT THE DESIGN OF EPIDEMIOLOGICAL STUDIES

277

Table 19.2 Reported adverse events related to device materials, by tissue reaction, device involved, and specific description Tissue reaction Irritant contact dermatitis

Examples of devices involved Gloves

External part of cochlear implant system

Allergic contact dermatitis, allergic contact hypersensitivity, or delayed hypersensitivity Immediate allergic local contact hypersensitivity Immediate allergic general hypersensitivity, anaphylaxis type

Anaphylactoid reaction

Latex gloves

Description Dried, cracked, or split skin is caused by the physical irritation of the glove, glove powder, or moisture [5] Skin irritation can be caused by rubbing, necessitating changing the magnet strength (the magnet holds the external part in place on the skin) or temporary removal of the external part [6,7] Blistering, itching, crusting, or oozing skin lesions appear 1–3 days after the exposure [5,8]

Latex gloves

Skin urticaria (hives) occurs immediately and can progress to a generalized reaction [5,9,10]

Latex

Symptoms include allergic rhinoconjunctivitis and can progress rapidly to asthma, swelling of lips and airways, shortness of breath, shock, and death. This can occur with contact on the skin, a mucous membrane, the lungs (via airborne particles), or surgically exposed tissue [5,9] The hydrogel, meant to soften after insertion and make the catheter more comfortable, has been reported to cause immediate severe allergic reaction [11,12]

Polyurethane– elastomeric hydrogel composite midline venous catheters Dialysis membranes that have been improperly prepared for their first use Polyacrylonitrile (PAN) hemodialysis membranes Leukocyte filters Low-density lipoprotein filtering columns

This is called ‘First Use Syndrome’ and is an anaphylactic reaction to sterilizing substances left on the membrane after production [13,14]

The reaction is clinically similar to anaphylaxis, but is caused by a different mechanism (bradykinin pathway) and is potentiated by angiotensin-converting enzyme (ACE) inhibitor medications [15–25]

(Continued)

278 Table 19.2

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

(Continued)

Tissue reaction

Examples of devices involved

Blood coagulation

Prosthetic heart valves

Neurotoxicity

At least 10 year-old cellulose acetate hemodialysis membranes

Foreign body reaction, granulomatous type

Silicone gel breast implants, either intact or ruptured

Hip prostheses

Mercury toxicity

Mercury amalgam tooth fillings

Liver toxin

Polyurethane breakdown products

Ulcer

Soft hydrophilic polymer contact lenses

Description Unless anticoagulants are taken, the prosthesis can cause blood coagulation, increasing the risk of blood clots [26] Severe neurologic symptoms, including decreased vision and hearing, conjunctivitis, and headache, developed in seven patients within 7–24 hours after being exposed to the cellulose acetate breakdown products in membranes that were over 11 years old [27,29] Granulomatous reactions to the silicone membrane and to the escaped gel have been noted in surgical specimens. The result of the process is scarring around the implant that forms a capsule, which can contract, resulting in hardness and pain (see Chapter 27) Wear can cause particles to be ground off the prosthesis, provoking a granulomatous reaction in the joint, which leads to aseptic loosening [30] There are controversial claims that enough mercury vaporizes or is ground off the surface of the fillings to gradually poison the patient or the patient’s fetus. Reported effects have included impaired neurologic development during childhood, neurotoxicity, kidney dysfunction, and reduced immunocompetence [31,32] Polyurethane foam has been used to cover implanted electric lead wires [34], silicone breast implants [35], and as a potting material in hemodialyzers [33]. It has been shown to be hepatotoxic in humans [33] These reduce oxygen flow to the cornea; if wear is prolonged, an ulcer can form [36,37]

the outcomes that are assessed in a study that compares these two types of hip joints may be affected more by the different patient characteristics than by the hip joints used. Another example is surgical gloves: vinyl is much less allergenic than latex, but latex has much better performance characteristics. Studying latex sensitivity among users of vinyl, compared to users of latex, could be confounded by the users’ actual or feared latex sensitivity. Latex has many interesting features that impact epidemiologic study design, as discussed in the following section.

EXAMPLE: NATURAL RUBBER AND LATEX ALLERGY

279

Figure 19.2 Latex medical or surgical gloves. Photograph by S. Lori Brown, PhD, MPH, used with her permission

Example: natural rubber and latex allergy Reactions to latex in medical devices can be used to illustrate many of the general considerations. Latex is a component in a wide variety of medical [including medical and surgical gloves (see Figure 19.2), condoms, diaphragms, cervical caps, barium enema cuffs, dental dams, airway tubes, anesthesia masks, tympanostomy tubes, douche bulbs, elastic bandages, ostomy bags, manual ventilation bags] and consumer products (including rubber bands, cleaning gloves, balloons, elastic, baby bottle and pacifier nipples, carpet backing, erasers, eye dropper bulbs, hot water bottles, balls, toys, hand grips on equipment). Dry rubber is generally made in a different way from latex and is used for other consumer, automotive, and household products, such as tires and shoe soles [39].

280

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

Natural latex is a complex material. It is currently recovered from rubber trees (most are on rubber tree plantations in Malaysia) by tapping the bark to extract the milkylooking sap (latex) and removing some of the clear serum by centrifugation. The concentrated latex is then treated with ammonia, chlorine, or other chemicals, depending on the desired product. The shape of the final product is created by techniques such as dipping (hand-shaped blocks are dipped in the liquid and pulled back out to create gloves), foaming to form a sponge, or extrusion into thread or elastic. To stabilize the object, it is rinsed and then vulcanized (cured) in ovens. After the vulcanization, the products are rinsed again to leach allergenic proteins. The original sap is a mix that includes various carbohydrates and proteins [39], which vary by the immediate environment of the trees and tree genetics [40]. Natural rubber usage changed in the late 1980s, when it was recognized as a barrier to the transmission of the virus that causes acquired immune deficiency syndrome (AIDS). Until then, gloves had been used for sterility and protection from known infections; after the call for the use of universal precautions to protect all patients and workers as if all of them were infectious, glove use for all medical and some non-medical situations rose sharply [41,42]. The rubber industry changed its production processes, presumably by making them faster, as part of its effort to meet demand [42]. In the 1980s, several case reports of latex allergy among patients and healthcare workers were published [42]. In 1990, the Center for Devices and Radiological Health (CDRH) in the Food and Drug Administration (FDA) received reports of two instances of severe anaphylaxis. Each patient had seemed robust and then suddenly went into anaphylactic shock and died as they were being prepared for barium enemas for imaging their colons. FDA already had many reports on file of allergic reactions associated with barium, but these were the first where the barium solution had not yet been given. The exposures considered as potentially causing these severe reactions were the lubricating jelly and the barium enema cuffs, which were made of latex. Given the sensitivity of rectal mucosal tissues, the recent reports of latex allergy, and the lack of any particular reasons to suspect the lubricating jelly, the barium enema cuffs were implicated [43]. Laboratory studies have shown that the processing of natural rubber results in a mesh that physically, rather than chemically, traps the various (up to 60) allergenic proteins [40,44–46]. Rinsing during production eliminates most of the overall protein, and rinsing after production removes the remaining proteins on the surface [39]. After production, the proteins continue to migrate to the surface of the product, but less rinsing during production results in more surface proteins later [39]. Although the demand for barium enema cuffs had not particularly increased in the late 1980s, all medical latex production had increased and the production quality for barium enema tips might have decreased, resulting in more surface proteins on the surface of the tips. The surface proteins can be carried to skin by direct contact or, more effectively, by a liquid, such as perspiration [47]; they can also be carried into the lungs on particles of powder that become airborne, such as during glove donning [48–50]. The rubber industry has reacted to this recent understanding of the causes of latex allergy by developing and using processing protocols that reduce the amount of protein in their products [39].

EXAMPLE: NATURAL RUBBER AND LATEX ALLERGY

281

Almost any type of tissue can be exposed. Tissue reactions can be more severe for mucous membrane, lung, and wound exposure, and are generally allergic in nature. The allergic reactions are driven by immunoglobulins and the predominant immunoglobulin type varies in different tissues. An individual’s allergic reaction to latex depends on route of exposure, history of exposure, quantity of exposure, and genetic susceptibility. Designers of epidemiologic studies have to consider definitions of exposure, outcome, and the population to be studied. Almost everyone has at least some exposure to latex, since consumer products are common and almost universally used. Therefore, in assessing latex exposure in epidemiologic studies, one needs to consider the magnitude, duration, and route of exposure, as well as the sensitivity of the exposed tissues. This strategy has led to the frequent choice of healthcare workers, who are often exposed on their skin and airways for long periods of time, and patients with frequent surgeries, who are exposed via their airways and surgical incisions, for ‘exposed’ cohorts (see Tables 19.3 and 19.4 for more specifics). Other strategies have been to examine the relative allergenicity of types of latex gloves by comparing, among healthcare workers, use of powdered vs. nonpowdered gloves [51,52], or use of gloves with and without large amounts of surface proteins [52,53], or use of gloves made by different manufacturing processes [39]. Measurement of latex allergy (the outcome in this example) is complicated by the need to consider delayed vs. immediate reaction, the severity of reaction, and whether one will rely on clinical history or some type of ‘objective’ test. Skin prick tests rely on a solution of latex allergens, which may or may not include the particular protein to which the subject is allergic, because of the large number of candidate allergens. The major problem with skin prick tests is that they place the subject at risk of a severe anaphylactic reaction. Another test, which does not place the subject at risk, is the radioallergosorbent test (RAST), where blood is drawn and tested for the presence of reactive immunoglobulins [41]. Like the skin prick test, the RAST test also may fail to test for the exact protein to which the subject is allergic. Two brand name antigen supplies that have been used in published studies include AlaSTAT [54–59] and CAP [60,61]. The lack of accuracy of these methods of measuring latex allergy has been reflected in discordance between the results of any two or all three of the allergy measurement methods used on individuals [62,63]. Discordant results have been documented between self-reports and skin tests [60,64–68], between self-reports and blood tests [58,60,67], and between skin tests and blood tests [60,68]. There have also been discrepancies in skin tests that were performed with different extracts [68] and between blood tests done with different assays [56]. Large numbers of study subjects are desirable because rates of latex allergy tend to be low and the outcome measures are not precise. Since willingness to participate may be correlated with latex sensitivity, high participation rates are important. Unfortunately, high participation rates are difficult to achieve when subjects are asked to undergo risky, uncomfortable, or invasive tests. In attempting to determine the prevalence of latex sensitization in the general population, large study sizes and reasonably high participation rates (see Table 19.3) were obtained in studies of students who were beginning an apprenticeship, blood donors,

282

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

Table 19.3 Prevalence of latex allergy reported in the literature for people who are not occupationally exposed to latex, arranged by population, measure of latex allergy, finding, study size, and participation rate

Population General population Students before starting apprenticeship Blood donors Blood donors in midwinter Blood donors at a non-hospital worksite Blood donors in midsummer Patients Emergency room patients Adult surgery patients Surgery patients with history of drug allergy or atopy Adult surgery patients Adult surgery patients Ambulatory surgery patients Emergency room patients Patients being evaluated for any type of allergy

Latex allergy measure

Estimated prevalence (%)

Study size (n)

Participation rate (%) Reference

Skin test

0.70

769

100

Gautrin et al. [71]

Latex-specific IgE, by AlaSTAT Latex-specific IgE, by AlaSTAT Latex-specific IgE, by AlaSTAT

4

2000

100

Merrett et al. [55]

6.4

1000

100

Ownby et al. [56]

7

5000

100

Merrett et al. [55]

1.6

1027

89

Grzybowski et al. [58] Rueff et al. [72] SanchezFernandez et al. [81]

Latex sensitivity, by questionnaire Rubber intolerance, by questionnaire Skin prick test

5.8

325

75

0.8

1000

100

Skin prick test

4.9

285

66

Latex-specific IgE

7.1

323

75

Latex-specific IgE, by AlaSTAT

7

996

100

Latex-specific IgE, by AlaSTAT Latex-specific IgE, by AlaSTAT

8.2

1027

89

12

200

Not reported

Rueff et al., 2001 [72] Rueff et al. [72] LebenbomMansour et al. [57] Grzybowski et al. [58] Reinheimer et al. [59]

surgical patients, emergency room patients, and patients being evaluated for any type of allergy. The blood donors and patients may have experienced higher latex exposures than the general population, because highly exposed patients, such as those with spina bifida, have been shown to have especially high latex allergy rates (24–60%) [69]. The different

EXAMPLE: NATURAL RUBBER AND LATEX ALLERGY

283

Table 19.4 Prevalence of latex allergy reported in the literature for people who are traditionally occupationally exposed to latex, arranged by population, measure of latex allergy, finding, study size, and participation rate

Population

Documented or presumed frequent latex exposure Medical, surgical, and laboratory staff Operating room nurses Hospital staff Hospital staff Operating room nurses Anesthesiologists and nurse anesthetists Medical, surgical, and laboratory staff Dental school staff

Dental school staff

Dental school staff

Hospital staff

Anesthesiologists and nurse anesthetists Hospital operating room staff Hospital infectious disease and outpatient nurses Hospital nurses, cleaning staff, and technologists

Latex allergy measure

Estimated prevalence (%)

Study size (n)

Participation rate (%)

Reference

Symptoms, by questionnaire

17.2

1165

95

Larese Filon et al. [73]

Symptoms, by questionnaire Latex allergy, by questionnaire Symptoms, by questionnaire Symptoms, by questionnaire Contact dermatitis, by questionnaire

41

248

93

40

504

55

7

1827

55

247

20

168

Not reported Not reported 98

Lagier et al. [64] Kaczmarek et al. [41] Sinha et al. [74] Mace et al. [66] Brown et al. [60]

Contact urticaria, by questionnaire

3

1165

95

Larese Filon et al. [73]

Glove-related dermatitis, by questionnaire Prompt gloverelated urticaria, by questionnaire Glove-related non-hand symptoms, by questionnaire Non-cutaneous symptoms, by questionnaire Type I latex allergy, by questionnaire Type I latex allergy symptoms, by questionnaire Type I latex allergy symptoms, by questionnaire Glove-related urticaria, by questionnaire

22

177

77

Katelaris et al. [75]

9

177

77

Katelaris et al. [75]

33

177

77

Katelaris et al. [75]

8.1

1294

50

Nettis et al. [68]

2

168

98

Brown et al. [60]

18

314

49

Smedley et al. [54]

12

58

43

Smedley et al. [54]

4.7

273

94

Vandenplas et al. [65]

(Continued)

284 Table 19.4

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

(Continued)

Population

Hospital nurses, cleaning staff, and technologists Anesthesiologists and nurse anesthetists Operating room nurses Hospital staff

Latex allergy measure

Estimated prevalence (%)

Study size (n)

Participation rate (%)

Reference

Skin prick test

4.7

273

94

Vandenplas et al. [65]

Skin prick test

11

154

90

Skin prick test

11

197

74

Skin prick test

12

1351

66

Skin prick test

2

202

Skin prick test

2.9

512

Operating room nurses Operating room physicians Hospital anesthesiologists, radiologists, and surgeons Hospital nurses and physicians Hospital laboratory staff Hospital staff

Skin prick test

5.6

Skin prick test

7.4

Skin prick test

9.9

Not reported Not reported 101

Not reported Not reported Not reported Not reported Not reported

Brown et al. [60] Lagier et al. [64] Liss et al. [76] Wrangsjo et al. [61] Turjanmaa [77] Turjanmaa [77] Turjanmaa [77] Arellano et al. [78]

Skin prick test

13.3

Skin prick test

17

Skin prick test

17

Not reported Not reported 224

Anesthesiologists and nurse anesthetists Hospital nurses

Latex-specific IgE, by CAP Latex-specific IgE

8

168

Not reported Not reported Not reported 98

8.9

741

90.6

Operating room and dental staff

Latex-specific IgE, by RAST with CAP Latex-specific IgE, by RAST

2

193

Not reported

Liss et al. [76] Liss et al. [76] Yassin et al. [79] Brown et al. [60] Grzybowski et al. [58] Wrangsjo et al. [61]

3

226

Not reported

Waclawski [80]

Symptoms, by questionnaire Glove-related rhinoconjunctivitis, by questionnaire Glove-related urticaria, by questionnaire

0

227

0

57

Not reported 36

Sussman et al. [51] Saary et al. [52]

4

57

36

Saary et al. [52]

Operating room staff and dental staff Hospital staff

Nurses Documented reduced or minimal latex exposure Hospital staff Dental students

Dental students

(Continued)

EXAMPLE: NATURAL RUBBER AND LATEX ALLERGY

285

Table 19.4 (continued)

Population

Dental school staff

Latex allergy measure

Hospital staff

Glove-related urticaria, by questionnaire Glove-related rhinoconjunctivitis, by questionnaire Skin prick test

Hospital staff

Estimated prevalence (%)

Study size (n)

Participation rate (%)

Reference

11

36

Not reported

Saary et al. [52]

19

36

Not reported

Saary et al. [52]

1

227

Skin prick test

12

1351

Not reported 66

Dental students

Skin prick test

0

57

36

Dental school staff

Skin prick test

8

36

Not reported

Sussman et al. [51] Liss et al. [76] Saary et al. [52] Saary et al. [52]

Dental school staff

studies used different methods to assess latex allergy, such as questionnaires, skin prick tests, and blood tests. The estimates of latex allergy prevalence varied (0.7–12%), with no obvious pattern in the type of people most likely to be sensitive, type of measure, participation rate, or year of study. Table 19.4 summarizes studies of the prevalence of latex sensitization among healthcare workers who are traditionally occupationally exposed to latex. The study sizes and participation rates were highly variable. For the study populations with documented or presumed frequent latex exposure, the questionnaire method resulted in prevalence estimates in the range 3–55%, the skin prick test 2–17%, and the blood test 2–8.9%. For studies of healthcare workers where latex exposure had been systematically and dramatically reduced, the questionnaire method resulted in prevalence estimates of 0–19% and the skin prick test 0–12%. The prevalence rates from the studies of healthcare workers also do not show a clear relationship with type of healthcare worker, latex allergy measure, study size, participation rate, or year of study. One problem with interpreting latex allergy prevalence rates among healthcare workers is that those who are least willing to tolerate latex exposure are most likely to discontinue direct healthcare work, resulting in a presumed underestimate of the impact of latex exposure. A better measure of the burden of latex allergy on healthcare workers is the incidence (rate of new onset) of latex sensitization onset. Data from the Finnish Register of Occupational Diseases showed that the reported incidence rates of latex contact urticaria were 0.2/10 000 per year for the overall working population, 2.2/10 000 per year for working nurses, 3.9/10 000 per year for working physicians, 6.0/10 000 per year for working dentists, and 11.8/10 000 per year for working dental assistants [70]. A study of 122 apprentices in dental hygiene technology throughout their training showed much higher incidence rates of self-reported measures: 249/10 000 per year for skin

286

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

sensitization, 177/10 000 per year for asthma, and 35/10 000 per year for rhinoconjunctivitis [9]; this rate may have been much higher because it was self-reported rather than verified in some way. A study of the incidence of latex allergy onset among Mayo Clinic employees who regularly use gloves showed that it was very low (2.5/10 000 per year) through 1987, then rose to a peak of 27/10 000 per year for 1990–1993; after highallergen gloves were removed from the work areas in early 1994, the incidence in 1994 fell to 12/10 000 per year and thereafter to 0 to 2/10 000 per year [53].

Summary Awide variety of synthetic and natural materials are used in the manufacture of medical devices, which are then used in a diverse patient population. The materials used in medical devices can cause a wide variety of reactions. Surveillance and epidemiologic studies play an important role in characterizing and understanding the roles of these different materials in device performance and safety. Natural rubber sensitivity was used as an example to illustrate a material widely used in medical devices and for which numerous epidemiologic studies have improved our understanding of the prevalence of this problem.

References 1. Blot WJ, Ibrahim MA, Ivey TD et al. Twenty-five-year experience with the Bjork–Shiley convexoconcave heart valve: a continuing clinical concern. Circulation 2005; 111(21): 2850–2857. 2. Kao HL, Wang SS, Chen WJ, Huang SK, Lee YT. Migration of a fractured retention wire in the pulmonary artery from an active fixation atrial lead. Pacing Clin Electrophysiol 1995; 18(10): 1966–1967. 3. Saliba BC, Ardesia RJ, John RM, Venditti FJ, Schoenfeld MH. Predictors of fracture in the Accufix Atrial ‘J’ lead. Am J Cardiol 1997; 80(2): 229–231. 4. Ulatowski TA. Warning letter. US Food and Drug Administration, 2004: http://www.fda.gov/foi/ warning_letters/g5089d.htm [accessed November 2005]. 5. Kurup VP, Fink JN. The spectrum of immunologic sensitization in latex allergy. Allergy 2001; 56: 2–12. 6. James AL, Daniel SJ, Richmond L, Papsin BC. Skin breakdown over cochlear implants: prevention of a magnet site complication. J Otolaryngol 2004; 33(3): 151–154. 7. Benefits and risks of cochlear implants. US Food and Drug Administration, October: http:// www.fda.gov/cdrh/cochlear/riskbenefit.html [accessed November 2005]. 8. Gottlober P, Gall H, Peter RU. Allergic contact dermatitis from natural rubber. Am J Contact Dermatitis 2000; 12(3): 135–138. 9. Archambault S, Malo J-L, Infante-Rivard C, Ghezzo H, Gautrin D. Incidence of sensitization, symptoms, and probable occupational rhinoconjunctivitis and asthma in apprentices starting exposure to latex. J Allergy Clin Immunol 2001; 107: 921–923. 10. Sommer S, Wilkinson SM, Beck MH, English JS et al. Type IV hypersensitivity reactions to natural rubber latex: results of a multicentre study. Br J Dermatol 2002; 146: 114–117.

REFERENCES

287

11. Mermel LA, Parenteau S, Tow SM. The risk of midline catheterization in hospitalized patients. A prospective study. Ann Intern Med 1995; 123(11): 841–844. 12. Blum DY. Untoward events associated with use of midterm i.v. devices. J Intraven Nurs 1995; 18(3): 116–119. 13. Nicholls AJ, Platts MM. Anaphylactoid reactions due to haemodialysis, haemofiltration, or membrane plasma separation. Br Med J (Clin Res Ed) 1982; 285: 1607–1609. 14. Hakim RM, Breillatt J, Lazarus JM, Port FK. Complement activation and hypersensitivity reactions to dialysis membranes. N Engl J Med 1984; 311: 878–882. 15. Benson JS. FDA Safety Alert: Anaphylactoid Reactions Associated with ACE Inhibitors and Dialyzer Membranes. Rockville, MD: US Food and Drug Administration; 1992. 16. Bright RA, Torrence ME, Daley WR, McClellan WM. Preliminary survey of the occurrence of anaphylactoid reactions during haemodialysis. Nephrol Dial Transplant 1999; 14(3): 799–800. 17. Bright RA. Dialysis therapy. N Engl J Med 1998; 339(14): 1004–1005. 18. John B, Anijeet H, Ahmad R. Anaphylactic reaction during haemodialysis on AN69 membrane in a patient receiving angiotensin II receptor antagonist. Nephrol Dial Transplant 2001; 17(5): 943–944. 19. Kammerl M, Schaefer R, Schweda F, Schreiber M, Riegger G, Kramer B. Extracorporeal therapy with AN69 membranes in combination with ACE inhibition causing severe anaphylactoid reactions: still a current problem? Clin Nephrol 2000; 53(6): 486–488. 20. Peces R. Anaphylactoid reaction induced by ACE-I during hemodialysis with a surface-treated AN69 membrane. Nephrol Dial Transpl 2002; 17(10): 1859–1860. 21. Olbricht CJ, Schaumann D, Fischer D. Anaphylactoid reactions, LDL apheresis with dextran sulphate, and ACE inhibitors. Lancet 1992; 340(8824): 908–909. 22. Kroon AA, Mol MJ, Stalenhoef AF. ACE inhibitors and LDL-apheresis with dextran sulphate adsorption. Lancet 1992; 340(8833): 1476. 23. Keller C, Grutzmacher P, Bahr F et al. LDL-apheresis with dextran sulphate and anaphylactoid reactions to ACE inhibitors. Lancet 1993; 341(8836): 60–61. 24. Sano H, Koga Y, Hamasaki K, Furuyama H, Itami N. Anaphylaxis associated with white-cell reduction filter. Lancet 1996; 347(9007): 1053. 25. Zoon KC, Jacobson ED, Woodcock J. Hypotension and bedside leukocyte reduction filters. US Food and Drug Administration Notice, May 1999: http://www.fda.gov/cdrh/safety/hypoblrf. html [accessed November 2005]. 26. FDA Heart Health Online; prosthetic heart valve: http://www.fda.gov/hearthealth/treatments/ medicaldevices/prostheticheartvalve.html [accessed November 2005]. 27. Hutter JC, Kuehnert MJ, Wallis RR, Lucas AD et al. Acute onset of decreased vision and hearing traced to hemodialysis treatment with aged dialyzers. JAMA 2000; 283(16): 2128–2134. 28. Hutter JC, Long MC, Luu HM, Schroeder LW. Prediction of the shelf life of cellulose acetate hemodialyzers by Monte Carlo simulation. American Society for Artificial Internal Organs Journal 2001; 47(5): 522–527. 29. Kessler LG, Jarvis WR. Important information relating to cellulose acetate dialyzers. US Food and Drug Administration Notice, December 1996: http://www.fda.gov/cdrh/dialyzer.html [accessed November 2005]. 30. Hirakawa K, Jacobs JJ, Urban R, Saito T. Mechanisms of failure of total hip replacements: lessons learned from retrieval studies. Clin Orthop Relat Res 2004; 420: 10–17. 31. Eley BM. The future of dental amalgam: a review of the literature. Part 6: possible harmful effects of mercury from dental amalgam. Br Dent J 1997; 182(12): 455–459.

288

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

32. Counter SS, Buchanan LH. Mercury exposure in children: a review. Toxicol Appl Pharmacol 2004; 198(2): 209–230. 33. Do Luu HM, Hutter JC. Pharmacokinetic modeling of 4,40 -methylenedianiline released from re-used polyurethane dialyzer potting materials. J Biomed Mater Res 2000; 53(3): 276–286. 34. Santerre JP, Woodhouse K, Laroche G, Labow RS. Understanding the biodegradation of polyurethanes: from classical implants to tissue engineering materials. Biomaterials 2005; 26(35): 7457–7470. 35. Luu HM, Hutter JC, Bushar HF. A physiologically based pharmacokinetic model for 2,4-toluenediamine leached from polyurethane foam-covered breast implants. Environ Health Perspect 1998; 106(7): 393–400. 36. Poggio EC, Glynn RJ, Schein OD et al. The incidence of ulcerative keratitis among users of daily-wear and extended-wear soft contact lenses. N Engl J Med 1989; 321: 779–783. 37. Schein OD, Glynn RJ, Poggio EC, Seddon JM, Kenyon KR. The relative risk of ulcerative keratitis among users of daily-wear and extended-wear soft contact lenses. A case-control study. N Engl J Med 1989; 321: 773–778. 38. American Academy of Orthopaedic Surgeons. Hip implants: The URL is http://orthoinfo. aaos.org/fact/thr_report.cfm?Thread_ID¼271&topcategory¼Hip. [accessed November 2006]. 39. Yip E, Cacioli P. The manufacture of gloves from natural rubber latex. J Allergy Clin Immunol 2002; 110: S3–14. 40. Sussman GL, Beezhold DH, Kurup VP. Allergens and natural rubber proteins. J Allergy Clin Immunol 2002; 110: S33–39. 41. Kaczmarek RG, Silverman BG, Gross TP et al. Prevalence of latex-specific IgE antibodies in hospital personnel. Ann Allergy Asthma Immunol 1996; 76: 51–56. 42. Ownby DR. A history of latex allergy. J Allergy Clin Immunol 2002; 110: S27–32. 43. Tomazic VJ, Withrow TJ, Fisher BR, Dillard SF. Short analytical review: latex-associated allergies and anaphylactic reactions. Clin Immunol Immunopathol 1992; 64: 89–97. 44. Kurup VP, Alenius H, Kelly KJ, Castillo L, Fink JN. A two-dimensional electrophoretic analysis of latex peptides reacting with IgE and IgG antibodies from patients with latex allergy. Int Arch Allergy Immunol 1996; 109(1): 58–67. 45. Posch A, Chen Z, Wheeler C, Dunn MJ et al. Characterization and identification of latex allergens by two-dimensional electrophoresis and protein microsequencing. J Allergy Clin Immunol 1997; 99: 385–394. 46. Alenius H, Kurup V, Kelly K, Palosuo T et al. Latex allergy: frequent occurrence of IgE antibodies to a cluster of 11 latex proteins in patients with spina bifida and histories of anaphylaxis. J Lab Clin Med 1994; 123(5): 712–720. 47. Turjanmaa K, Laurila K, Makinen-Kiljunen S, Reunala T. Allergenic properties of 19 brands of latex gloves. Contact Dermatitis 1988; 19: 362–367. 48. Tomazic VJ, Shampaine EL, Lamanna A, Withrow TJ et al. Cornstarch powder on latex products is an allergen carrier. J Allergy Clin Immunol 1994; 93(4): 751–758. 49. Korniewicz DM, Kelly KJ. Barrier protection and latex allergy associated with surgical gloves. Association of Operating Room Nurses Journal [LAW5] 1995; 61(6): 1037–1044. 50. Charous B, Schuenemann P, Swanson M. Passive dispersion of latex aeroallergens in a healthcare facility. Ann Allergy Asthma Immunol 2000; 85: 285–290. 51. Sussman GL, Liss GM, Deal K et al. Incidence of latex sensitization among latex glove users. J Allergy Clin Immunol 1998; 101: 171–178. 52. Saary MJ, Kanani A, Alghadeer H, Holness DL, Tarlo SM. Changes in rates of natural rubber latex sensitivity among dental school students and staff members after changes in latex gloves. J Allergy Clin Immunol 2002; 109: 131–135.

REFERENCES

289

53. Hunt LW, Kelkar P, Reed CE, Yunginger JW. Management of occupational allergy to natural rubber latex in a medical center: the importance of quantitative latex allergen measurement and objective follow-up. J Allergy Clin Immunol 2002; 110: S96–106. 54. Smedley J, Jury A, Bendall H, Frew A, Coggon D. Prevalence and risk factors for latex allergy: a cross-sectional study in a United Kingdom hospital. Occup Environ Med 1999; 56: 833–836. 55. Merrett TG, Merrett J, Kekwick R. The prevalence of immunoglobulin E antibodies to the proteins of rubber (Hevea brasiliensis) latex and grass (Phleum pratense) pollen in sera of British blood donors. Clin Exp Allergy 1999; 29: 1572–1578. 56. Ownby DR, Ownby HE, McCullough J, Shafer AW. The prevalence of anti-latex IgE antibodies in 1000 volunteer blood donors. J Allergy Clin Immunol 1996; 97: 1188–1192. 57. Lebenbom-Mansour MH, Oesterle JR, Ownby DR et al. The incidence of latex sensitivity in ambulatory surgical patients: a correlation of historical factors with positive serum immunoglobulin E levels. Anesth Analg 1997; 85: 44–49. 58. Grzybowski M, Ownby DR, Peyser PA, Johnson CC, Schork MA. The prevalence of anti-latex IgE antibodies among registered nurses. J Allergy Clin Immunol 1996; 98: 535–544. 59. Reinheimer G, Ownby DR. Prevalence of latex-specific IgE antibodies in patients being evaluated for allergy. Ann Allergy Asthma Immunol 1995; 74: 184–187. 60. Brown RH, Schauble JF, Hamilton RG. Prevalence of latex allergy among anesthesiologists: identification of sensitized but asymptomatic individuals. Anesthesiology 1998; 89: 292–299. 61. Wrangsjo K, Osterman K, Van Hage-Hamsten M. Glove-related skin symptoms among operating theatre and dental care unit personnel (II): clinical examination, tests, and laboratory findings indicating latex allergy. Contact Dermatitis 1994; 30: 139–143. 62. Ebo DG, Stevens WJ. IgE-mediated natural rubber latex allergy: an update. Acta Clin Belg 2002; 57: 58–70. 63. Hamilton RG. Diagnosis of natural rubber latex allergy. Methods 2002; 27: 22–31. 64. Lagier F, Vervloet D, Lhermet I, Poyen D, Charpin D. Prevalence of latex allergy in operating room nurses. J Allergy Clin Immunol 1992; 90: 319–322. 65. Vandenplas O, Delwiche JP, Evrard G et al. Prevalence of occupational asthma due to latex among hospital personnel. Am J Resp Crit Care Med 1995; 151: 54–60. 66. Mace SR, Sussman GL, Liss G et al. Latex allergy in operating room nurses. Ann Allergy Asthma Immunol 1998; 80: 252–256. 67. Chaiear N, Foulds I, Burge PS. Prevalence and risk factors for latex allergy. Occup Environ Med 2000; 57(7): 501. 68. Nettis E, Assennato G, Ferrannini A, Tursi A. Type I allergy to natural rubber latex and type IV allergy to rubber chemicals in healthcare workers with glove-related skin symptoms. Clin Exp Allergy 2002; 32: 441–447. 69. Sapan N, Nacarkucuk E, Canitez Y, Saglam H. Evaluation of the need for routine preoperative latex allergy tests in children. Pediatr Int 2002; 44: 157–162. 70. Jolanki R, Estlander T, Alanko K, Savela A, Kanerva L. Incidence rates of occupational contact urticaria caused by natural rubber latex. Contact Dermatitis 1999; 40: 329–331. 71. Gautrin D, Infante-Rivard C, Dao TV, Magnan-Larose M et al. Specific IgE-dependent sensitization, atopy, and bronchial hyperresponsiveness in apprentices starting exposure to protein-derived agents. Am J Respir Crit Care Med 1997; 155: 1841–1847. 72. Rueff F, Kienitz A, Schopf P et al. Frequency of natural rubber latex allergy in adults is increased after multiple operative procedures. Allergy 2001; 56: 889–894. 73. Larese Filon F, Bosco A, Fiorito A, Negro C, Barbina P. Latex symptoms and sensitisation in healthcare workers. Int Arch Occup Environ Health 2001; 74: 219–223.

290

CH19

SURVEILLANCE AND EPIDEMIOLOGY AS TOOLS

74. Sinha A, Harrison PV. Latex glove allergy among hospital employees: a study in the north-west of England. Occup Med 1998; 48: 405–410. 75. Katelaris CH, Widmer RP, Lazarus RM. Prevalence of latex allergy in a dental school. Med J Aust 1996; 164: 711–714. 76. Liss GM, Sussman GL, Deal K et al. Latex allergy: epidemiological study of 1351 hospital workers. Occup Environ Med 1997; 54: 335–342. 77. Turjanmaa K. Incidence of immediate allergy to latex gloves in hospital personnel. Contact Dermatitis 1987; 17: 270–275. 78. Arellano R, Bradley J, Sussman G. Prevalence of latex sensitization among hospital physicians occupationally exposed to latex gloves. Anesthesiology 1992; 77: 905–908. 79. Yassin MS, Lierl MB, Fischer TJ, O’Brien K et al. Latex allergy in hospital employees. Ann Allergy 1994; 72: 245–249. 80. Waclawski E. Prevalence and risk factors for latex allergy: a cross sectional study in a United Kingdom hospital. Occup Environ Med 2000; 57(7): 501. 81. Sanchez-Fernandez C, Quirce S, Sanchez-Cano M. Preoperative screening for general anesthesia. Allergy 1998; 53(5): 542–543. 82. Reinheimer .. et al. 1995

20 Exploring methods for analyzing surveillance reports on electromagnetic interference with medical devices S. Lori Brown US Food and Drug Administration, Bothell, WA, USA

Nilsa Loyo-Berrı´os, Miche`le G. Bonhomme, Donald M. Witters, Nancy A. Pressly, and Jeffrey L. Silberberg US Food and Drug Administration, Rockville, MD, USA

Surveillance of adverse medical device events (AMDEs) is an essential component of assuring the postmarket safety of medical devices. Timely identification and assessment of problems allows the Food and Drug Administration (FDA), manufacturers, and healthcare providers to take remedial actions that prevent future problems and decrease the severity of events should adverse events occur. The written reports of adverse events involving FDA-approved medical devices, which FDA receives through the Medical Device Reporting (MDR) system and the agency’s MedWatch program, are a key source of AMDE surveillance data. These reports are entered into the Manufacturer and User Facility Device Experience database (MAUDE). We explored different methodologies for analyzing surveillance reports on electromagnetic interference (EMI) with medical devices. Two FDA studies of EMI adverse event reports that were submitted through the MDR system, conducted at different times and using different methodologies, are presented. These analyses illustrate how the MAUDE database and its adverse event Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

292

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

problem codes can be used in the surveillance of AMDEs. The difficulties of assessing the incidence and severity of EMI with medical products, as well as determining risk factors for exposed patients, are discussed.

Electromagnetic interference with medical devices Medical device EMI is the disruption of the normal function of electrically powered medical devices by conducted or radiated electromagnetic energy or by electrostatic discharge (ESD). EMI can affect the safety or effectiveness of active, electrically powered medical devices in ways that can pose significant risks to patients and medical device users, and it has been reported to result in patient injury and death[1,2]. These events may occur in the hospital, the home, the workplace, the shopping mall, the airport, or other public places. Reports of EMI-related medical device problems are scrutinized by medical device manufacturers and regulators. Many manufacturers point to the paucity of confirmed reports and question the extent of the problem. Regulators and those involved in writing standards for electromagnetic compatibility (EMC) believe that EMI problems are underreported [3], that the reported cases are indicative of more widespread issues, and strive to prevent future EMI problems by drafting and revising standards and recommendations for EMC [4–6]. Electromagnetic compatibility (EMC) is essentially the opposite of EMI: EMC is the condition in which the electromagnetic emissions from a device are limited so that they do not affect other active medical devices or other equipment in the vicinity, and the device or equipment is immune to the electromagnetic emissions, also known as electromagnetic disturbance (EMD), expected in its use environment. Understanding of medical device EMI and EMC has increased over the past three decades; however, despite measures to ameliorate EMI with medical devices and its consequences, new adverse event reports of EMI with medical devices continue to be reported.

The MAUDE database The MAUDE database consists of reports of device-related deaths, serious injuries and device malfunctions – the three major types of adverse medical events reported to the FDA. The database includes both voluntary and mandatory reports. The section on ‘Postmarket Surveillance’ of Chapter 2 in this book gives a brief historical perspective on voluntary and mandatory medical device reporting, describes the reporting requirements for manufacturers, user facilities and importers, and defines reportable events. Injuries or illness are defined as serious injuries or illnesses that are life-threatening, result in permanent impairment of a body function or permanent damage to a body structure, or necessitate medical or surgical intervention to preclude permanent impairment of a body function or permanent damage to a body structure [7]. Malfunctions are defined as a failure of a medical device to meet its performance

ADVERSE EVENT REPORTS TO FDA

293

specifications or otherwise perform as intended. The sources of AMDE reports are mandatory reports from device manufacturers, user facilities, and importers, and voluntary reports from healthcare professionals, device users, and others. User facilities include, among others, hospitals, nursing homes, ambulatory surgical facilities, and diagnostic and treatment facilities. As noted in Chapter 4, the FDA Modernization Act of 1997 repealed mandatory reporting by device distributors and suggested that FDA develop a network from a ‘. . . subset of user facilities that constitutes a representative profile of user reports . . .’. In response, the Medical Product Surveillance Network (MedSun; see Chapter 5) was established in 2002 to obtain a subset of user facilities, which join the network voluntarily. The reports submitted through MedSun are included in MAUDE. Most of the AMDE reports FDA receives are sent by medical device manufacturers, based on reports they receive from customers, which may include healthcare providers, patients and others. Manufacturers are required to report deaths, serious injuries and malfunctions to FDA. User facilities are required to report deaths and serious injuries to the manufacturer and deaths to FDA. If the manufacturer is not known, the user facility must send the report to FDA [8]. Relatively few AMDEs are reported directly to FDA by healthcare providers or patients. The MAUDE database includes information about the patient, the event or problem, the device and the outcome of the event, the reporter, and the user facility where the event occurred (see Table 20.1). For additional details regarding the data collection form, the review of reports and actions taken by FDA, and the limitations of the FDA MDR system, please see Chapter 2. Modified versions of the MAUDE database, MDR database [formerly called the Device Experience Network (DEN) database], and MedWatch database are available to the public [8,9].

Adverse event reports to FDA (December 1984–October 1995): EMI with cardiovascular devices In 1995, the Center for Devices and Radiological Health (CDRH) EMC Work Group performed a review of the DEN database for adverse events related to EMI [10]. At the time this project was initiated, the regulation implementing user facility reporting was not yet final, thus the search of the database was restricted to manufacturer reports (the MDR database) [11]. Due to the large volume of reports in the database, the initial review was limited to the cardiovascular device panel, which includes implanted pacemakers (see Figure 20.1) and defibrillators. These devices were examined first because of an increase in the number of reports of EMI related to cellular telephones, and because past experience suggested that these critical devices were potentially more susceptible. Data were gathered during December 1984–October 1995. The group developed a plan for identifying EMI-related reports in the database, then reviewed and separated out those reports that were identified as being EMI-related. Between December 1984 and October 1995, reports entered into the database were not routinely coded as being EMI-related, so a method was needed to identify EMI reports.

294

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

Table 20.1 Data elements of adverse event reportsa Patient information Patient name or other identifier Patient’s age at the time of the event Gender Weight Event information Outcome attributed to adverse event (death, serious injury, or malfunction) Date of event Date of report by initialing reporter Event description (nature of event, patient follow-up, required treatment, environmental conditions) Relevant tests (including dates and laboratory data) Other relevant history including pre-existing medical conditions Device information Brand name Type of device Manufacturer name and address Operator of the device (health professional, patient, lay user, other) Expiration date Model number, catalog number or other identifying number Date of device implantation Date of device explantation Whether device was available for evaluation or returned to the manufacturer (if so, date returned) Concomitant medical products and therapy dates (excluding therapies used to treat the event) Initial reporter information Name, address, and telephone number of person who reported event to the user facility, manufacturer or distributor Whether initial reporter was a health professional Occupation Whether initial reporter sent copy of report to FDA User facility information Whether reporter is a user facility User facility number User facility address Contact person and contact person’s telephone number Date user facility became aware of event Type of report (initial or follow-up) Date of user facility report Approximate age of device Event problem codes Whether a report was sent to FDA and date sent Location where event occurred Whether reports was sent to manufacturer and date sent Manufacturer’s name and address a

These variables are not all included in the public database.

ADVERSE EVENT REPORTS TO FDA

295

Figure 20.1 St George’s implantable axilla cardiac pacemaker, 1967. The cardiac pacemaker was designed to regulate the beating of the heart by producing small electrical signals. Reproduced with permission from the Science Museum/Science and Society Library

A list of terms related to EMI was developed and AMDE text fields were searched for these terms to identify potential EMI-related events. The terms in the text search were: EMI, EMC, electromagnetic, ESD, electrostatic, static, surge, glitch, AC, power line, radio frequency, radiofrequency, radio wave, radiowave, susceptibility, immunity, interact, cellular phone, walkie talkie, mobile phone, false alarm, RFI, unexplained, noise, interfere, lock-up, and locks-up. An EMI categorization scheme for assessing the reports was developed, tested, and revised, using a random sample of reports. The final scheme included four categories: (a) the likelihood that the adverse event resulted from EMI;

296

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

(b) the most likely source of the EMD; (c) the patient outcome of the EMI-caused adverse event; and (d) the device outcome. Out of approximately 150 000 reports in the cardiovascular panel at the time, a total of 1749 reports were reviewed, based on the results of the text search. Of these adverse event reports, 576 (33%) were categorized as being highly probable, probable, or possibly related to EMI. Of the 576 reports that were categorized as EMI-related, 456 (79%) involved pacemakers or implantable cardiac defibrillators (ICDs). The most frequent sources of EMDs reported to interfere with the pacemakers and ICDs (67% of 456 reports) were an activated electrosurgical unit (ESU) or external defibrillator in the vicinity of the pacemaker or ICD. In most cases, this left the device inoperable, with operation not restorable. Exposure of the implanted pacemakers and ICDs to the relatively high energy levels emitted by ESU devices and other electrical devices falls into a special category, referred to as ‘electrical overstress’, which is distinguished from typical EMI. This distinction is also made in international standards for active implantable medical devices, including such devices as implanted cardiac pacemakers and ICDs. Electrical overstress from ESUs, external defibrillators, and other electrical energy sources is distinguished from EMI and has led to specific testing and requirements to mitigate the effects (e.g. ISO 14708-1, EN 45502-2-1, and EN 45502-2-2) [12–14]. Another source of EMDs that caused interference with pacemakers and ICDs was permanent magnets (10% of 456 reports). In 16% of the 456 reports involving pacemakers or ICDs, the source was unknown. The EMDs for the remaining 7% of ICDs and pacemaker came from various sources. Most of the pacemaker or ICD reports were coded with a patient outcome of ‘intervention required to prevent adverse outcome’. The other medical devices most frequently affected by EMI included external defibrillators (8% of 576) and cardiac monitors (5% of 576). The remaining reports (8% of 576) mentioned various other medical devices, including devices such as pulse oximeters, balloon pump controllers, pacemaker programmers, and blood pumps. In many of these cases, the source of the EMD was not known. The review of the cardiovascular device panel was an update to previous examinations of the adverse event report database for EMI-related reports. The findings from that study were useful in helping to define problem codes that would subsequently be used in the new FDA adverse event report database that was implemented once the user facility reporting regulation took effect [7]. Problem codes are used to code device or patient problems reported to occur with medical devices. Additional examinations and reviews of the older event report database (MDR database) and the newer MAUDE database have been performed periodically by the CDRH EMC Work Group over the years, looking at general trends in EMI reports as well as looking for reports in targeted areas (e.g. EMI with ventilators [15] and from electronic security systems [16,17]). Reviews of the adverse event databases and reports provide valuable information about EMI incidents involving a number of vital medical devices and have helped lead to the development of testing methods [18] and improvements in the EMC requirements of national and international consensus standards.

297

ADVERSE EVENTS REPORTED TO FDA

Adverse events reported to FDA (January 1994–March 2005): a recent epidemiological analysis Methodology This epidemiological analysis includes reports on suspected EMI cases with any medical device; reports from user facilities, manufacturers, voluntary sources, and distributors were included in the study. The unit of analysis was the adverse event report and a suspected EMI was defined as any report in which the event description suggested the medical device’s performancewas disrupted or degraded by another medical device or other equipment in the environment in which the medical device was used. This can occur when there is electromagnetic energy or EMD emitted from a source in the vicinity of the active medical device. This study was used to evaluate problem codes in MAUDE as a tool for EMI surveillance. A secondary objective was to describe and characterize the reports identified by the search criteria. Table 20.2 presents the problem codes used to identify adverse event reports in MAUDE that could be related to EMI. Reports entered in the database during January 1994–March 2005 were included in the analysis. The reports selected by the search criteria were downloaded into an Excel database for data cleaning and management. Then a review of the adverse event description in each report was performed to determine whether the report was possibly related to EMI and to identify the affected device and the EMD source, if given. We excluded reports that were clearly, unambiguously, not related to EMI. For example, using problem code 1194 we retrieved the report ‘during the stent deployment, the stent fell off the stent delivery system’, but reviewing this report indicated that it was not EMI-related. Eliminating this type of report left reports that were suspected to be related and were classified as possibly related to EMI. No attempt was made to further classify these remaining reports as to the likelihood that they were EMI-related, as was done in the previous analysis.

Statistical procedures Descriptive statistical analysis was conducted using Stata 9 [19] and trends were evaluated using the statistical software R 2.2.0 [20]. The appropriateness of the search Table 20.2 Problem codes used for the adverse event reports search in MAUDE, January 1994– March 2004 Description

Code

Electromagnetic interference (EMI) Interference Interference with monitoring device Interference with pacemaker Radiofrequency interference (RFI) Electromagnetic interference (EMI), compatibility/incompatibility

1194 1327 1328 1329 2314 2345

298

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

criteria was evaluated by determining the number of suspected EMI cases identified and the number of reports that were not EMI-related. Additionally, for those that were not EMI-related, the coding was evaluated further and an estimate of the proportion of reports that were miscoded is presented. Descriptive statistics are presented for the suspected EMI cases on report source, patient’s age and gender, event location and type of event. EMI-affected devices were classified following the definition for device type (implanted, monitoring, therapeutic, surgical, diagnostic, and prosthetic) from the Code of Federal Regulations (CFR) [21]. Medical EMD sources were also classified by device type and non-medical sources were classified as products related to ‘Conducted Electromagnetic Energy (Alternating Current)’ or ‘Radiated Electromagnetic Energy’ (wireless technology).

Results Appropriateness of search criteria A total of 550 reports meeting our initial criteria were retrieved. Figure 20.2 presents the distribution of the reports and how they were classified. After reading all reports, 116 (21%) were determined either to be not related to EMI (13%) or the information needed to even tentatively ascribe the adverse event to EMI was inadequate (8%). Twenty-nine (5%) reports involved a magnetic resonance imaging (MRI) device and a metal object or another medical device. There are several concerns besides EMI that relate to adverse events among medical devices and MRI devices, such as tissue heating near conductive leads, torque induced onto ferrous metal implants, objects drawn into the MRI bore area, and imaging artifacts. In fact, 14 of the MRI-related reports (3%) were accidents that involved burns or injuries to the patient or health professional, and 15 (3%) were considered possible EMI cases, as the exposure to the MRI magnet resulted in a malfunction of the exposed medical device. The MRI reports were analyzed separately because of the unique electromagnetic environment inside the MRI machine and its suite, and are further described in the section on Description of MRI-related adverse events, below. After eliminating the 116 unrelated reports and the MRI-related reports, 405 (74% of 550) reports describing suspected or possible EMI incidents remained. These included a small number of reports (5% of 405) about interference from an electronic component of the medical device that degraded its own performance (Figure 20.2). The 405 reports also included cases of pacemakers affected during electrosurgery (9% of 405) or by external defibrillation (10% of 405) (Figure 20.2). As stated earlier, the pacemaker ESU and AED events are instances of electrical overstress that should be distinguished from EMI. These results highlight the importance of reviewing the narrative event description, if it is provided in the report, to assess whether the event is an actual EMI case or something else. Some of the reports selected using the initial search criteria were not suspected EMI cases. Most of the codes used in the search are specific for EMI except for code 1327 (Interference), which can be used for any type of interference, such as cross-reactions

Inconclusive 46 (8.4%) MRI accidents 14 (2.5%)

MRI-related 29 (5.3%)

EMI–MRI 15 (2.7%)

Device components 19 (4.7%)*

Pacemakers/ESU 38 (9.4%) *

Suspected EMI 405 (73.6%)

Pacemakers/AED 41 (10.1%)*

Figure 20.2 Distribution of reports selected with the initial search criteria. *Percentages were calculated as a proportion of 405. All other percentages were calculated as a proportion of the 550 selected reports

Not EMI 70 (12.7%)

116 (21.1%)

Total Selected Reports 550

300

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

MAUDE EMI coded as something else (?)

Coded incorrectly (10.2%)

Other EMI codes* 56 (10.2%)

Search criteria

n = 550

Not EMI 70 (12.7%) Code 1327† 14 (2.5%)

Figure 20.3 Distribution of codes in the reports that were not EMI-related. *Coded with any of EMI codes 1194, 1328, 1329, 1378, 2314, 2345; {Correctly coded as some other type of interference 1327 (Interference).

with reagents. Therefore, the reports retrieved by the initial search criteria, that were determined to be unrelated to EMI based on the review of the event narrative description, may have resulted from other types of interference. Figure 20.3 presents the results of the review of the event description to confirm EMI-relatedness. Further evaluation of these reports showed that 2.5% of the 550 reports were correctly coded, as they involved some type of interference, but not EMI. However, 10% (of 550) were incorrectly coded with at least one EMI-specific code. These are two examples of miscoded reports: ‘The coil was stretched during the procedure.’ Coded as 1194 (Electro-magnetic interference). ‘Recent data from customer and internal investigations have confirmed a positive interference from specific samples when using the suspect reagent on the analyzers.’ Coded as 1378 (Magnetic interference).

Finally, EMI cases could have been coded as another problem that was not related to EMI. Because EMI-specific problem codes were used in this study, an estimate of the number of EMI cases coded as another problem could not be obtained. The reports included in the analysis do not represent the entire universe of EMI cases.

Description of EMI-related reports Figure 20.4 shows the monthly frequency of EMI reports received during January 1994– March 2005. The number of EMI reports has increased since 1996. The maximum number of reports received was 12/month and the mean was 2.9/month. Because of the large volume of reports for all devices received by FDA, entry of death and serious injury reports are a priority.

301

12

ADVERSE EVENTS REPORTED TO FDA

6 0

2

4

No. reports

8

10

Smoothing degree = 1/2

1Jan94

1Jan96

1Jan98

1Jan2000

1Jan2002

1Jan2004

Calendar Time

Figure 20.4 Monthly numbers of reports on suspected EMI cases received at FDA and entered into the system, January 1994–March 2005. *Decreasing trend during the last 2 years of follow-up may be related in part to the use of summary reporting

The observed increasing trend may be due to a number of causes. First, the increase in EMI-related AMDEs may have resulted from increased use of EMD sources, such as wireless communications (cell phones, PDAs, wireless internet access, etc.), which has become more common in recent years. Second, increased public awareness and changes in reporting regulations may have improved reporting. In fact, the final MDR regulation requiring user facilities to report was published in December 1995 and became effective in July 1996, when the increasing trend began (Figure 20.4). These may partially explain improved reporting and the increase in the number of reports received. Additionally, FDA conducted a campaign targeted at healthcare professionals and engaged in other activities to increase awareness, recognition, and reporting of AMDEs related to EMI. Despite these efforts, underreporting continues to be a limitation of postmarket surveillance for adverse events related to medical devices. Table 20.3 characterizes the EMI-related reports. Most of the reports were manufacturers’ reports (71%) and voluntary reports (21%). Reports from user facilities and distributors were less common (7% and 1%, respectively). Patient age was not documented for over one-third of the reports; among reports in which the age of the patient was known, over half of the EMI adverse events occurred among patients over the age of 40. Men and women were represented in similar proportions, and gender was undocumented for 26% of the reports. Most of the EMI-related adverse events occurred in the hospital or other medical setting (47%), or in a public place (22%). One might expect most problems to occur in hospitals, because this is where most medical devices are concentrated. A few reportedly occurred at home, at the workplace, or in a helicopter or

302

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

Table 20.3 EMI-related adverse events reported to FDA, January 1994–March 2005 (n ¼ 405) Frequency Variable Report source Manufacturer User facility Voluntary reports Distributors Patient age ¼ 20 21–40 41–60 61þ Unknown Gender Female Male Unknown Event location Hospital or other medical setting Public place Home Workplace Helicopter/ambulance Unknown Event type Death Injury Malfunction Other Not specified

(n)

(%)

289 27 86 3 405

71.4 6.7 21.2 0.7 100.0

11 32 71 142 149 405

2.7 7.9 17.5 35.1 36.8 100.0

142 156 107 405

35.1 38.5 26.4 100.0

189 89 13 9 4 101 405

46.7 22.0 3.2 2.2 1.0 24.9 100.0

6 170 167 43 19 405

1.5 42.0 41.2 10.6 4.7 100.0

ambulance while the patient was traveling to a medical center. The location of the incident was unspecified in 25% of the reports. Adverse event reports are categorized as malfunction, injury, death, or other. The type of adverse event was unknown for a small percentage of reports (5%). The majority of event reports described a patient injury (42%) or a device malfunction (41%) or a patient

ADVERSE EVENTS REPORTED TO FDA

303

injury (42%) related to EMI. In 43 (11%) reports the event type was classified as ‘other’ (that is, not a death, injury or malfunction). In most reports classified as ‘other,’ the EMI resulted in the inappropriate delivery of therapy (72%). For example, EMI reportedly caused implanted nerve stimulators to decrease stimulation or stop stimulation altogether (30%). However, the most common example of inappropriate therapy (42%) caused by EMI involved overstimulation by implanted stimulator devices (e.g. spinal cord stimulators for pain, intracerebral/subcortical stimulators). In 12% of the reports classified as ‘other’ the event type could not be determined, and in 16% the event type included problems with a hearing aid, cardiac monitors, an electrosurgical unit, a motorized-wheel chair, or a telemetry system. Electromagnetic interference can have serious adverse consequences. Some automatic implantable cardioverter defibrillators (ICDs) affected by EMI were reported to have delivered shocks when they were not needed, or to have failed to provide a shock when one was needed. In one instance, the patient had to be resuscitated after an ICD failed to shock the patient. In another case, a pacemaker was reported to have stopped during surgery. Although the patient suffered no ill-effects, this device malfunction could have resulted in serious patient injury or death. This event may have been due to EMI, electrical overstress alone, or a combination of these. Table 20.4 shows examples of reported incidents of EMI. These reports demonstrate that a variety of medical devices are potentially affected by EMI and that the source of EMD may also vary. The effect of EMI on patients may be benign or life-threatening. Six deaths were reported to be related to EMI with a medical device. In two of these death reports, the medical device affected was a pacemaker and the source of EMD was either an automatic external defibrillator or an electrocautery device. As mentioned above, this is usually electrical overstress rather than EMI. In two other deaths the device affected was a telemetry system and the EMD was reported to have been picked up by the telemetry unit’s antenna system, interfering with patient monitoring. In the remaining two deaths, the affected medical devices were an automated external defibrillator and a detector and alarm for arrhythmia. The six deaths reported are described in detail in Table 20.5. Table 20.6 presents the types of medical devices most frequently affected by EMI. Implanted devices (72%) were most frequently affected, followed by monitoring devices (12%) and therapeutic devices (7%). Interference with diagnostic, surgical, and prosthetic devices was infrequently reported. Table 20.7 lists the specific devices that were most frequently affected by electromagnetic interference: pacemakers and implantable spinal cord/nerve stimulators and intracerebral/subcortical stimulators were the leading affected devices. EMD sources are described in Table 20.8. A potential source of EMD was identified in 82% of the reported incidents. Of the total reports, another medical device was most frequently reported as the source of EMD (45%). The most common medical devices included in the reports as EMD sources were therapeutic devices (19%) and surgical devices (15%); diagnostic and implanted devices accounted for 6% of the reports. Of the EMI-related incidents attributed to non-medical devices, 28% were sources emitting radiated electromagnetic energy and 3% were sources related to conducted

304

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

Table 20.4 Examples of suspected EMI-related adverse events coded as EMI and reported to FDA, January 1994–March 2005 Affected medical device

EMD source

Details

Cardiac pacemaker

Electrosurgery generator

Cardiac pacemaker

Automatic external defibrillator

Cardiac pacemaker

Cell phones/ in-store satellite system

Cardioverterdefibrillator

Electronic article surveillance system

Spinal cord stimulator for pain relief

Anti-theft system in library

During a breast surgery, the doctor was using the 9900 electrosurgery generator and the patient’s pacemaker stopped working. Cardiopulmonary resuscitation (CPR) was used to resuscitate the patient. Telemetry indicated the device’s programmed pulse amplitude and pulse width had changed since the last follow-up visit, but the nurse was positive that no-one had reprogrammed the device. The cell voltage was 2.76 V and capture and sensing were appropriate. Upon further questioning, the patient stated that he had recently been defibrillated. The reporter indicated that the patient experienced symptoms of syncope while shopping in Lowe’s department store, which she was visiting for the first time. It was stated that Lowe’s has an ‘in-store satellite system’ for cellular phone users. The patient was at a pharmacy waiting for his wife at the main entrance. While waiting, he was leaning against an electronic article surveillance system (EAS) and received a shock after a few moments. When the patient went in for his routine follow-up, the physician observed noise on the stored EMG [This report says ‘EMG’ (electromyogram) but we believe it means ECG (electrocardiogram)] and learned of the patient’s experience with the EAS system on that date. A healthcare provider reported that as a patient walked through the theft detector at a university library, he received a shock that knocked him to the floor. The patient’s implanted device was reset to zero by the shock. The patient sustained no known injury as a result of the shock. The healthcare provider reprogrammed the device and it is functioning properly. The library has been investigating this event and the healthcare provider stated that the security system was reportedly ‘set too high’.

ADVERSE EVENTS REPORTED TO FDA

305

Table 20.4 (Continued) Affected medical device

EMD Source

Oxygen panel

During surgery, every time an electrosurgiElectrosurgical cutting cal device (cutting and coagulation) was and coagulation device activated, the oxygen panel would alarm. (a device intended to The electrosurgical device was inspected remove tissue and and found to be operating normally. The control bleeding by staff contacted the oxygen alarm panel use of high-frequency manufacturer, who acknowledged an electric current) EMI susceptibility and offered an upgrade that would correct the problem. Two-way radio While a hospital staff member was in a patient’s room, staff used a two-way radio, causing the ventilator to shut off. No negative outcome; a registered nurse reset the ventilator. Further testing after this incident revealed that facilities’ two-way radios can not be used within 8 feet of patients. Signage regarding the use of electronic equipment has been posted and in-services to the staff completed. Cell phone Patient receiving ventilator support in the intensive care unit with high-frequency oscillatory ventilator. The ventilator spontaneously shut off. Patient needed manual ambu-bag ventilation to survive while the ventilator was restarted. A cell phone used in the unit at the time of the event was the only suspected cause of the malfunction. User reports a number of incidents of Harbor radio, citizen unexpected movement of his wheelchair. band radio from The patient had to turn off the wheelchair trucks, radio-controlled while in crosswalk and was afraid of targets on naval base careening off of the sidewalk.

Ventilator

Ventilator

Powered wheelchair

Details

electromagnetic energy. In the remaining reports, the source was either work-related (2%) or was ambiguous or not reported (21.5%).

Description of MRI-related adverse events The 29 MRI reports describe three types of events: metal objects pulled into the active MRI system, tissue heating involving conductive leads or implants, and possible EMI from the MRI emissions. AMDEs in which the magnet attracted a metal object in the MRI

306

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

Table 20.5 Examples of suspected EMI-related death reports coded as EMI and reported to FDA, January 1994–March 2005 Affected medical device

EMD source

Details

Pacemaker

Automated external defibrillator

Pacemaker

Electrocautery device

Automated external defibrillator

Radio-frequency waves, source unknown, possibly helicopter radio

Telemetry system

Unknown, likely radio-frequency interference

The medical device affected was a pacemaker and the source of EMD was an automatic external defibrillator. The report indicates that the device was ‘removed from an unexpected death. The patient was cardioverted and the pacemaker was subsequently said ‘not to work’. The report did not specify the location of the event, the patient’s age or gender. The medical device affected was a pacemaker and the source of interference was electrocautery equipment. A 69 year-old male ‘presented to the hospital for an AV graft procedure. Post-surgery, no pacing was seen. Epinephrine was administered to increase the patient’s heart rate. The patient’s rhythm deteriorated from the 80s to a brady rhythm to ventricular fibrillation. The physician believed that the loss of pacing was due to loss of capture from the electrocautery used during the graft procedure or ischemia. The patient subsequently expired. The medical device affected was an automated external defibrillator and the EMD source was radio-frequency waves. The device manufacturer reported that while a female patient (age unknown) was being transported by helicopter, the patient’s pacing was intermittently interrupted by radio frequency interference. The patient subsequently expired. The medical device affected was a telemetry system and the source of interference was the device’s antenna. The facility administering care filed this report regarding a 64 year-old male: ‘This is a precautionary notice. Interference of telemetry equipment caused lapses in patient monitoring. Patient was attended by Registered Nurse who left room for about 1 minute. Patient stopped breathing and a code was called. The telemetry equipment failed and the alarm did not go off. It is not hospital’s belief that the equipment failure contributed to the

307

ADVERSE EVENTS REPORTED TO FDA

Table 20.5 (Continued) Affected medical device

Telemetry system

Detector and alarm for arrhythmia

EMD source

Details

patient’s death because the patient was attended within 60 seconds. However, it is impossible to say for certain, so the hospital is reporting. The equipment failed because of interference coming across the antennas’. Unknown, likely The medical device affected was a telemetry system and the source of interference was radio-frequency radio-frequency waves. The patient was an interference 82 year-old female. ‘Rented telemetry unit picked up lots of interference. When a nurse went to check on the patient, the patient was found unconscious, with no pulse,’ The patient was ‘coded but to no avail. Equipment failed to notify staff of abnormal occurrence’. Unknown The medical device affected was a detector and alarm for arrythmia and the source of interference was unknown. The report filed by the manufacturer states: ‘The doctors claimed they were misled by the ‘‘pacemaker synch. marker’’ which seemed like an ‘‘R’’ complex on a patient’s waveform that had VF with pacing. During above situation the monitor gave HR readings and did not alarm for ventricular fibrillation or asystole.’ Patient’s age and gender are unknown.

Table 20.6 Distribution of medical devices identified as EMI-affected in reports received January 1994–March 2005 (n ¼ 405) EMI-affected devices

Medical devices Implanted devices Monitoring devices Therapeutic devices Surgical devices Diagnostic devices Prosthetic devices Other/unknown

Frequency (n)

(%)

290 47 29 9 9 3 18 405

71.6 11.6 7.2 2.2 2.2 0.7 4.5 100.0

308

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

Table 20.7 Devices most frequently affected by EMI in the suspected EMI-related reports received by FDA, January 1994–March 2005 (n ¼ 405) Device definition Cardiac pacemaker: a device that has a power supply and electronic circuits that produce a periodic electrical pulse to stimulate the heart. The device is used as a substitute for the heart’s intrinsic pacing system to correct both intermittent and continuous cardiac rhythm disorders. 21 CFR §870.3610 Implanted spinal cord/nerve stimulator and implanted intracereral/ subcortical stimulator: a device that is used to stimulate electrically a patient’s spinal cord to relieve severe intractable pain. The stimulator consists of an implanted receiver with electrodes that are placed on the patient’s spinal cord and an external transmitter for transmitting the stimulating pulses across the patient’s skin to the implanted receiver. 21 CFR §882.5880 and §882.5840 Monitors: these include electrocardiogram, electroencephalogram, and other patient monitoring devices. Implantable infusion pump (or insulin infusion pump): a device used to pump fluids into a patient in a controlled manner. 21 CFR §880.5725 Automatic implantable cardiac defibrillator: a device that monitors heart rhythms, and delivers shocks if dangerous rhythms are detected. Motorized scooter/wheelchair (powered wheelchair): a battery-operated device with wheels that is intended for medical purposes to provide mobility to persons restricted to a sitting position. 21 CFR §890.3860 Ventilator: a device intended to mechanically control or assist patient breathing by delivering a predetermined percentage of oxygen in the breathing gas. 21 CFR §868.5895 Automated external cardiac defibrillators (external transcutaneous cardiac pacemaker, non-invasive: a device used to supply a periodic electrical pulse intended to pace the heart. The pulse from the device is usually applied to the surface of the chest through electrodes such as defibrillator paddles. 21 CFR §870.5550 Electrocardiogram and electrocardiogram monitor (diagnostic): a device used to process the electrical signal transmitted through two or more electrocardiograph electrodes and produce a visual display of the electrical signal produced by the heart. 21 CFR §870.2340 1

(n) (% total reports)1 195 (48.1)

65 (16.0)

46 (11.3) 15 (3.7) 12 (3.0) 12 (3.0)

8 (2.0)

7 (1.7)

5 (1.2)

Table includes only specific devices that account for at least 1% of the affected devices.

suite that injured the patient or a healthcare professional accounted for 31% (9 of 29) of the MRI-related reports. The objects included monitors, power adapters, an IV pole, ferrous objects, infusion pumps, syringe pumps, a portable oxygen cylinder, and an anesthesia machine. The second type of MRI-related event involved medical devices that were brought into the MRI suite and affected by the strong magnetic fields emitted by the MRI system, resulting in burns, blisters, pain or tugging sensations. There were five (17% of 29) of these events and the affected devices included: EKG leads, ECG patches, a metal pin for a

309

ADVERSE EVENTS REPORTED TO FDA

Table 20.8 (n ¼ 405)

Distribution of EMD sources in reports received by FDA, January 1994–March 2005

EMD source

Was the source a medical device? Yes No Multiple sourcesa Unknown source Medical devices Diagnostic devices Monitoring devices Surgical devices Therapeutic devices Implanted devices Unspecified medical devicesb Non-medical devices Conducted electromagnetic energy Radiated electromagnetic energy Work-related Other/unknown Total

Frequency (n)

(%)

183 146 4 72 405

45.2 36.0 1.0 17.8 100.0

13 10 61 77 12 10 183

3.2 2.4 15.3 19.0 2.9 2.7 45.2

13 114 7 88 222

3.2 28.1 1.7 21.8 54.8

405

100.0

a

Includes medical devices and non-medical equipment. Occurred in the operating room.

b

leg fracture, a metal clip for aneurysm, and an electrical brain stimulator for Parkinson’s disease. The third type of MRI-related event involved active medical devices brought into the MRI suite where their exposure to the electromagnetic field resulted in malfunction of the medical device. These were considered likely to be actual EMI events. One report was difficult to classify because the exposure to the MRI magnetic fields resulted in both device malfunction and patient injury. This adverse event is the last report listed in Table 20.9. Because a device malfunction was included in the event description, this report was classified as an EMI MRI-related event. EMI MRI-related events accounted for 15 (52% of 29) of the MRI-related reports. Table 20.9 presents some example reports for a variety of medical devices, including cardiac pacemakers, ICDs, infusion pumps, and spinal cord stimulators. The infusion pump events are especially disconcerting, as they resulted in drug delivery overdose and in one case caused the pump motor to reverse.

310

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

Table 20.9 Examples of reports of suspected EMI events related to magnetic resonance imaging (MRI) Affected device

Event description

Pacemaker

The patient presented to the hospital with the device in back-up mode. The patient had recently had an MRI. A software download was successful, but post-download measured battery data was 2.28 V, 9 k, 41 mA. The device was subsequently replaced. During a routine follow-up, interrogation revealed dashes for parameters with a message of back-up operation. Attempts to reprogram and perform a software download were unsuccessful. The device was reportedly exposed to MRI. The device was replaced. The day after an MRI was performed, the AICD was in a HWVVI reset mode upon interrogation. It was discovered that all defibrillation therapies were off as a result of the HWVVI. Report received of an overdelivery of propofol due to the pump programming changing from a primary infusion to a secondary infusion while in MRI. Approximately 40 ml propofol had infused in a 15 minute time period. The patient had no blood pressure, a heart rate of 40 beats/minute and was treated with an unspecified amount of atropine. The patient responded to the administration of atropine with an increased heart rate and blood pressure, and required no further interventions. The pump was replaced and the same tubing was used to resume the infusion on a different pump. The pump in use during the event reported never had a secondary infusion programmed. During MRI, the pump infused all drug in the reservoir. The patient experienced burning sensation and baclofen/morphine overdose. The device was explanted because patient developed pinpoint pupils and became drowsy after the MRI. Patient wore infusion pump during MRI. Following inability to prime infusion set tubing, the patient returned pump for evaluation. The evaluation confirmed that the pump motor was running backwards. A patient wore his/her pump into an MRI. During that night glucose levels dropped to 21. Patient called 911 and began to eat. When the paramedics arrived, patient’s blood glucose was 60. At 8.00 a.m. it was at 21 again. Patient called physician and was provided with a replacement pump. The patient received a shock down the right arm during a MRI exam. The arm became swollen and any attempt to reprogram the device resulted in additional shocking sensations in the arm. Reportedly the unit was off and all parameters were at zero before the patient went into the MRI exam. The patient had two stimulators implanted, one for occipital purposes and one for arm pain. No pain was observed during the MRI from the occipital stimulator. The device was explanted.

Pacemaker

Automatic implanted cardioverter defibrillator(ICD) Infusion pump

Implanted infusion pump

Insulin infusion pump Insulin infusion pump

Implanted spinal cord stimulator

DISCUSSION

311

Discussion Both of the studies described in this chapter reviewed reports of AMDEs received by the FDA to identify and characterize EMI-related adverse events. The earlier study focused on medical devices in the cardiovascular panel, relied on text searches to identify EMI events, and for the reports selected rated the likelihood that the event was EMI-related. The more recent study focused on reports that met the definition for suspected EMI cases, used problem codes to identify suspected EMI cases, excluded reports in which the information needed to determine whether the event was EMI-related was insufficient or ambiguous, and included reports sent by manufacturers, distributors, user facilities, and voluntary sources. Although the methodologies used to identify EMI events differed, the results of the two studies were similar. In both studies, pacemakers were the medical devices most frequently reported to be affected and deaths were rarely reported. In the earlier FDA study, an activated electrosurgical unit or an external defibrillator in the vicinity of the pacemaker or ICD were the most frequent potential EMD sources. In the more recent study, the most frequent non-medical EMD sources were radiated electromagnetic energy emitted from products that included wireless communication devices and security systems in public places. The different study periods and the recent proliferation of new wireless technology may explain the difference in EMD sources identified by each study. The adverse events reported in the MAUDE database mirrored the reports in the medical literature. Cellular or mobile phones were reported as potential causes of interference with a variety of medical devices. Cellular phones have been reported to interfere with implanted devices, such as a bone-anchored hearing aid [22], monitoring devices, such as ionizing radiation dose monitoring equipment [23], and therapeutic devices, such as mechanical ventilators [24,25] and infusion pumps [26]. Appropriate medical device design, testing, and separation distance appear to be adequate in some cases, resulting in recommendations for patients with implanted electronic devices, such as pacemakers, implantable cardioverter-defibrillators, and implantable deep brain stimulators, to keep a separation distance between their cellular phone and their implanted device [27,28]. Cellular phone and personal digital assistant (PDA) use in the hospital or clinic has been controversial, with some asserting that medical device filtering, shielding and separation distance from the radio-wave emitter are adequate with modern medical devices and modern cellular phones, and others unable to agree [29–32]. A case of acute epinephrine poisoning caused by cellular phone interference reported in a medical journal and to FDA illustrates the serious health consequences of EMI with a medical device [33]. An 18 year-old with septic shock was admitted to the pediatric intensive care unit of a US hospital. An epinephrine infusion was initiated, and after 9 hours the patient abruptly complained of headache and abdominal and chest pain. The patient’s blood pressure increased, and he developed pulmonary edema, electrocardiogram (ECG) changes and myocardial damage, as evidenced by elevated cardiac enzymes. After noticing that the epinephrine infusion bag contained 24 ml less than it should have, the hospital staff noted that a relative of the patient had received a phone call

312

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

on a personal cellular telephone just before the patient’s acute episode. The hospital’s engineering department removed the infusion pump from service, reset its rate and observed the device as several cell phones were held at different distances from the pump while receiving a call. The pump spontaneously delivered a higher rate of epinephrine during several test calls. Although the engineering department was unable to consistently reproduce the infusion malfunction, it noted that the rate change was more frequent when the cell phone was closer to the infusion pump. A review of the history log of the pump showed no record of the event. It was concluded that EMD from the cell phone caused the infusion pump to malfunction and deliver an overdose of epinephrine, which led to epinephrine poisoning. This case also illustrates how the problem was recognized, investigated by the clinical and engineering staff, and reported to FDA. Patients with pacemakers or implanted cardioverter defibrillators are made aware of the potential for EMI and are typically warned to avoid contact with potential sources of EMD, including instructions not to put a cellular phone in their shirt pocket or near their implanted device. While informing patients is important, this is not enough to mitigate the problem. Adequate EMCs designed into devices and policies to prevent EMI and the adverse patient outcomes that may result are also needed. As with reports to the FDA, medical device interference caused by other medical devices was identified or considered in a number of situations. Examples include infusion pumps failing during radiotherapy sessions [34], interference with pacemakers by endoscopic video capsules [35], potential interference with cochlear implants by dental devices [36], the interference of a thalamic stimulator with an electrocardiogram monitor [37], and the interference with implantable cardioverter-defibrillators by surgical devices or interventional procedures [38,39]. Interference with implanted devices has also been reported or considered from common household electronic products, including convection ovens [40] and washing machines [41]. The workplace may also present potential problems and consideration of the workplace environment may be necessary for patients’safety when returning to work after surgery [42,43]. Reports of airport metal detectors and electronic article surveillance (antitheft) device interference with implanted devices are also considered and dismissed [44.45]. However, such conclusions would seem to ignore the concerns raised by the significant number of reports identified by FDA as attributable to medical device EMI [17]. Also, some types of medical devices are more susceptible to these EMD sources than others or their failure has a more significant adverse impact on the patient. The prevalence and incidence of EMI events with medical devices and their significance are difficult to gauge. Underreporting of EMI appears to be common, and a number of factors contribute to it [3]. The healthcare staff may not be aware of the causes of EMI or how EMI is manifested. The intermittent nature of the effects of EMI on device function often makes it difficult to recognize EMI events. The device may function normally after the incident or the user may find a ‘work around’ that allows him/her to use the device. Fear of liability or litigation may also inhibit a potential reporter. Typically, events that result in a patient injury are more likely to be noticed and reported. While some events may be distinctly recalled by patients, such as the patient with a spinal cord stimulator for pain relief who reported being thrown to the floor from

DISCUSSION

313

the interference caused by a library antitheft system (Table 20.2), others, such as the silent resetting of a patient’s ICD, may go unnoticed until noted during routine follow-up with the patient. One study reported the incidence of antitachycardia therapy suspension due to possible EMI in implantable cardioverter defibrillators as 11% per patient year [46]. In this unique study, medical records for 46 patients with ICDs capable of storing information on magnet reversions were retrospectively reviewed for such events. In none of these patients were tachyarrhythmias present at the time of the possible EMI-related ICD inactivation. To complicate matters, it is not certain that all of these types of events can be attributed to EMI. It is possible that device malfunction, exclusive of EMI, may contribute to deactivation of ICDs or inappropriate behavior of implanted electronic devices. Ruling out device malfunctions, particularly when they are idiopathic, is not always possible. Suspected EMI is difficult to confirm and few attempt to do so. A review of the potential for mobile phone interference with medical equipment found considerable heterogeneity in conduct of tests for EMI with specific medical devices [29]. A clear understanding of when EMI becomes clinically relevant is generally lacking, which leaves the importance of EMI with medical devices difficult to interpret. A major limitation of our analyses is that not all EMI-related AMDEs are reported to FDA, so the reports we identified do not represent the universe of cases that occur. Among the other limitations of our analyses was the high proportion of missing values for some of the study variables, which limited our ability to characterize the EMI-related events. Data on many study variables are not systematically collected. The search strategies used in the studies we reported retrieved many EMI-related AMDEs, but it is likely that we may not have found all the relevant reports. We relied on problem codes that might be assigned by the manufacturer when they make the report or assigned by a contractor who enters the data into the FDA database. We found that 10% of the reports coded as EMIrelated were in fact not related to EMI. The problem codes in the MAUDE database could be improved. For example, modifying problem codes to differentiate between electromagnetic interference and interference related to the reactions with reagents would focus the search results on relevant reports. Similarly, developing more specific problem codes might make it easier to distinguish EMI events from events that resulted from electrical overstress. Information in the event report may be inadequate to distinguish between these two types of events. We recommend using text searches to complement the search by problem codes; text searches could help to identify additional relevant reports not identified by problem codes and may also be useful to validate the results of searches conducted with problem codes. Despite the limitations noted, the analyses we reported were based on a nationwide database that provided over 10 years of data on EMI events. The earlier study helped to define problem codes for the MAUDE database developed when user facility regulations went into effect. The information gathered from the reports also contributed to the development of test methods and improved EMC requirements for national and international consensus standards. For the more recent study, the MAUDE database provided problem codes designed specifically to identify EMI events. Furthermore, these studies

314

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

illustrate the methodologies that can be used to evaluate other problem codes designed to identify other postmarket problems. The tip-of-the-iceberg paradigm diagrammed in Figure 20.5a and b describes the data we currently have and the data we would ideally have to conduct postmarket surveillance of EMI-related adverse events. The base of the pyramid shows the universe of adverse events, which must be identified, reported, properly coded, and entered in the MAUDE database in a timely manner. We currently only see the tip of the iceberg; efforts to increase public awareness of the injuries that result from EMI must continue. FDA established the CDRH EMC Work Group and, in collaboration with professional organizations, is disseminating information about EMI and approaches to assure EMC by posting public health advisories when necessary, publishing and presenting scientific papers, and participating in the development of guidelines and recommendations for EMC/EMI management and standards for medical device EMC. One of the EMC Work Group’s goals is to educate the public and increase recognition of the problem, and in turn increase the percentage of events reported. Improving the quality and completeness of the information reported about those events is a related goal. Postmarket surveillance of EMI-related ADMEs would be more informative if the reports systematically documented the patient’s history and outcomes, the indications for the use of the device, the nature of the malfunction, the likelihood that the event is EMI-related, the duration of the EMI’s effect on the device, and whether the device function is restored spontaneously, restored with the patient’s intervention, restored with clinical intervention, or fails irreparably. As mentioned above, MedSun is an FDA program that is designed to improve postmarket surveillance for adverse events. MedSun is expected to provide more complete clinical and laboratory data and information about use errors and the conditions under which new devices are used after FDA approval. Online reporting is also being considered to solve the limitation of the time lag in entering reports into the system. Finally, event reports must be evaluated, coded properly, and entered in the database. About 10% of the reports we retrieved as EMI-related were coded incorrectly, according to our criteria. Some non-EMI events were coded as EMI, and some EMI events were not coded as such. The proportion of EMI events incorrectly coded as non-EMI events remains unknown. Ideally, the number of EMI adverse events identified by the postmarket surveillance system, documented and entered in MAUDE, would be closer to the universe of actual EMI events (the bottom of the ‘ocean floor’ in Figure 20.5b). This is difficult to achieve. As public awareness increases, and the recognition of events, reporting, coding and data management improves, more accurate estimates of the number of EMI-related adverse events can be obtained. Information provided by an improved surveillance system can be used to further increase public awareness about the medical devices that could be affected by EMI and the types of malfunctions that EMI can cause. Additionally, an improved postmarket surveillance system can help identify environments and conditions of use that may need to be mitigated or studied further; and it can help identify appropriate actions to be taken in response to new EMI-related adverse events.

Identified

Universe of events

Universe of events

Reported

Coded

Entered

Identified

Reported

Coded

Entered

(b)

Improved public awareness

Improved problem recognition

Improved reporting and data quality

Improved data processing - Timely data entry - Improved coding

Figure 20.5 (a) The tip-of-the-iceberg paradigm: current postmarket surveillance of EMI adverse events. (b) Improved postmarket surveillance system to identify EMI adverse events

(a)

316

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

With the rising dependence on medical devices of all kinds and particularly on implants and other devices with electronic components, ultimately design that protects medical devices from unwanted EMI will be necessary to protect patients from the wide variety of electromagnetic energy and sources increasingly found in the environment in which medical devices are used [47]. In the meantime, medical device manufacturers and healthcare providers should provide patients with adequate information about the prevalence of EMI with medical devices, particularly with devices that are lifesustaining, inform them about the risks inherent in using a particular device, and report EMI problems to the manufacturer or to MedWatch.

References 1. Kainz W, Neubauer G, Alesch F, Schmid G, Jahn O. Electromagnetic compatibility of electronic implants – review of the literature. Wien Klin Wochenschr 2001; 24: 903–914. 2. Silberberg JL. Performance degradation of electronic medical devices due to electromagnetic interference. Compliance Eng 1993; 10(5): 25–39. 3. Silberberg, JL. What can reports of medical device electromagnetic interference tell us? Compliance Eng 1997; Annual Reference Guide: D17–D23. 4. Sykes S (ed.). Electromagnetic compatibility for medical devices: issues and solutions. FDA/ AAMI Conference Report, Association for the Advancement of Medical Instrumentation. Arlington, Virginia, USA (AAMI), 1996. 5. International Electrotechnical Commission (IEC). 60601-1-2 Ed. 2.1 (2004) and Amendment 1 (2004), Medical electrical equipment. Part 1, general requirements for safety; part 2, collateral standard: electromagnetic compatibility – requirements and tests. Geneva, Switzerland: IEC, 2004. 6. Association for the Advancement of Medical Instrumentation. Guidance on electromagnetic compatibility of medical devices for clinical/biomedical engineers, Part 1: radiated radiofrequency electromagnetic energy. AAMI TIR No. 18. Arlington, Virginia, USA: AAMI, 1997. 7. Medical device reporting: http://www.fda.gov/cdrh/mdr/ [accessed December 14 2005]. 8. Manufacturer and User Facility Device Experience (MAUDE): http://www.fda.gov/cdrh/ databases.html 9. MedWatch: the FDA safety information and adverse event reporting program: http://www.fda. gov/medwatch/ 10. Pressly N. Review of MDR reports reinforces concern about EMI. FDA User Facility Reporting No. 20, Summer 1997. www.fad.gov/cdrh/fuse20.pdf (accessed November 24, 2006). 11. Medical device reporting database: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfmdr/ search.CFM 12. ISO 14708–1 (2000). Implants for surgery – active implantable medical devices, Part 1: general requirements for safety, marking and information to be provided by manufacturer. International Standards Organization: www.iso.org 13. European Committee for Electrotechnical Standardization. EN 45502–2-1:2003 Active implantable medical devices, Part 2–1: particular requirements for active implantable medical devices intended to treat bradyarrhythmia (cardiac pacemakers): www.cenelec.org 14. European Committee for Electrotechnical Standardization. EN 45502–2-2 (draft). Active implantable medical devices, Part 2–1: particular requirements for active implantable medical devices intended to treat tachyarrhythmia (includes implantable defibrillators): www.cenelec.org

REFERENCES

317

15. Casamento JP, PS Ruggera. Applying standardized electromagnetic compatibility testing methods for evaluating radiofrequency interference with ventilators. Biomed Instrument Technol 1996; Sept/Oct. 16. FDA Center for Devices and Radiological Health. Important information on anti-theft and metal detector systems and pacemakers, ICDs, and spinal cord stimulators. Letter to cardiologists, cardiac surgeons, neurosurgeons, and emergency physicians, September 28 1998. This article can be accessed at http://www.fda.gov/cdrh/safety/easnote.html (accessed November 24, 2006). 17. Witters D, Portnoy S, Casamento J, Ruggera P, Bassen H. Medical device EMI: FDA analysis of incident reports, and recent concerns for security systems and wireless medical telemetry. Symposium Record 2001 IEEE EMC International Symposium, August 13–17 2001, Montreal, Canada;1289–1291. 18. Kainz W, Casamento JP, Ruggera PS, Chan DD, Witters DM. Implantable cardiac pacemaker electromagnetic compatibility testing in a novel security system simulator. IEEE Trans Biomed Eng 2005; 52(3): 520–530. 19. Intercooled Stata 9. Copyright ( 2005, StataCorp LP, 4905 Lakeway Drive, College Station, TX 77845, USA: www.stata.com 20. The R Foundation for Statistical Computing. Version 2.2.0 (2005-06-20), ISBN 3-900051-07-0: http://www.r-project.org/ 21. Code of Federal Regulations. Title 21: Food and Drugs: http://www.fda.gov/cdrh/aboutcfr.html [accessed October 25 2005]. 22. Kompis M, Negri S, Ha¨usler R. Electromagnetic interference of bone-anchored hearing aids by cellular phones. Acta Otolaryngol 2000; 120: 855–859. 23. Gilligan P, Somerville S, Ennis JT. GSM cell phones can interfere with ionizing radiation dose monitoring equipment. Br J Radiol 2000; 73: 994–998. 24. Close range EMI sends Nellcor Puritan Benett 840 ventilators into ‘vent inop’ mode. Health Devices 2003; 32:128–130. 25. Shaw CI, Kacmarek RM, Hampton RL et al. Cellular phone interference with the operation of mechanical ventilators. Crit Care Med 2004; 32:928–931. 26. Richardson NH. Interference from mobile phones in the ICU. Br J Intensive Care 1995; 7: 233–234. 27. Electromagnetic Compatibility Program: http://www.fda.gov/cdrh/emc/pace.html [accessed October 4 2005]. 28. Kainz W, Alesch F, Chan DD. Electromagnetic interference of GSM mobile phones with the implantable deep brain stimulator, ITREL-III. Biomed Eng OnLine 2003: 2: 11–20. 29. Lawrentschuk N, Bolton D. Mobile phone interference with medical equipment and its clinical relevance: a systematic review. Med J Aust 2004; 181: 145–149. 30. Klein AA, Djaiani GN. Mobile phones in the hospital – past, present and future. Anaesthesia 2003; 58: 353–357. 31. Aziz O, Sheikh A, Paraskeva P, Darzi A. Use of mobile phone in hospital: time to lift the ban? Lancet 2003; 361: 788. 32. Wireless communication devices and electromagnetic interference. ECRI’s updated recommendations. Health Devices 2001; 30: 403–409. 33. Hahn I, Schnadower D, Dakin RJ, Nelson LS. Cellular phone interference as a cause of acute epinephrine poisoning (letter). Ann Emerg Med 2005; 46: 298–299. 34. Electromagnetic interference from linear accelerators can affect electronic devices. Health Devices 2001; 30: 260–262. 35. Dubner S, Dubner Y, Gallino S, Spallone L et al. Electromagnetic inteference with implantable cardiac pacemakers by video capsule. Gastrointest Endosc 2005; 61: 250–254.

318

CH20

EXPLORING METHODS FOR ANALYZING SURVEILLANCE REPORTS

36. Roberts S, West LA, Liewehr FR, Rueggeberg FA et al. Impact of dental devices on cochlear implants. J Endodont 2002; 28: 40–43. 37. Khan IA. Differential electrocardiographic artifact from implanted thalamic stimulator. Int J Cardiol 2004; 96: 285–286. 38. Madigan JD, Choudhri AF, Chen J, Spotnitz HM et al. surgical management of the patient with an implanted cardiac device: implications of electromagnetic interferences. Ann Surg 1999; 230: 639–647. 39. Fiek M, Dorwarth U, Durchlaub I, Janko S et al. Application of radiofrequency energy in surgical and interventional procedures: are there interactions with ICDs? Pacing Clin Electrophysiol 2004; 27: 293–298. 40. Rickli H, Facchini M, Brunner H, Ammann P et al. Induction ovens and electromagnetic interferences: what is the risk for patients implanted with pacemakers? Pacing Clin Electrophysiol 2003; 26: 1494–1497. 41. Kolb C, Schmieder S, Schmitt C. Inappropriate shock delivery due to interference between a washing machine and an implantable cardoverter defibrillator. J Intervent Card Electrophys 2002; 7: 255–256. 42. Gurevitz O, Fogel RI, Herner ME, Sample R et al. Patients with an dICD can safely resume work in industrial facilities following simple screening for electromagnetic interference. Pacing Clin Electrophysiol 2003; 26: 1675–1678. 43. Gustrau F, Bahr A, Goltz S, Eggert S. Active medical implants and occupational safety: measurement and numerical calculation of interference voltage. Biomed Tech 2002; 47(suppl 1, pt 2): 656–659. 44. Kolb C, Schmieder S, Lehmann G, Zrenner B et al. Do airport metal detectors interfere with implantable pacemakers or cardioverter-defibrillators? J Am Coll Cardiol 2003; 41: 2054–2059. 45. Niehaus M, Tebbenjohanns J. Electromagnetic interference in patients with implanted pacemakers or cardioverter-defibrillators. Heart 2001; 86: 246–248. 46. Kolb C, Deisenhofer I, Weyerbrock S, Schmieder S et al. Incidence of antitachycardia therapy suspension due to magnet reversion in implantable cardioverter-defibrillators. PACE 2004; 27: 221–298. 47. Render ML. Research and redesign are safer than warnings and rules. Crit Care Med 2004; 32: 1074–1075.

21 Alternative and complementary medical devices S. Lori Brown US Food and Drug Administration, Bothell, WA, USA

Joannie C. Shen US Food and Drug Administration, Rockville, MD, USA

Introduction In 1991, the US Congress passed legislation to provide funding to establish an office at the National Institutes of Health to investigate and evaluate the use of unconventional medical practices. Today, the National Center for Complementary and Alternative Medicine (NCCAM) concentrates on investing in research into unconventional medical practices, encouraging the use of vigorous scientific evidence to evaluate these practices. Conventional medicine is defined as medicine practiced by medical doctors (MD) or doctors of osteopathy (DO) and is also called ‘Western medicine’ or ‘allopathy’. Unconventional medical practices include complementary and alternative medicine and are described as a group of diverse medical and healthcare systems that are not part of conventional medicine. Complementary medicine is defined as medicine that is used with conventional medicine, usually to supplement or enhance conventional treatments. The practice of using complementary treatments with conventional medicine is sometimes referred to as ‘integrative medicine’. Alternative medicine is used instead of conventional medicine. The US Food and Drug Administration (FDA) currently has no legal or regulatory definition of complementary or alternative medicine [1].

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

320

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

Patients may prefer alternative or complementary treatments because of their distrust of conventional medicine or concerns about the safety or effectiveness of conventional medical treatments and therapies [2]. Patients may have tried conventional therapies that provide no relief from their symptoms or that have side effects that are unacceptable to the patient. If patients’ mistrust of conventional medicine drives them to seek alternative treatments, they may not trust the results from randomized controlled-clinical trials performed by those at the heart of the system of conventional medicine. The NCCAM’s mission includes a budget for providing training to potential investigators and support for studies which are not necessarily performed by scientists or others from the dominant health system. Studies of the use of alternative therapies indicate that, despite the lack of scientific evidence of safety or effectiveness for most modalities, their use is widespread and increasing in the USA and Europe. Nationally representative random household telephone surveys in 1991 and 1997 indicate that use of specified alternative therapies has increased from 33.8% in 1990 to 42.1% of those surveyed in 1997 [3]. One of the therapies among those that reportedly increased the most was ‘energy healing’, which includes the use of magnets and energy-emitting machines. The use of alternative or complementary therapies is reportedly growing worldwide [4]. Most of the attention has been focused on the use of alternative or complementary ‘botanicals’ or ‘nutriceuticals’ and less on medical devices. There are many medical devices which could be classified as ‘alternative’ or ‘complementary’. We discuss here examples of some of these devices and the type of information that is available. Some of these products are exempt from premarket review by FDA, but many require review and clearance/approval before marketing. In some cases they have been marketed illegally without seeking FDA clearance, or have been promoted for a range of claims never cleared/approved by the FDA. It bears repeating that FDA has established no definition of complementary or alternative devices. Surveillance or epidemiologic studies on these devices are usually absent. Unless there are known serious injuries associated with these medical devices, consistent FDA oversight of these products may not be a public health priority. The extent to which these devices are being used by the public, the benefit of these devices, and the number of adverse events or full spectrum of problems associated with these products is often unknown. An example of devices that have been cleared by the FDA through the 510(k) process, but are used for off-label treatment or diagnosis for purposes for which they have not been cleared, is electronic muscle stimulators. These devices are indicated for use in physical therapy for relaxation of muscle spasm, prevention or retardation of disuse atrophy, increasing local blood circulation, muscle re-education, immediate postsurgical stimulation of calf muscles to prevent venous thrombosis, or maintaining or increasing range of motion. They have also been advertised for uncleared purposes, such as weight loss or physical conditioning without exercise, i.e. to achieve ‘rock-hard abs’ [6,7]. The example of the Relaxicisor, shown in Table 21.1, illustrates that there may be a fine line between off-label use of a cleared or approved device as an alternative or complementary device and intentional health fraud. Table 21.1 illustrates some devices and claims associated with these devices. The internet has extensive listings of information

INTRODUCTION

321

Table 21.1 A few alternative devices found on the internet. None of these devices have been cleared or approved by FDA and the associated claims are considered fraudulent Device

Description or claims

Electrodiagnostic devices [41]

By measuring disturbances in the body’s flow of electro-magnetic energy along acupuncture meridians, these devices can reportedly be used to diagnose and treat a broad spectrum of problems, including allergies, asthma, bronchitis, gastritis, depression, health and circulatory illness, food intolerance, pancreatitis, rheumatic disorders, and pain This is described as an electric gas barbecue grill igniter outfitted with finger grips. When pressed against the skin, the device sparks and causes a small electric shock. Purveyors of the device claim that it can relieve headaches, back pain, arthritis, stress, menstrual cramps, earaches, sinus, nosebleeds, ‘flu and other ailments Provides electrical shocks to the body through contact pads, reportedly making weight reduction and muscle toning possible without diet or exercise Reportedly cures cancer and makes surgery, chemotherapy, and other conventional cancer treatments unnecessary. Bioresonant therapy can be used to strengthen the immune system, for treatment of allergy, treatment of badly healing wounds, and treatment of diseases Among other things, the master essence vial reportedly boosts the immune system and enhances the balance of all bodily systems. It is especially effective when worn over the breastbone area, where it supports the thymus gland. The vial comes with a cloth bag and can also be used with a radionics machine to send the healing energy to another The radionics practitioner reportedly taps into the ‘vibrational frequencies’ of a patient’s bioenergy to diagnose the cause of illness. Patients complete a questionnaire about their medical history and supply a hair sample or a drop of blood. The sample is placed on a radionic black box which has frequency settings tuned to the various disorders. When the diagnosis is complete the machine broadcasts healing energy to the patient. The list of illnesses that are reportedly diagnosed and treated with radionics is extensive This item, also referred to as the ‘Rife’, reportedly works to cure cancer and other maladies, such as leprosy, tuberculosis, and typhoid, because the chemical nature of each organism has a resonant frequency determined by its chemical make-up. Destruction of bad organisms such as viruses and bacteria by the rays cures cancer and disease without hurting human tissues. This technology is reportedly safe and painless

Stimulator, Crystaldine pain reliever [42,43]

Relaxacisor [44]

BioResonance and molecular enhancer device [45]

Master essence vial [46]

Radionic machines [47]

Bioray Light and Sound Generator [48]

322

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

related to these devices. There is no information on the safety or effectiveness of these devices, but for many of them the safety issue may be nebulous, that is, related to discouraging patients from seeking conventional treatment for curable ailments. There is the potential for serious injury with these devices, for instance, interference with a pacemaker or defibrillator (see Chapter 20). The amount of money spent on these treatments is also unknown, but the issues associated with cost have to do with a supplier providing a consumer with a product (or service) that will not provide the benefits advertised.

Acupuncture needles The line between alternative and complementary medicine may shift over time. An example of this, involving a medical device, is acupuncture. At one time in the USA, acupuncture was viewed as an alternative practice, but it is now used as a complementary treatment. Acupuncture is part of a tradition of ancient Chinese healing practice that inserts needles at specified acupuncture points to redirect and reposition the flow of energy (see Figure 21.1). This practice is used to relieve pain, such as migraine pain or

Figure 21.1 Engraving of moxabustion acupuncture being performed, Japan, 1863–1864. Reproduced with permission from the Science Museum/Science and Society Library

323

ACUPUNCTURE NEEDLES

Table 21.2 Acupuncture was until recently considered an alternative medicine. The FDA has now cleared numerous products and brands for use in acupuncture Product Acupuncture needle Acupuncture point locator Electro-acupuncture stimulator

Year first cleared 1996 1978 1979

Number cleared to date 66 3 7

menstrual pain, or to treat chronic medical conditions, such as arthritis or asthma. It is now recommended or prescribed by some MDs or DOs for their patients who have, for instance, nausea from chemotherapy or chronic pain. Auricular acupuncture is also used in conjunction with conventional practices in detoxification programs and in addiction treatment. Table 21.2 shows approved or cleared medical devices that are used in acupuncture, based on searching the public 510(k) database [8]. The acupuncture needle is now defined in regulations as a hospital and general personal use device: An acupuncture needle is a device intended to pierce the skin in the practice of acupuncture. The device consists of a solid, stainless steel needle. The device may have a handle attached to the needle to facilitate the delivery of acupuncture treatment [9].

Another acupuncture device, the electro-acupuncture stimulator, is defined as a prescription device (meaning that it must be prescribed by an MD or DO). In the 1970s, acupuncture needles were ‘investigational’devices in the USA and could not be legally sold. In April of 1996, the FDA announced that it had reclassified the acupuncture needle from Class III, a category usually requiring clinical studies to establish safety and effectiveness, to Class II, a category which involves less stringent controls by FDA but requires good manufacturing and proper labeling (see Chapter 2 for a more complete description). This change was largely based on a workshop held by the Office of Alternative Medicine (now NCCAM) in 1994 to investigate the current state of knowledge on the safety and effectiveness of acupuncture needles [1]. After the workshop, petitioners filed with the FDA to reclassify acupuncture needles. The FDA agreed to this change, although they did not agree that there was adequate clinical information to recommend approval for any of the proposed indications. This change made it possible to consider acupuncture needles under Section 510(k). Data from the workshop supported use of acupuncture needles for a number of indications, including acute and chronic pain, nausea and vomiting, and substance abuse. The acupuncture needles used in human body acupuncture are generally 0.2–0.5 mm in thickness. The length of the needle shaft varies considerably, according to therapeutic indications and practitioners’ styles; 20–40 mm long needles are commonly used, but longer needles are often selected by practitioners trained in traditional Chinese acupuncture practice. There are also hair-thin Japanese needles used by practitioners. Veterinarian use requires longer and larger needles for treating large animals, as commonly seen in equestrian use.

324

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

In addition to the commonly used types of body needle described above, other types of needle are also used. Among them are press tacks, intradermal needles, and stud needles used for body acupuncture, finger acupuncture, and auricular acupuncture. These are often short needles intended to be left in place for varying lengths of time. Patients may be instructed to self-simulate the needles according to a schedule set forth by the practitioner or according to the patient’s own symptoms. Aside from the above standard body and ear acupuncture needles, there are a variety of related devices in acupuncture practice. Moxibustion (or Moxa) is the burning of dried fiber herbs to deliver heat stimulation, often wrapped around the needle handle after acupuncture needle insertion. The electrical acupuncture point detector is a device used for accurately locating acupuncture points by measuring the electrical resistance of the skin, assuming that acupuncture points have a lower electrical resistance. The value of acupuncture point detectors is debated, even among acupuncture practitioners. The electrical acupuncture simulator is used to deliver electricity to stimulate acupuncture points once the acupuncture needles are inserted. The current and frequency can be selected by the acupuncture practitioners according to therapeutic indications. Clinical evidence is lacking regarding whether electrically stimulated acupuncture is more effective than manual acupuncture. Other devices are used with acupuncture for cupping, which is thought to eliminate stagnation and improve the flow of vital energy. The practitioner creates suction in the cup (most often by lighting a flame in the cup) and applies the cup to the body. The skin rises under the cup vacuum force and often results in a local bruising. Currently the FDA registered general medical devices list includes acupuncture point locators, cupping sets, and various acupuncture needles. Although today’s acupuncture needles are made mostly of steel, one may find traditional practitioners using gold or silver needles. These metals are considered by some to have special therapeutic indications and properties. Re-usable acupuncture needles impose serious public health concerns; note, for instance, an outbreak of acupunctureassociated hepatitis B [10]. Because of increased awareness of the risk of communicable diseases, most practitioners in the USA now use disposable needles. Firms marketing acupuncture needles in the USA are now required to obtain premarket clearance from FDA through the premarket notification (510k) process. Manufacturers are required to label the needles for single use only, restrict their use to qualified practitioners as determined by the states, and provide information about device material biocompatibility and sterility. Foreign manufacturers must meet the same premarket clearance and manufacturing quality requirements as US manufacturers. Although there are no available data demonstrating the safety effects of this reclassification, with acupuncture needles marketed as a single-use, sterile product, to be used only by qualified acupuncture practitioners; together with the acupuncture profession’s emphasis on clean needle technique in training, written examination, and a requirement for licensure throughout the USA, acupuncture patients may now donate blood without a 12 month deferral (if the needle sterilization procedure cannot be verified, then the donor should be deferred for 12 months).

ACUPUNCTURE NEEDLES

325

A recent review on the effectiveness of acupuncture concluded that, despite the difficulties and methodological constraints in conducting high-quality acupuncture clinical trials, a general international agreement has emerged that acupuncture appears to be effective for postoperative dental pain, postoperative nausea and vomiting, and chemotherapy-related nausea and vomiting [11]. For numerous other conditions, the evidence is considered promising, but rigorous research is still needed to add to the body of evidence. For conditions such as asthma, drug addiction, and certain pain conditions, the evidence is considered inconclusive and difficult to interpret. The review also concludes that acupuncture is a relatively safe procedure, although it is not free from serious adverse events. Regardless of the evidence on the effectiveness of acupuncture practice, the increasing popularity of acupuncture worldwide has resulted in a growing literature on adverse events associated with acupuncture, including case reports on communicable diseases such as hepatitis, and mechanical trauma from needle placement, such as pneumothorax and spinal cord injury. There were also reported fatalities from infections. A common cause of adverse events was due to needle breakage [12–15]. The best evidence in supporting the safety of acupuncture practice thus far has come from prospective studies relying on both practitioner and patient reports [13,16,17], supporting the assertion that ‘acupuncture is safe in competent hands’; but more studies are expected to add to the international evidence base. A recent study estimated severe underreporting of acupuncture-associated serious adverse events, and suggested a means of verifying the accuracy of the reported serious adverse event rate and correcting it to the statistically expected rate [18]. A recent review pooled information from computerized databases, previous reviews of case reports, population surveys, and prospective surveys of acupuncture practice to indicate the range of significant adverse events associated with acupuncture [19]. In this review, a total of 715 adverse events were included. Reports of adverse events were classified by the author as ‘primary’ if written by a clinician who dealt with the case, and as ‘secondary’ if written by another author, e.g. from retrospective studies which were considered to be more likely to subject to recall bias and which might include double reporting. Trauma accounted for 90 primary reports and 186 secondary reports; the most common were pneumothorax and injury to the central nervous system. Infection accounted for 204 primary reports and 91 secondary reports. Over 60% of these cases were hepatitis B. The next most common infection was of the external ear, as a complication of auricular acupuncture. The 144 miscellaneous events comprised mainly seizures and drowsiness judged severe enough to cause a traffic hazard. There were 12 primary reports of deaths. In addition, this review also pooled from 12 prospective studies which surveyed more than a million treatments; the risk of a serious adverse event with acupuncture was estimated to be 0.05 per 10 000 treatments, and 0.55 per 10 000 individual patients. The review thus concluded that the risk of serious events occurring in association with acupuncture is low and below that of many common medical treatments. In addition to safety concerns associated with acupuncture needles themselves, there are also concerns about their potential interactions with other medical devices. For example, with the growing and common use of magnetic resonance imaging (MRI),

326

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

patients with embedded auricular needles should be forewarned about the interactions with the large field magnet and safety concerns. Also, localized argyria and often invisible metallic remnants from prior acupuncture treatment may pose potential risk from heating and localized tissue injuries when patients undergo MR scanning. Other types of device interactions include the interaction of electro-acupuncture with other medical devices powered by electricity. Acupuncture practitioners often avoid using electro-acupuncture across the thorax because of the theoretical risk of interfering with the heart’s conduction system. One case report described electromagnetic interference from low-frequency acupuncture on a pacemaker [20]. Therefore, caution should be exercised when using electro-acupuncture on patients with any electrically powered medical devices. As previously stated, surveillance or epidemiologic studies on ‘alternative and complementary’ devices are usually absent. Surveillance or epidemiologic studies for acupuncture needles are feasible, and could be helpful in reducing future serious injuries and improving the safety of acupuncture practice. The current acupuncture professional education and licensure does not have an emphasis on adverse events reporting. It is important for the public and acupuncture professionals to know that, with their involvement in the surveillance or epidemiologic studies, the benefits of these devices and the types of adverse events and full spectrum of device issues associated with these products can be better analyzed and understood by healthcare providers and patients.

Ear candles Ear candles, also known as ear cones, are hollow candles which are used by placing the candle in the outer ear canal and lighting the outer end of the candle. The candle may be placed in the ear through a hole in a cloth or metal or paper plate with a hole cut in the middle to prevent wax from dripping on the face or into the ear during treatment. After the treatment, the end of the candle reportedly contains a dark brown substance which is described as cerumen (ear wax) and impurities pulled from the ear. A light powder is also reported. Claims regarding the beneficial effects of this practice can be found on various websites and at health stores that sell ear candles or promote ear candling. Reported benefits include treating wax build-up, ear pain, sore throats, hearing problems, hearing loss, ringing in the ear (tinnitus), rhinitis, sinus problems, headaches, ear infections, allergies, and Candida (fungus) growth in the ear, which causes itching. Ear candles also reportedly reduce stress and tension, purify the blood, clear the mind and senses, cure Me´nie`re’s syndrome, and improve hearing. Ear candles with herbs may also oxygenate and strengthen the brain and fortify the central nervous system. In addition, they reportedly have preventive properties, such as reducing the occurrence of earaches and ear-related headaches. They also reportedly have antiinflammatory, antiseptic, antibiotic, and antibacterial properties. All this sounds almost too good to be true. FDA considers ear candles to be devices that are actively regulated and require clearance or approval before marketing and for specific claims. However, companies sometimes make and sell these products without following regulatory requirements.

EAR CANDLES

327

While most websites that sell ear candles list some of the reported beneficial effects of ear candling, many of them now have a disclaimer at the end that explains that ear candles cannot be used to diagnose or treat disease. Despite disclaimers, onewebsite notes that ear candling is a safe and gentle treatment for children. Another website has a photo of a child undergoing ear candling. In addition to information on the internet, there are several books that describe the practice and benefits of ear candling. Some of these books are available from popular online bookstores; others are available from ear candle merchants. Advertised ear candles are manufactured from beeswax, paraffin, or mixtures of the two. Some may also have herbs or natural fibers in the candles which reportedly enhance the benefits. One website warns consumers to be careful of false claims from some ear candle purveyors who claim that their candles are made from 100% beeswax when they are really made from 100% paraffin [21]. Ear candling purportedly works by creating a downward spiraling energy flow of smoke and heated air which acts as a vacuum and cleanses the ear canal of wax and debris. Candling supposedly pulls the residue out of the ears, sinuses, and mind. The smoke from the candle also moves by osmosis through the skin to achieve some of its beneficial effects. The practice of ear candling is presented as being widely used by ancient Egyptians, Chinese, Tibetan, Romans, Greeks, Atlanteans, American Indians (including Hopi), pre-Columbian Americans (also Mayans), and some German doctors, too. Ear candling may be performed by a practitioner (medical doctor, naturopath, chiropractor, massage therapist, or other healthcare provider, according to one website) or it may be performed at home. If performed at home, it is usually recommended that a partner should be present to help, particularly because ear candling is reportedly so relaxing, people may fall asleep during the process. It is conceivable that sleeping with a burning candle in the ear could result in an adverse event. A small clinical trial of ear candling assessed cerumen removal but none of the other claimed benefits [22]. Eight patients, four with and four without cerumen impaction, underwent treatment with an ear candle. There was no improvement in cerumen status after ear candling in patients with cerumen impaction and in two patients without cerumen impaction, candle wax was observed in the ear after the procedure. Because of concerns over the potential for injury, no further studies were performed with humans. Analysis of the candle stub or powder by gas chromatography indicated that the sample was composed of multiple alkanes which are typically found in candle wax (not cerumen or ‘impurities’, as claimed by ear candling advocates). An investigation of the theoretical underpinning of ear candles using a Plexiglas canal the same dimensions as the ear and a tympanometer to measure a pressure change (vacuum) during ear candling indicated that no vacuum was created during this procedure [22]. Similarly, Health Canada reports that their laboratory testing did not show significant heating or suction of the ear [23]. A survey by the Northwest Academy of Otolaryngology–Head and Neck Surgeons queried whether members were aware of this practice and whether they had seen any complications due to ear candle use [22]. Of the 122 otolaryngologists responding to the survey, 40 responded that they had patients who had practiced ear candling. Fourteen physicians reported treating patients with complications due to this practice, which included 13 burns of the auricular and external auditory canal, seven partial or complete

328

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

occlusions of the ear canal with candle wax, and one tympanic membrane perforation. External otitis and temporary hearing loss were also reported. Other adverse events are reported on websites, including a case on the Video Otoscopy forum [24]. In this case report, a 55 year-old woman was seen at an audiology clinic after sustaining a burn during ear candling. She reported some loss of hearing and bleeding from her ear after the experience. Candle wax was removed from the surface of the tympanic membrane and several days later hearing was improved, consistent with restoration of tympanic membrane mobility. Two reports in the FDA adverse events reporting system were identified by searching on the term ‘ear candle’ [25]. Neither of these reports described a particular adverse event but simply asserted that ear candles were dangerous and that FDA should do something about them. The impact that FDA enforcement has had on the sale of ear candles has been minimal, as evidenced by the number of websites advertising ear candles. Occasional seizures or import alert actions have encouraged labeling on websites, which make numerous claims on the miraculous effects of ear candles, to add a disclaimer which says that ear candles cannot be used to diagnose or treat disease. Neither the prevalence of ear candling, nor the incidence or severity of adverse events, with this alternative therapy are known. Internet websites selling or describing the use of ear candles are numerous (a Google search for ‘Ear Candles’ on July 28 2005 retrieved 1 010 000 web citations). A few of these sites warn against the practice of ear candling but many more are sites selling ear candles or testimonials proclaiming the benefits of ear candling.

Magnets Therapeutic magnets or pulsed magnetic fields are offered in numerous forms, including magnetic jewelry, magnetic insoles, shoes, mattresses, wraps, and devices to provide pulsed magnetic fields. Magnets are variously advertised on websites and in printed advertisements to prevent, reduce, or relieve pain and stress, thereby warding off problems such as headaches, hypertension, gastritis, arrhythmias, depression, fatigue, arthritis flairs, and immune deficiencies. Pulsed magnetic fields also reportedly help by improving oxygenation of tissues, improving circulation, reducing pain (including acute pain, chronic pain, wound pain, pain from cramps, and pain from burns), slowing the aging process, and enhancing energy levels. One website describes a pulsed magnetic field device that will give the user more energy: ‘Stiff and sore? With pulsed magnetic fields no more! Have you ever noticed that pain drains you of your energy? You won’t believe how easy it is to lie down and regain energy and relief with the Magnopro. You don’t have to work out, don’t have to meditate, don’t have to do yoga or all the things the practitioners have asked you to do. All you have to do is lie down while opening your mail, talking on the phone, reading or watching TV. You don’t have to move your body. Yes, you can really have more energy without much effort using pulsed magnetic fields . . .’ [26].

MAGNETS

329

These products are in many cases devices regulated by FDA, with specific oversight determined by the product design, claims, and perceived risks. However, they are often sold and advertised by firms that do not seek proper FDA review. Several published studies describe clinical trials of magnets. The results of these studies, whether they show an effect or lack of effect for magnets, are typically contested by others. For example, a randomized, double-blind, placebo-controlled cross-over pilot study examined the effect of bipolar magnets on lower back pain [27]. Nineteen men and women with stable lower back pain of duration 6 months and greater were eligible and participated in the study. Real (300 Gauss) and sham magnetic devices which were similar in appearance were applied to the patient’s skin, covered with cloth, and then wrapped with a smooth gold-colored foil. Participants underwent 6 hour treatments three times per week for 2 weeks. They rated their pain on a visual analogue scale before and after each treatment. At the end of each week’s treatment, subjects responded to the Pain Rating Index of the McGill Pain Questionnaire and were rated for range of motion in the lumbrosacral spine by a clinician. There were no statistically significant differences in any of the outcome measures between magnet and sham treatments. Others commenting on this study argue that the study was flawed because of sex bias (19 men, 1 woman), selection bias (disabled, retired, veterans), and that the intervention was weak because the subjects only wore the permanent static magnets (300 Gauss) 6 hours per day, 3 days per week, and the fact that the magnets may have slipped during the treatment [28]. This simply emphasizes the difficulty of performing studies on alternative treatments: there may be no standards for treatments or methods for applying them, so that any negative finding may be construed as being due to inappropriate use of the treatment, rather than a generally ineffective treatment. That being said, a recent study of follow-up on outcomes for patients reporting to two general practices in Manchester with lower back pain indicated that 3 months after their first consultation, only 23% reported being free of lower back pain and disability [29]. While bipolar permanent magnet treatment for persistent low back pain is of questionable value, allopathic treatment may not be much more effective in some cases. It may be worth examining whether unproven alternative devices or treatments are more likely to thrive when effective allopathic devices or treatments are lacking. Another example of a condition with claims of treatment by magnets is diabetic neuropathy, a painful and disabling complication of high blood glucose, which is treated with drugs, including antidepressants, anticonvulsants, and analgesics, with mixed results. A small single-site pilot study of the effectiveness of magnetic foot insoles for the treatment of pain due to diabetic neuropathy reportedly showed a reduction in symptoms of burning and numbness and tingling [30]. Consecutive patients (14) with diagnosed advanced chronic peripheral neuropathic pain secondary to diabetes, who had failed conventional pharmacologic treatments, and 10 patients with peripheral neuropathies not related to diabetes, were recruited for this study. These patients wore active (475 Gauss) or sham magnetic insoles on opposing feet (in shoes). They rated the level of pain on a continuous five-point visual analog scale twice daily for each foot. After 30 days, the active and sham insoles were switched and after the second month, all insoles were switched to active magnets. Nineteen of the 24 subjects finished the

330

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

study – the other five withdrew for various reasons, including foot surgery, inability to ‘tolerate application to their painful feet’, and loss to follow-up. These patients were not included in the analysis, although arguably they should have been considered failures. In addition, not all the patients were randomized with respect to which side received the active magnet, as required by the protocol, and neither patients nor examiner were blinded as to which side had an active magnet. The authors report that 90% of the diabetic peripheral neuropathy patients had a statistically significant reduction in neuropathic pain compared to the non-diabetic neuropathy group. However, it is not clear what magnitude of pain change on a five-point scale would be significant clinically, neither is it clear that the comparison group is an appropriate control group. It is also unclear from the methods section whether participants were currently on medication for peripheral neuropathy – although the method section does state that there were no new pharmacologic interventions allowed during the study. A more sophisticated clinical trial of magnetic field therapy for diabetic neuropathy, led by the same investigator, was performed at 48 centers in 27 states and enrolled 375 participants [31]. While an improvement over the earlier study, this clinical trial also had certain weaknesses. In contrast to the preliminary study, patients were randomly assigned to the treatment or sham group. The report explicitly mentioned that patients were not to take any new analgesic drugs but could continue any medication for neuropathic pain that they already used. Patients wore a magnetized insole (450 Gauss) or a sham insole which looked similar. Participants rated their pain on an 11 point scale, three times daily for the 16 week period. They also rated their sleep (whether disturbed because of neuropathic pain) on a visual analogue scale and reduction of exerciseinduced foot pain. The drop-out rates for the 199 member treatment group and the 176 member sham group were 23% and 25%, respectively. Foot pain scores decreased 31% for the treatment group and 25% for the sham group (baseline compared to month 4), a statistically significant decline.However, comparison of the mean ratings for groups suggests that clinically this was not a notable improvement (both groups had baseline pain ratings of 5.8 2.3; at 4 months the active magnet group pain rating was 4.1 2.7 and the sham group 4.3 2.8). Sleep disturbance secondary to neuropathic pain also declined from the baseline in both magnet and sham groups, but this difference was not statistically significant. Note that treatment was not significantly different from placebo results. An issue which repeatedly arises in these clinical studies is blinding patients and clinicians to the patient’s treatment. Clinicians, study personnel, or study participants may check to see whether metal objects (such as paper clips) stick to the surface of the test device or whether the test device (such as an insole) sticks to the refrigerator door. In one study, there were efforts to improve blinding by arranging magnets in both the test and placebo device so that they emitted a comparable magnetic field on the external surface facing away from the treatment area, without exposing the treatment area to a magnetic field which exceeded background magnetic field in the placebo device [32]. This study was designed to test the feasibility of studying static magnetic therapy for treatment of osteoarthritis pain and to assess whether the sham device design was effective at concealing the group assignment from patients. Twenty-six patients were randomly

ADVERSE EVENTS ASSOCIATED WITH ALTERNATIVE OR COMPLEMENTARY DEVICES

331

assigned to the active magnet or placebo group. Thirteen patients received active magnets (maximum 100 Gauss/mm) sewn into a knee sleeve and the placebo group received the placebo knee sleeve. The groups were comparable on baseline patient characteristics, except for the Kellgren/Lawrence X-ray grade, a scale to measure the degree of osteoarthritis, with patients with placebo magnets having had significantly lower grades than those in the active treatment group. Patients were encouraged not to use their usual analgesic medications for several days prior to starting the treatment or were given acetaminophen but asked not to take it for 48 hours prior to each visit (although non-compliance with this did not disqualify participants from the study). Patients were asked to wear their knee sleeve for 6 hours per day. Pain was assessed at baseline, 4 hours, 1 week, and 6 weeks after the start of the trial. Patients responded to the five-item pain subscale from the Western Ontario and McMaster Universities (WOMAC) Osteoarthritis Index and demonstrated their pain on a visual analogue scale. After 4 hours of treatment, patients with active magnets reported a small but statistically significant change in pain compared to the placebo group. However, at the 1 and 6 week follow-ups, there was no difference between pain reported by the active magnet and the sham placebo treatment groups. Concern over the participants discovering their treatment group was borne out by the finding that 69% of the active magnet group and 46% of the placebo group admitted to either intentionally testing for, or inadvertently observing, their device’s magnetic properties. In this study [32], as in others [27,31], there are reductions in pain ratings for both active treatment and sham control groups over time. A meta-analysis of 130 clinical trials that compare a placebo to no treatment [33] indicates that placebo is not an effective treatment when binary outcomes are used but shows some effectiveness when continuous outcomes are used, particularly for subjective outcomes, such as pain relief. The placebo effect was diminished by larger sample size in this study. This is of interest, since in most of the clinical trials cited above, the placebo/sham treatment was virtually as effective as the active treatment. This study re-emphasizes the importance of successful blinding, adequate sample size, and appropriate outcome measures for an honest evaluation of the effectiveness of complementary or alternative devices. It is notable that studies for effectiveness for magnets are typically for medical issues that involve pain for which conventional medicine does not provide an adequate solution: diabetic neuropathy [30,31], chronic back pain [27], plantar heel pain [34], and fibromyalgia [35], to name a few. These ailments are notoriously resistant to successful treatment.

Adverse events associated with alternative or complementary devices While there are many advocates for treatments with alternative and complementary devices, there is no successful central repository for collecting information on adverse events [36]. For most of these devices, there is little (if any) information in the medical literature. Because many of these products have low safety risks and often do not receive

332

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

priority from FDA, reporting is not well enforced and adverse event reports are not typically reported and entered into the FDA Adverse Event Reporting System, where they could be used to signal widespread adverse events due to the use of these devices. Approved or cleared devices have a product code that is assigned to them at the time of approval or clearance [37]. Products in reported adverse events are classified with these codes. Because there are no codes for unapproved or uncleared devices, even if adverse events are reported to the FDA, they may not be entered into the surveillance system and, if they are, they may end up under diverse product codes. Finding them under these circumstances would be very unlikely. There are clear-cut examples of adverse events directly related to the use of the alternative therapy, such as in burns caused by ear candling [22] or pain at an acupuncture site [38]. Known adverse events may be minor and easily reversible, for instance, the reported adverse effect of mild discomfort caused by wearing a magnetic knee sleeve reported in a clinical trial, which was reversed by limiting or discontinuing wear [32]. There is also concern over the potential for adverse interactions between alternative and allopathic devices. Examples of this would be the potential interaction of magnets with insulin pumps [31] or pacemakers [32]. In some cases, the only known adverse event associated with the use of an alternative device is that its use might discourage or prevent patients from seeking allopathic therapies. For instance, patients using alternative treatment for cancer may be discouraged from seeking conventional treatment and succumb to cancer [39]. Proponents for alternative medicine argue that Western medicine can fail patients who will additionally (needlessly) experience the adverse side effects of allopathic therapies, which may be serious, and then succumb to cancer despite treatment. The lack of clear evidence-based medical practice or supporting health technology assessment to establish the safety or effectiveness of medical devices or drugs is used as an argument that alternative medical devices or practices are as good as allopathic medical practice (see e.g. [40]). In fact, in these cases, they are only equivalent in the lack of evidence supporting use. This is a strong argument for the importance of evidencebased medicine – a standard which should apply to any treatment, whether it is allopathic or alternative.

References 1. Eskinazi D, Hoffman FA. Progress in complementary and alternative medicine: contributions of the National Institutes of Health and the Food and Drug Administration. J Altern Complement Med 1998; 4: 459–467. 2. Van Haselen R, Fisher P. Evidence influencing British health authorities’ decisions in purchasing complementary medicine [letter]. J Am Med Assoc 1998; 280: 1564–1565. 3. Eisenberg DM, Davis RB, Ettner SL, Appel S et al. Trends in alternative medicine use in the United States, 1990–1997: results of a follow-up national survey. J Am Med Assoc 1998; 280: 1569–1575. 4. Goldbeck-Wood S. Complementary medicine is booming worldwide. Br Med J 1996; 313: 131–133.

REFERENCES

333

5. Guidance for industry, FDA reviewers/staff, and compliance-guidance documents for powered muscle stimulator 510(k)s: http://www.fda.gov/cdrh/ode/2246.html 6. Consumer information on electrical muscle stimulators: http://www.fda.gov/cdrh/consumer/ ems.html 7. Abdominal muscle stimulators. What price perfection?: http://www.fda.gov/cdrh/fdaandyou/ issue01.html#4 8. FDA 510(k) database: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm [accessed January 7 2004]. 9. Code of Federal Regulations. Food and Drugs, Title 21, Chapter 1, Part 880.5580. Accupuncture needle. 10. Kent GP, Brondum J, Keenlyside RA, LaFazia LM, Scott HD. A large outbreak of acupunctureassociated hepatitis B. Am J Epidemiol 1988; 127: 591–598. 11. Birch S, Hesselink JK, Jonkman FA, Hekker TA, Bos A. Clinical research on acupuncture. Part 1. What have reviews of the efficacy and safety of acupuncture told us so far? J Altern Complement Med 2004; 10: 468–480. 12. Yamashita H, Tsukayama H, White AR, Tanno Y et al. Systematic review of adverse events following acupuncture: the Japanese literature. Complement Ther Med 2001; 9: 98–104. 13. White A, Hayhoe S, Hart A, Ernst E; British Medical Acupuncture Society (BMAS) and Acupuncture Association of Chartered Physiotherapists (AACP). Survey of adverse events following acupuncture (SAFA): a prospective study of 32 000 consultations. Acupunct Med 2001; 19: 84–92. 14. White A, Hayhoe S, Hart A, Ernst E. Adverse events following acupuncture: prospective survey of 32 000 consultations with doctors and physiotherapists. Br Med J 2001; 323: 485–486. 15. Yamashita H, Tsukayama H, White AR, Tanno Y et al. Systematic review of adverse events following acupuncture: the Japanese literature. Complement Ther Med 2001; 9: 98–104. 16. MacPherson H, Thomas K, Walters S, Fitter M. A prospective survey of adverse events and treatment reactions following 34 000 consultations with professional acupuncturists. Acupunct Med 2001; 19: 93–102. 17. MacPherson H, Scullion A, Thomas KJ, Walters S. Patient reports of adverse events associated with acupuncture treatment: a prospective national survey. Qual Safety Healthcare 2004; 13: 349–355. 18. Endres HG, Molsberger A, Lungenhausen M, Trampisch HJ. An internal standard for verifying the accuracy of serious adverse event reporting: the example of an acupuncture study of 190 924 patients. Eur J Med Res 2004; 9: 545–551. 19. White A. A cumulative review of the range and incidence of significant adverse events associated with acupuncture. Acupunct Med 2004; 22: 122–133. 20. Fujiwara H, Taniguchi K, Takeuchi J, Ikezono E. The influence of low frequency acupuncture on a demand pacemaker. Chest 1980; 78: 96–97. 21. Health, wealth, and happiness: http://www.relfe.com/ear_candles.html 22. Seely DR, Quigley SM, Langman AW. Ear candles – efficacy and safety. Laryngoscope 1996; 106: 1226–1229. 23. Ear candling. It’s your health: http://www.hc-sc.gc.ca/english/pdf/iyh/candeling-e.pdf 24. Harris R. Untoward otological and audiological consequences of ear candling: http://www.rcsullivan.com/www/forum/harris/candle.htm 25. Manufacturer and user facility device experience (MAUDE) database: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfMAUDE/search.CFM 26. Magnets and health: http://dedi130.your-server.de/maghelth/catalog/index.php? osCsid¼4477c9f2124c15a569a2ce1d1f3d6d7c [accessed February 22 2005].

334

CH21

ALTERNATIVE AND COMPLEMENTARY MEDICAL DEVICES

27. Collacott EA, Zimmerman JT, White DW, Rindone JP. Bipolar permanent magnets for the treatment of chronic low back pain: a pilot study [preliminary communication]. J Am Med Assoc 2000; 283: 1322–1325. 28. Weintraub MI. Are magnets effective for pain control? [letter]. J Am Med Assoc 2000; 284: 564– 566. 29. Croft PR, Macfarlane GJ, Papageorgiou AC, Thomas E, Silman AJ. Outcome of low back pain in general practice: a prospective study. Br Med J 1998; 316: 1356–1359. 30. Weintraub MI. Magnetic biostimulation in painful diabetic peripheral neuropathy: a novel intervention – a randomized, double-placebo crossover study. Australasian Journal of Podiatric Medicine (AJPM) 1999; 9: 8–17. 31. Weintraub MI, Wolfe GI, Barohn RA, Cole SP et al.; Magnetic Research Group. Static magnetic field therapy for symptomatic diabetic neuropathy: a randomized, double-blind, placebocontrolled trial. Arch Phys Med Rehabil 2003; 84: 736–745. 32. Wolsko PM, Eisenberg DM, Simon S, Davis RB et al. Double-blind placebo controlled trial of static magnets for the treatment of osteoarthritis of the knee: results of a pilot study. Altern Ther Health Med 2004; 10: 36–43. 33. Hrobjartsson A, Gotzsche PC. Is the placebo powerless? N Engl J Med 2001; 344: 1594–1602. 34. Winemiller MH, Billow RG, Laskowski ER, Harmsen WS. Effect of magnetic vs. shammagnetic insoles on plantar heel pain. J Am Med Assoc 2003; 290: 1474–1478. 35. Holdcraft LC. Complementary and alternative medicine in fibromyalgia and related syndromes. Best Pract Res Clin Rheumatol 2003; 4: 667–683. 36. Biley FC. Primum non nocere: thoughts on the need to develop an ‘adverse events’ register for complementary and alternative therapies. Complement Ther Nurs Midwifery 2002; 8: 57–61. 37. Classification database: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPCD/ classification.cfm?ID¼2551. 38. MacPherson H, Scullion A, Thomas KJ, Walters S. Patient reports of adverse events associated with acupuncture treatment: a prospective national survey. Qual Saf Healthcare 2004; 13: 349– 355. 39. Rife Machine Operator Sued: http://www.quackwatch.org/04ConsumerEducation/News/ rife.html 40. Entrenched medical practices: time to take action. HealthFacts, August 1990: http://www.findarticles.com/p/articles/mi_m0815/is_n135_v15/ai_9324637 41. Quack Electrodiagnostic Devices: http://www.quackwatch.org/01QuackeryRelatedTopics/electro.html 42. Gas Grill Igniters: the Stimulator and Crystaldyne Pain Reliever: http://www.mtn.org/quack/ devices/stimul.htm 43. FDA Consumer Alert on the Stimulator: http://www.fda.gov/bbs/topics/ANSWERS/ ANS00817.html 44. Relaxacisor warning: http://www.mtn.org/quack/devices/relaxacisor.htm 45. Plaintiff Federal Trade Commission’s complaint against David L. Walker: http://www.ftc.gov/ os/2002/04/davidwalkercmp.pdf 46. Radionics home study: http://home.earthlink.net/cj31/ 47. What is radionics?: http://www.copenlabs.com/whatsradionics.htm#medicine 48. FDA warning letter: http://www.fda.gov/foi/warning_letters/m5140n.pdf

22 Drug-eluting coronary stents Hesha J. Duggirala, and Thomas P. Gross US Food and Drug Administration, Rockville, MD, USA

David E. Kandzari Duke Clinical Research Institute, Durham, NC, USA

In April 2003, the US Food and Drug Administration’s (FDA) approval of the first drugeluting stent (DES) was followed by widespread adoption of this novel therapy for percutaneous coronary revascularization. Drug-eluting stents are a good example of the epidemiology of break-through technologies and implantable devices. Break-through technologies: (a) transform the standard of care; (b) are immediately adopted; and (c) experience widespread use. The latter is particularly important from a public health standpoint, from both a risk and benefit perspective. This chapter discusses the unique clinical and regulatory considerations for breakthrough technologies. We provide a description of the differences between bare metal and drug-eluting stents and discuss the epidemiologic studies showing the clinical benefits of drug-eluting stents. A device that penetrates the market rapidly poses a unique potential for public health concerns (especially implantables); therefore, there is a need for increased postmarket rigor commensurate with the potential for risk. We present the postmarket regulatory requirements for drug-eluting stents and discuss some postmarket issues that have come to light since the first DES was approved.

History and evolution of coronary stents: clinical perspective Each year, over 1 million percutaneous coronary interventions are performed worldwide, surpassing the annual number of coronary artery bypass operations. Coronary Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

336

CH22

DRUG-ELUTING CORONARY STENTS

stents are generally permanently placed to keep the treated vessel open after angioplasty. In the USA, use of coronary stents in interventional procedures increased from 10% in 1994 to over 90% by 2002 [1]. Compared with balloon angioplasty, stents significantly reduce recurrent ischemia, restenosis, and revascularization. Nevertheless, despite the enthusiasm for stenting in percutaneous coronary revascularization, angiographic restenosis rates (range 20–50%) due to intimal hyperplasia with conventional bare metal stents remain unacceptably high. The persistent problems of intimal hyperplasia and thrombosis following stent implantation have motivated the design of stents with biocompatible coatings to serve as a matrix for adjunctive drug delivery. In most instances, a conventional stainless steel or cobalt alloy stent (‘bare metal stent’) serves as the scaffolding over which an antiproliferative agent is applied by either ionic bonding, incorporation into a polymer matrix, or fixation to a polyamine layer adsorbed to the stent surface. The rationale for the development of drug-eluting stents includes the following. First, local pharmacologic delivery maximizes the drug effect where it is required, allowing for controlled release and reducing the potential for systemic toxicity. Second, as a combined, one-step method, drug-loaded stents may also offer a savings benefit in time and cost. Finally, the potential for stents as drug delivery devices has broad applications in both de novo and restenotic coronary disease in addition to peripheral vascular disease.

Stents as antiproliferative agents: results from first in man to pivotal clinical trials Pharmacologic agents delivered to the site of vascular injury via DES are designed to address the continuing problem of restenosis. At present, both observational registries and randomized clinical trials comparing DES with bare metal stents have examined the safety and efficacy of sirolimus- and paclitaxel-eluting stents (SES and PES, respectively) in percutaneous coronary intervention (PCI).

Sirolimus Sirolimus (rapamycin or RapamuneTM) is a natural macrocyclic lactone approved initially as a potent immunosuppressive agent for the prevention of renal transplant rejection. In preclinical studies, sirolimus exhibits antiinflammatory properties and is effective in preventing the smooth muscle cell proliferation and intimal thickening induced by mechanical injury. Sirolimus binds to a smooth muscle cell cytosolicreceptor protein and elevates p27 levels, leading to the inhibition of cyclin-dependent kinase complexes and preventing cell growth. By preventing cell replication, the drug is considered cytostatic rather than cytotoxic. In the ‘First in Man’ trial, 30 patients were treated with a BX VelocityTM stent coated with sirolimus in one of two preparations, slow- (SR) and fast-release (FR) [2]. At 4 month angiographic follow-up, no in-stent or edge restenosis was observed. Intimal

HISTORY AND EVOLUTION OF CORONARY STENTS: CLINICAL PERSPECTIVE

337

hyperplasia measured by intravascular ultrasound was minimal (11:0 3:0% in the SR group and 10:4 3:0% in the FR group; p ¼ NS), and late loss was negligible (in-stent late loss measured by quantitative angiography, 0:09 0:3 mm [SR] and 0:02 0:3 mm [FR]). Clinical follow-up at 12 months revealed no major adverse events and no restenosis (defined as > 50% reduction in minimal luminal diameter [MLD]) [3]. Four-year follow-up showed that revascularization occurred in two patients, with one patient undergoing repeat PCI for restenosis at the proximal edge of a sirolimus stent [4]. Among 26 patients undergoing angiographic follow-up at 4 years, late lumen loss measured by intravascular ultrasound was 0:09 0:23 mm in the SR group compared with 0:41 49 mm in the FR group, a distinction that led to further clinical trial development with the SR preparation. Although the initial clinical experience with SES involved a small number of patients, these promising results led to the development of larger pivotal, placebo-controlled, randomized clinical trials. In the RAVEL trial, 238 patients with a de novo coronary lesion 2.5–3.5 mm in diameter and < 18 mm in length were randomized in a doubleblinded fashion to the rapamycin-coated BX VelocityTM stent (Cypher1 stent, Cordis Corporation, Miami Lakes, FL) or the bare metal BX VelocityTM stent [5]. At 6 months, the binary rates for restenosis were 0% for the SES and 26.6% for the uncoated stents (p < 0:001). In-stent late loss was 0:80 0:53 mm in the control uncoated stent group and 0:01 0:33 mm for the SES group (p < 0:001). One-year rates of target lesion revascularization (TLR) were 24% in the control group and 0% in the sirolimus-coated group (p < 0:001). Overall event-free survival was 94.1% in the sirolimus group and 70.9% among patients receiving the uncoated stent. In addition, the observed reduction in restenosis was independent of vessel diameter. At 3 years, event-free survival of TLR was 93.7% for patients randomized to SES vs. 75.0% for those assigned to the bare metal stent group (p < 0:001) [6]. In February 2001, the SIRIUS trial was initiated, randomizing 1101 patients undergoing elective percutaneous coronary intervention in a blinded fashion to either the sirolimus-eluting Cypher1 stent or the bare metal BX VelocityTM stent [7] (Table 22.1). The SIRIUS trial enrolled patients with more complex lesions and greater cardiovascular comorbidity than did the RAVEL trial. At 8 month angiographic follow-up, in-stent binary restenosis (defined as diameter stenosis at follow-up 50%) occurred in 3.2% of patients treated with sirolimus-eluting stents and 35.4% of patients receiving bare metal stents (p < 0:001). However, inclusion of 5 mm on either end of the stent as part of the ‘in-segment’ angiographic analysis raised the binary rates for restenosis to 8.9% and 36.3%, respectively (p < 0:001). The primary endpoint of 9-month target vessel failure (TVF), defined as the composite of death, myocardial infarction, or target vessel revascularization, was 8.8% and 21.0% for the sirolimus-eluting and bare metal stent cohorts, respectively (p < 0:001). At 1 year follow-up, TLR occurred in 4.9% of patients in the sirolimus-eluting group and 20.0% of patients assigned to treatment with bare metal stents (p < 0:001) [8]; 1 year rates of TVF were 9.8% of SES patients compared with 24.8% of patients treated with bare metal stents (p < 0:001). Despite the use of overlapping stents (up to two 18 mm stents) and a 3 month prescription of combined antiplatelet therapy, stent thrombosis rates were comparable among both groups

338

CH22

DRUG-ELUTING CORONARY STENTS

Table 22.1 Randomized trials comparing drug-eluting stents with bare metal stents Trial

(n)

SIRIUS [7]

1058

TAXUS IV [17]

1314

ENDEAVOR 1197 II [21]

Stents compared

Primary endpoint

Outcome

Polymer-based, 9 Month Target vessel failure (21.0% vs. 8.6%; sirolimus-eluting target p < 0:001) and angiographic binary stent (Cypher1) vessel in-segment restenosis (36.3% vs. 8.9%; vs. bare metal, failure* p < 0:001) significantly decreased with stainless steel sirolimus-eluting stent stent (BX VelocityTM) Polymer-based 9 Month Target vessel revascularization (12.0% paclitaxel-eluting ischemia- vs. 4.7%; p < 0:001) and angiographic stent (TAXUS1) driven binary in-segment restenosis (26.6% vs. bare metal, target vs. 7.9%; p < 0:001) significantly stainless steel vessel reduced with paclitaxel-eluting stent stent (ExpressTM) revascularization Polymer-based ABT-578-eluting 9 Month Target vessel failure (15.4% vs. 8.1%; (sirolimus analog) target p < 0:0005) and angiographic binary stent vessel in-stent restenosis (32.7% vs. 9.5%; (ENDEAVORTM) failure* p < 0:0001) significantly decreased vs. cobalt alloy with ABT-578-eluting stent stent (DriverTM)

*Target vessel failure defined as any cardiovascular death, myocardial infarction or repeat percutaneous or surgical revascularization of the target vessel.

(0.8% and 0.4% for SES and bare metal stents, respectively; p ¼ 0:45). Given the favorable safety and efficacy of the sirolimus-eluting stent, the Cypher1 stent received European CE mark approval in April 2002 followed by US FDA approval in April 2003. In addition to the SIRIUS trial, the corollary European SIRIUS (e-SIRIUS) and Canadian (c-SIRIUS) trials were performed [9,10]. Each trial randomized patients (e-SIRIUS, n ¼ 352; c-SIRIUS, n ¼ 100) to either the sirolimus-eluting stent or the bare metal stent to examine the primary endpoint of in-stent minimal luminal diameter at 8 month angiographic follow-up. In both studies, the efficacy of SES was further confirmed, with very low rates of angiographic restenosis (5.9% E-SIRIUS, 2.3% C-SIRIUS) and TLR (4.0% E-SIRIUS, 4.0% C-SIRIUS).

Paclitaxel In spite of an unexpectedly high rate of in-stent thrombosis and late restenosis with a taxane derivative in the preliminary SCORE trial [11], several studies have since been performed demonstrating the safety and efficacy of stents coated with taxol formulations

HISTORY AND EVOLUTION OF CORONARY STENTS: CLINICAL PERSPECTIVE

339

to prevent restenosis. Paclitaxel is a chemotherapeutic agent that interrupts cell division by stabilizing microtubule formation. Preliminary animal studies evaluating paclitaxelcoated stents have demonstrated dose-dependent inhibition of neointimal proliferation, although higher doses may be associated with medial toxicity and stent thrombosis [12]. In comparison with SES trials, studies evaluating PES have been more varied in trial design and with regard to the dose and formulation of drug elution. In the ELUTES study, 190 patients were randomized to stents eluting one of five, non-polymer-based, paclitaxel dose densities (Cook Inc., Bloomington, IN) [13]. For all angiographic parameters, a clear dose-dependent response was observed. At 6 months, angiographic restenosis (> 50% decrease in the minimum luminal diameter) occurred in 21% of patients receiving the bare stent and 3% of patients treated with the 2.7 mg/mm2 stent (p ¼ 0:055). Late lumen loss was 0.73 mm for the control stent and 0.10 mm for the 2.7 mg/mm2 stent (p < 0:005). At 1 year, rates of TLR were 5% and 16% for patients receiving the 2.7 mg/mm2 and uncoated stents, respectively. The absence of late stent thromboses or additional deaths further confirmed that both efficacy and safety were sustainable with paclitaxel as an antirestenotic therapy. The DELIVER trial was a prospective, randomized trial evaluating the non-polymerbased paclitaxel-eluting MultiLink Penta stent (Achieve stent, Guidant Corporation, Santa Clara, CA) in 1043 patients undergoing elective percutaneous coronary revascularization [14]. Neither 9 month TVF (11.9% vs. 14.5%; p ¼ 0:12) nor restenosis at 8 month angiographic follow-up (14.9% vs. 20.6%; p ¼ 0:076) were significantly reduced with the paclitaxel-coated stent. As a result of these findings, further clinical development of this particular DES program has been halted. Most clinical development with PES has been associated with the TAXUS program (Boston Scientific, Natick, MA), evaluating a nonerodable, polymer-based PES. Based upon favorable clinical and angiographic outcomes in the TAXUS1 I and II trials [15,16], the pivotal, double-blind TAXUS IV trial was performed [17] (Table 22.1), randomizing 1326 patients to treatment with either the slow-release paclitaxel-eluting ExpressTM stent (TAXUS1 stent) or the control bare metal stent. At 8 month angiographic and 9 month clinical follow-up, both in-segment restenosis (7.9% vs. 26.6%; p < 0:0001) and TLR (3.0% vs. 11.3%; p < 0:0001) were significantly reduced with the TAXUS1 stent. At 9 month clinical follow-up, the occurrence of stent thrombosis did not significantly vary between groups (0.6% with PES vs. 0.8% with bare metal stent; p ¼ 0:75). As a result of these findings, use of the paclitaxel-eluting TAXUS1 stent was approved by the FDA in March 2004 for the treatment of de novo coronary lesions 28 mm and in vessels 2.5–3.75 mm in diameter. TAXUS VI was a randomized study (n ¼ 446 patients) performed outside the USA examining the safety and efficacy of the moderate-release TAXUS1 stent (1.0 mg/mm2 paclitaxel in a nonerodable polymer) [18]. Compared with previous TAXUS trials, patients enrolled in TAXUS VI had more complex lesion and clinical characteristics. Overall, the average lesion length was 20.6 mm, mean stent length was 22.4 mm, small vessels (< 2.5 mm diameter) were identified in 27.8%, and overlapping stents occurred in 27.8% of patients. The primary endpoint, 9 month target vessel revascularization (TVR), was significantly reduced with the TAXUS1 stent (9.1% vs. 19.4%; p ¼ 0:0027).

340

CH22

DRUG-ELUTING CORONARY STENTS

Treatment with PES appeared to be consistent among patients with higher-risk features for restenosis; for example, among patients with long lesions ( 26 mm) and with overlapping stents, TLR was significantly reduced by 85% and 93%, respectively. In 2005, the TAXUS V study (n ¼ 1172) examined the safety and efficacy of the slowrelease TAXUS1 stent in a broader patient and lesion subset [19]. As an expanded indication study, inclusion criteria allowed treatment of de novo coronary lesions in vessels of diameter 2.25–4.0 mm and lesion lengths 10–46 mm. Treatment with overlapping stents was permitted. Among patients randomized to PES, 31.7% of patients had diabetes, the mean lesion length was 17.3 mm, and 34.1% of patients received more than one stent. At 9 month clinical follow-up, TLR was significantly reduced with the TAXUS1 stent (15.7% vs. 8.6%; p ¼ 0:0003). In-stent restenosis was also significantly reduced with PES (13.7% vs. 31.9%; p < 0:0001). Despite consistent reductions in restenosis and TLR, however, the trial also raised concerns regarding safety among patients treated with multiple and/or overlapping stents. In particular, 30 day rates of major cardiac adverse events were significantly higher among patients with multiple stents (8.3% vs. 3.3%; p ¼ 0:047). Although this difference was attributed to more commonly reduced flow in side branch vessels among patients treated with overlapping stents, the exact mechanisms for these differences are uncertain.

Novel drug-eluting stent programs Aside from sirolimus and paclitaxel, several other antiproliferative agents have been evaluated in recently completed and/or forthcoming clinical trials (i.e. mid-2003 to mid2005) designed not only to better understand their safety and efficacy but also to facilitate regulatory approval for market use.

ABT-578 (Zotarolimus) As a sirolimus analog, ABT-578 is a novel lipophilic, tetrazole-containing macrocyclic immunosuppressant that possesses both antiproliferative and anti-inflammatory characteristics. Despite a subtle difference in chemical structure, ABT-578 functions similarly to sirolimus to bind to the intracellular mammalian target of rapamycin, inhibit cellular signal transduction, and thus prevent vascular smooth muscle cell proliferation. Recent clinical experience with coronary stents eluting ABT-578 (Abbott Laboratories, Abbott Park, IL) has demonstrated comparable clinical efficacy and safety to approved SES and PES clinical trial programs. In the preliminary ENDEAVOR I trial, 100 patients were treated with the ABT-578-eluting ENDEAVOR stent (Medtronic, Inc., Santa Rosa, CA; concentration 10 mg/mm stent length) using a phosphorylcholine polymer [20]. At 1 year clinical and angiographic follow-up, angiographic restenosis occurred in five patients, and TLR was performed in two patients. No episodes of late thrombosis, late stent malapposition or aneurysm formation was observed. Based on this favorable feasibility study, the pivotal ENDEAVOR II trial was performed, in which

NOVEL DRUG-ELUTING STENT PROGRAMS

341

1197 patients were randomized in a prospective, double-blind fashion to treatment with either the ENDEAVOR stent or a control bare metal cobalt alloy stent [21] (Table 22.1). At 8 month angiographic follow-up, both in-stent restenosis (9.5% vs. 32.5%; p < 0:0001) and in-segment late lumen loss (0.36 mm vs. 0.71 mm; p < 0:0001) were significantly reduced with the ENDEAVOR stent. Accordingly, the primary endpoint of 9 month TVF was significantly lower with the ENDEAVOR stent (8.1% vs. 15.4%; p < 0:0005), principally driven by reductions in the need for repeat TLR (4.6% vs. 12.1%; p < 0:0001). Aside from the ENDEAVOR program, ABT-578 and the phosphorylcholine polymer are currently under evaluation in the Zomaxx clinical trial program, in which the pharmacokinetic release profile of the drug and the stent platform differ from the ENDEAVOR stent.

Everolimus Everolimus (Novartis Pharma AG, Basel, Switzerland) is another antiproliferative agent with similar pharmacodynamic properties to sirolimus and its analogs. Combined with a non-erodable polymer, everolimus was evaluated in 48 patients in the randomized FUTURE I and FUTURE II trials (Guidant Corporation, Santa Clara, CA) [22]. Compared with bare metal stents, treatment with the everolimus-eluting stent was associated with statistically significant reductions in 6 month angiographic restenosis (27.7% vs. 4.3%; p < 0:05) and in-segment late loss (0.5 mm vs. 0.17 mm; p < 0:05). Similarly, the preliminary SPIRIT FIRST trial randomized 60 patients to treatment with the bare metal cobalt alloy MultiLink Vision stent or the everolimus-eluting Vision stent (Guidant Corporation) [23]. At 6 month angiographic follow-up, no restenosis occurred with the everolimus-eluting stent (26.9% vs. 0; p ¼ 0:01), and in-segment late loss was also significantly reduced (0.10 mm vs. 0.84 mm; p < 0:0001). Current trials with everolimus are designed to compare an everolimus-eluting cobalt alloy stent (Xience V, Guidant Corporation) with approved DES.

Biolimus A9 Clinical experience with biolimus (Biosensors International, Newport Beach, CA), a sirolimus analog, is presently limited. In the STEALTH First in Man trial, 120 patients were randomized in a 2:1 fashion to treatment with either the biolimus-eluting stent or a bare metal control stent [24]. At 6-month angiographic surveillance, both in-segment late loss (0:74 0:45 mm vs. 0:36 0:43; p < 0:001) and in-segment restenosis were lower with the biolimus-eluting stent (7.7% vs. 3.0%; p ¼ 0:40).

Paclitaxel Although not a novel antiproliferative agent, paclitaxel has also been evaluated with a novel stent-based method of drug delivery with the COSTARTM stent (Conor Medsystems,

342

CH22

DRUG-ELUTING CORONARY STENTS

Menlo Park, CA), in which the stent consists of multiple laser-drilled reservoirs that enable bidirectional delivery of one or multiple therapeutic agents to the arterial wall and/or the bloodstream. In the dose-finding PISCES trial [25], controlled, 30 day release of paclitaxel to the mural surface of the vessel was associated with incrementally lower late loss values compared with alternative release patterns or bare metal stents. Aside from ongoing European registry studies with the COSTARTM stent, a randomized, pivotal trial in the USA is currently ongoing.

Comparative trials of drug-eluting stents Until recently, most studies evaluating DES were limited to nonrandomized registries or randomized trials comparing DES with bare metal stents. However, there remains a need to better understand the relative safety and efficacy between approved DES beyond indirect trial comparisons. In addition, given the efficacy of approved DES, treatment with either SES or PES has now been adopted as the clinical measure against which novel DES programs are compared. Four trials randomizing patients to SES or PES for the treatment of de novo coronary lesions have been recently completed (Table 22.2). In the REALITY trial, 1353 patients undergoing elective percutaneous coronary revascularization were randomized to either the Cypher1 or TAXUS1 stent(s) to examine the primary endpoint of in-lesion binary restenosis (defined as 50% stenosis at angiographic follow-up) 8 months following the index procedure [26]. Overall, patient demographic, clinical, and angiographic characteristics did not significantly differ between treatment groups. Although in-lesion restenosis did not statistically differ between SES and PES (9.6% with Cypher1 vs. 11.1% with TAXUS1; p ¼ 0:31), other angiographic indices were significantly improved with the Cypher1 stent, including in-lesion late loss (0:04 0:38 vs. 0:16 0:40 mm; p < 0:001) and in-lesion diameter stenosis (29:1 15:8% vs. 31:1 15:4%; p ¼ 0:009). However, target lesion revascularization did not statistically vary between groups (5.0% with Cypher1 vs. 5.4% with TAXUS1; p ¼ 0:81), raising attention to the relative importance of, or emphasis placed on, differences between angiographic and clinical outcomes. These findings bring attention to the concern that while efficacy may be similar between SES and PES, differences in safety may motivate clinical decision making regarding stent selection. In the TAXi trial, 202 patients were randomized to treatment with either the Cypher1 or TAXUS1 stent(s) [27]. Outcomes including both initial procedural success and major adverse cardiac events did not significantly differ between groups. Target lesion revascularization did not statistically vary and occurred in 1% and 2% of patients in the TAXUS1 and Cypher1 groups, respectively. Unlike the REALITY trial, no differences in stent thrombosis were observed. In the SIRTAX trial, 1103 patients were randomized at two institutions in a singleblinded fashion to treatment with the Cypher1 or TAXUS1 stents [28]. At 9 months, the primary endpoint of major adverse cardiac events was significantly reduced with the Cypher1 stent (6.2% vs. 10.8%; p ¼ 0:009), a difference largely driven by a lower

Polymer-based, sirolimuseluting stent (Cypher1) vs. polymer-based paclitaxel-eluting stent (TAXUS1) Polymer-based, sirolimuseluting stent (Cypher1) vs. polymer-based paclitaxel-eluting stent (TAXUS1) Polymer-based, sirolimuseluting stent (Cypher1) vs. polymer-based paclitaxel-eluting stent (TAXUS1)

1 De novo native coronary lesion(s), 2.25– 4.0 mm diameter

1012

202

250

300

SIRTAX [29]29

TAXi [28]28

ISAR-DIABETES [30]30

ISARDESIRE [40]40

Polymer-based, sirolimuseluting stent (Cypher1) vs. polymer-based paclitaxel-eluting stent (TAXUS1) vs. conventional balloon angioplasty

Polymer-based, sirolimuseluting stent (Cypher1) vs. polymer-based paclitaxel-eluting stent (TAXUS1)

2 De novo native coronary lesions, 2.25–3.0 mm diameter

1353

REALITY [26]26

Treatment of any de novo or restenotic coronary lesion(s) 1 De novo native coronary lesion(s), 2.25– 4.0 mm diameter, in patients with diabetes mellitus In-stent restenosis

Stents compared

Indication

(n)

Trial

6 Month angiographic binary in-segment ( 50%) restenosis

9 Month composite of cardiovascular death, myocardial infarction, ischemia-driven target lesion revascularization 6 Month composite of cardiovascular death, myocardial infarction, ischemia-driven target lesion revascularization 6 Month angiographic in-segment late lumen loss

8 Month angiographic binary in-segment restenosis

Primary endpoint

(Continued)

Angiographic restenosis 44.6% with angioplasty, 14.3% with SES, 21.7% with PES (p 0:01 for any DES vs. angioplasty; p ¼ 0:19 SES vs. DES); target vessel revascularization significantly lower with SES vs. PES (8.0% vs. 19.0%; p ¼ 0:02)

Significant reduction in in-segment late loss with SES (0.43 vs. 0.67 mm; p < 0:001); target lesion revascularization, 6.4% SES vs. 12.2% PES; p ¼ 0:13

No significant difference in primary composite endpoint (6.0% SES vs. 4.0% PES; p ¼ 0:80) or target lesion revascularization (2.0% SES vs. 1.0% PES; p ¼ 0:90)

No significant difference in in-segment restenosis (9.6% SES vs. 11.2% PES; p ¼0:31) or target lesion revascularization (5.0% SES vs. 5.4% PES; p ¼ 0:81); higher stent thrombosis with PES in actual treatment analysis (0.4% vs. 1.8%; p ¼ 0:02) Significant reduction in primary endpoint with SES (6.2% vs. 10.8%; p ¼ 0:009) due to reduction in target lesion revascularization (4.8% vs. 8.3%; p ¼ 0:025)

Outcome

Table 22.2 Randomized trials comparing drug-eluting stent programs

COSTAR II

1700; 3:2 study stent to control randomization

ZOMAXX II 1670

Single de novo native coronary lesion, 2.5–3.75 mm diameter, lesion length 10–28 mm 1 De novo native coronary lesion(s), 2.5–3.5 mm in diameter, lesion length 30 mm

2 De novo native coronary lesion(s), 2.5– 3.5 mm diameter

SPIRIT III

1380; 2:1 study stent to control randomi-zation

Single de novo native coronary lesion, 2.5– 3.5 mm diameter, lesion length 27 mm

ENDEAVOR 1548 IV

Indication Single de novo native coronary lesion, 2.5– 3.5mm diameter, lesion length 14–27 mm

(n)

ENDEAVOR 436 [55]1 III

Trial

Month composite of cardiovascular death, myocardial infarction, ischemia-driven target vessel revascularization

Month composite of cardiovascular death, myocardial infarction, ischemia-driven target vessel revascularization Month angiographic in-segment late lumen loss and 9-month ischemia-driven target vessel revascularization Month ischemiadriven target vessel revascularization

8 Month angiographic in-segment late lumen loss

Polymer-based ABT-578eluting (sirolimus analog) stent (ENDEAVORTM) vs. polymer-based, sirolimus-eluting stent (Cypher1) Polymer-based ABT-5789 eluting (sirolimus analog) stent (ENDEAVORTM) vs. polymer-based paclitaxeleluting stent (TAXUS1) Polymer-based, everolimus- 8 eluting stent (XIENCE V1) vs. polymer-based paclitaxel-eluting stent (TAXUS1) Polymer-based ABT-5789 eluting stent (ZoMaxx1) vs. polymer-based paclitaxel-eluting stent (TAXUS1) Bioresorbable polymer8 based aclitaxel-eluting stent with reservoir wells (COSTARTM) vs. polymerbased paclitaxel-eluting stent (TAXUS1)

Primary endpoint

Stents compared

Table 22.2 (Continued)

Enrollment began May 2005

Enrollment began May 2005

Enrollment began May 2005

In-segment lumen loss with ENDEAVORTM was 0.34 mm at 8 months followup compared with 0.13 mm with SES (p < 0:001) Enrollment began April 2005

Outcome

APPLICATION OF DRUG-ELUTING STENTS IN COMPLEX CORONARY LESION SUBSETS

345

incidence of TLR among patients treated with the Cypher1 stent (4.8% vs. 8.3%; p ¼ 0:025). In-segment late loss was also significantly lower in the SES group compared with the PES cohort (0.19 vs. 0.32 mm; p ¼ 0:02). The ISAR-DIABETES trial also compared Cypher1 with TAXUS1 stents, although exclusively in diabetic patients [29]. Designed as a noninferiority trial, ISARDIABETES examined the primary endpoint of late lumen loss in 250 patients randomized to either SES or PES. At 6 months, patients in the SES arm of the study had significantly lower in-stent (0.19 vs. 0.36 mm; p ¼ 0:002) and in-segment late loss (0.43 vs. 0.67 mm; p < 0:001) and significantly lower rates of angiographic restenosis. Rates of TLR, however, did not differ between the two patient groups (6.4% with Cypher1 vs. 12.2% with TAXUS1; p ¼ 0:13). In addition to direct comparisons between SES and PES, several trials are ongoing, as of mid-2005, to compare novel DES platforms with either the approved Cypher1 or TAXUS1 stents. The recently completed ENDEAVOR III trial randomized 436 patients in a 3:1 fashion to either the ENDEAVOR stent or the Cypher1 stent, and final results are forthcoming regarding the primary endpoint of in-segment late lumen loss at 8 month angiographic follow-up. Similarly, the ongoing ENDEAVOR IV trial is designed to evaluate TVF (death, myocardial infarction and TVR) at 9 months among patients randomized to either the ENDEAVOR or TAXUS1 stent(s). Other ongoing randomized, comparative trials include the SPIRIT III trial (everolimus-eluting XIENCE1 stent vs. TAXUS1), the COSTAR II trial (paclitaxel-eluting COSTARTM stent vs. TAXUS1) and the ZOMAXX II trial (ABT-578-eluting ZoMaxx1 stent vs. TAXUS1).

Application of drug-eluting stents in complex coronary lesion subsets At present, the only DES approved by the US FDA are the sirolimus-eluting Cypher1 stent and the paclitaxel-eluting TAXUS1 stent. In both instances, based principally on the results of the pivotal SIRIUS and TAXUS IV trials, these DESs are indicated for treatment of stenotic de novo coronary lesions in vessels 2.5–3.5 mm diameter. In spite of these somewhat restricted indications, over 2 million patients have been treated worldwide to date with both the sirolimus-eluting and paclitaxel-eluting coronary stents, and the application of DES to broader clinical settings and more complex lesion characteristics is commonplace. For example, among patients with high-risk acute coronary syndromes and non-ST segment elevation myocardial infarction who underwent percutaneous revascularization in the USA, use of DES increased from approximately 50% in October 2003 to nearly 80% of cases in June 2004, despite the fact that DESs have not yet been well studied in this particular clinical setting [30]. Accordingly, there remains a need to better understand the safety and effectiveness of DES in expanded indications. However, not all trials are designed and conducted to demonstrate safety and effectiveness (i.e. promote routine clinical use) in addition to adhering to the requirements for regulatory approval (i.e. establish an indication for use). In addition to recent studies evaluating the effectiveness of DESs in the treatment of

346

CH22

DRUG-ELUTING CORONARY STENTS

small coronary vessels [31,32] and lengthy, more diffuse disease [18,19,33], several trials are ongoing to evaluate DESs in complex lesion subsets, including coronary bifurcation disease [34,35], chronic total coronary occlusions [36–38], in-stent restenosis [39,40] and saphenous vein bypass graft lesions [41]. In the ISAR-DESIRE trial, for example, 300 patients with in-stent restenosis were randomized to treatment with balloon angioplasty alone, SES or PES. Although the need for repeat revascularization was significantly lower with any DES compared with angioplasty, patients randomized to SES had significantly lower clinical restenosis and in-stent late lumen loss [39] (Table 22.2). In addition, randomized trials are ongoing to compare DESs with bare metal stents in primary PCI for acute myocardial infarction (HORIZONS) and with coronary bypass surgery for patients with multi-vessel and/or left main coronary artery disease (SYNTAX). The NIH funded FREEDOM compares drug-eluting stents with CABG in diabetics. If proven effective over long-term, DESs have the potential to broaden the utilization of stents in a variety of clinical settings and substantially improve clinical outcomes for patients undergoing percutaneous coronary revascularization. Considering the encouraging results of early trials of stents coated with antiproliferative agents, the enthusiasm for developing these technologies seems tempered only by concerns over their predicted cost. Amidst the enthusiasm of successive positive results, the early experience with DES has also identified several important unanswered questions. Unresolved issues include identifying the most effective agent(s), determining the optimal duration of antiplatelet therapy, and identifying the best delivery platform (i.e. polymer carrier or not). Moreover, the long-term certainty of preventing restenosis (e.g. in complex lesions, in-stent restenosis, chronic total occlusions) and maintaining safety is not fully established. For sirolimus, concern has been raised over the findings of negative late-lumen loss and stent malapposition during long-term follow-up. In the case of paclitaxel, remaining issues include the long-term effects of a polymer carrier and the optimal paclitaxel dose, considering its narrow therapeutic window. This latter finding, in addition to improper antiplatelet therapy, has been offered as a potential explanation for the trend toward higher rates of late thrombosis in recent clinical studies [42,43]. There are several epidemiologic study designs that can be employed to answer remaining questions of safety and effectiveness for drug-eluting stents. It may be costly and impractical to collect long-term effectiveness data in a randomized clinical trial. Multi-year registries designed as cohort studies are an important way to study issues such as the long-term certainly of preventing restenosis. Nested case-control designs can be incorporated into these cohorts to study emerging and unexpected safety issues.

Drug-eluting stents: regulatory perspective Postapproval requirements Drug-eluting stent manufacturers are required to follow similar postapproval requirements set by the FDA. In the order approving the device for marketing, FDA requires

DRUG-ELUTING STENTS: REGULATORY PERSPECTIVE

347

such manufacturers to collect and report to the FDA, on an annual basis, clinical outcomes through 5 years postprocedure on at least 80% of the patients enrolled in the pivotal trials presented in the premarket application. Manufacturers are also required to study implantation of the product in at least 2000 new US patients postmarket. The objectives of these studies, which are based on patient registries, are to collect safety surveillance and clinical outcomes data for DESs in routine clinical practice. These registries are needed to evaluate the potential for less frequent adverse events and product problems related to the device (including its delivery systems) and/or its use that (a) could not be detected in the initial clinical trials and/or (b) may have resulted from changes in the manufacturing process between the production of the clinical trial lots and commercial scale-up. The company is asked to choose study sites that are ‘representative’ of general ‘real world’ use, i.e. sites that are geographically diverse (including community-based facilities) and facilities with low, medium, and high stent implant volumes. In these postapproval study registries, the sponsor is asked to report extensive data on patient history, procedural characteristics, medications, and clinical outcomes. These data are used by the FDA and the manufacturer to help assess whether adverse events occurring in the ‘real world’ are what would be expected based on the clinical trial experience. In the early postmarket experience with DES, recruitment of study sites was slowed by the human subjects review process. As a result, subsequent DES applications, and some other cardiovascular device applications, have gone through a new ‘peri-approval’ process. This is a novel approach developed at the FDA which allows manufacturers to begin the process of institutional review board (IRB) approval of their required postapproval studies, and in some instances begin patient accrual before device approval. This is an experimental approach to approving postmarket studies and FDA is still considering how it might be used in the future or whether it will be expanded to other devices. Current DES postapproval study requirements, based largely on these patient registries, provide for mostly descriptive, and less analytic (i.e. hypothesis-testing), data. The data rigor may be increased in the future as more is learned from these registries. FDA may ask for more representative samples, larger samples to detect smaller effects, and for control data. In addition to these postapproval study requirements, and because the safety and effectiveness of new therapies are generally less understood when applied to patients outside of clinical trials, the FDA mandates manufacturer reporting of all device-related adverse events and malfunctions. Manufacturers of medical devices, including DESs, are required to report known occurrences to the FDA in accordance with the Medical Device Reporting (MDR) Regulation (21 Code of Federal Regulations Part 803). In addition, in order to better understand all product-related reports to manufacturers (including product complaints), the FDA requires DES manufacturers to submit summary information and interpretations of reportable and nonreportable events, first on a quarterly and then on an annual basis.

348

CH22

DRUG-ELUTING CORONARY STENTS

Postmarket experience: the sirolimus-eluting stent Based on the strength of the premarket clinical data, and the widely anticipated release of this break-through technology, the Cordis Cypher1 SES was rapidly adopted in the USA upon its marketing. In the first 5 months after the US launch, Cordis reported distributing 260 000 stents [44]. Even within the first 75 days, more than an estimated 50 000 patients received a Cypher1 stent [45]. Soon after this expansive launch, the FDA began receiving reports, through its MDR system, of potential Cypher1-related adverse events, the most concerning being reports of thrombosis and hypersensitivity. Approximately four dozen reports of subacute thrombosis (SAT) and one dozen reports of hypersensitivity were received within the first 2 months (by the end of June 2003). With regard to SATs, it was also noted that: (a) most of the reports of SAT originated from institutions reporting two or more cases; (b) the institutions reported SAT rates greater than 1% and/or not having had thrombosis cases for extended periods; (c) few SAT reports were received on the Cordis bare-metal stent (two reports over the prior 18 months); and (d) the proportion of all reports as SATs differed significantly between the Cordis bare-metal stent (approximately 1%) and the Cypher1 stent (approximately 30%). In addition, there was early evidence of off-label use (e.g. more complex lesions, in-stent re-stenosis, acute myocardial infarction) and use of improper technique. Although the dynamics of reporting of adverse events are well-recognized (e.g. reporting varying by publicity, notoriety of the device, time period of reporting or marketing, population exposure to the device) [46], the seriousness of the event, reporting clustering from reputable institutions, and evidence of inappropriate and off-label use, led to early precautionary action. Cordis, in conjunction with FDA, issued a ‘Dear Colleague’ letter in July 2003 that stressed using the product in accordance with indications for use and procedures noted in the package insert [45]. Although useful, the MDR system has its limitations in terms of deriving estimates of the true incidence of adverse events, including SATs, since the system typically reflects large-scale underreporting of events (the ‘numerator’ in incidence) and does not incorporate patient exposure information (the ‘denominator’) [46]. Thus, whether the incidence of SAT in Cypher1 stent recipients was greater than in comparable bare-metal stent recipients was impossible to determine. Furthermore, unearthing the potential etiology of a SAT (e.g. use in more complex patients, improper technique, improper use of adjunctive drugs, stent effects) was difficult given the incomplete and imprecise clinical information typically captured in a report. Reports on SATs, however, continued to be received and, based on FDA and Cordis’ investigations, revealed improper technique as well as significant off-label use (up to one-half of reports). In October 2003, after receiving more than 260 US reports of SATs and 50 reports of hypersensitivity, FDA issued its first public health notification on the Cypher1 stent, which reminded physicians to follow the instructions for use and to continue reporting adverse events [47]. Shortly after issuance of the notification, additional information from Cordis’ own ongoing clinical trials, in comparison to comparable bare-metal stent data, suggested that the incidence of SATs is within the expected

DRUG-ELUTING STENTS: REGULATORY PERSPECTIVE

349

rate for any stent. Furthermore, the hypersensitivity events were generally believed to be related to standard procedural drug therapy. An updated notification reflected these views [48]. Although not available at the time of these notifications, additional information has borne out these findings. Cordis, in accordance with a postapproval requirement imposed by FDA on their Cypher1 stent, conducted a 2000 patient postapproval study registry. A study using a nationwide registry found that about one-third of use was off-label [49]. In addition, and in testimony to its rapid adoption, by the last quarter of 2003 about 45% of all angioplasties involved placement of a Cypher1 stent.

Postmarket experience: the paclitaxel-eluting stent The Boston Scientific TAXUS Express1[2] paclitaxel-eluting coronary stent, released for marketing in March 2004, was the second DES commercialized in the USA. Similar to the Cypher1 stent, the market release of the TAXUS1 stent was widely embraced, resulting in more than 200 000 units used within the first 3 months and achieving a market share of about 70% [50]. During this period, approximately two dozen case reports of ‘nondeflate’, in which the stent delivery balloon deflated slowly or failed to deflate after stent deployment, were reported to FDA [51]. Boston Scientific, through engineering analysis, identified the causative manufacturing flaw and received FDA approval for a modified process in May 2004. During July–August, based on additional reports, the firm issued three recalls of suspected lots [52]. In addition, several hundred reports of balloon withdrawal ‘stickiness’ were received. Upon further investigation, this withdrawal resistance, poststent deployment, is considered to be potentially a combination of friction between the stent-delivery balloon and the drug–polymer coating on the stent, as well as certain coronary vessel characteristics (e.g. tortuosity) [52]. The association, if any, with reported adverse events is unclear.

Postmarket issues The varied adverse events associated with the two currently marketed DESs are revealing with respect to the system in place to capture such information. Whereas thrombosis is an ‘expected’ aspect of the underlying disease process and the stenting procedure (that can occur at the time of stenting or months later), nondeflation is a comparatively rare and unique occurrence occurring at the time of stent deployment. Making sense of thrombosis requires information on its rate of occurrence as well as detailed information related to its potential etiology (e.g. lesion-related, device-related, or procedure-related). Such information, as noted previously, is generally not available from passive reporting systems, such as FDA’s MDR system. In contrast, unique events such as ‘nondeflate’ may readily point to a device-related problem, which quality control systems and product testing may readily confirm. Passive surveillance, in the case of

350

CH22

DRUG-ELUTING CORONARY STENTS

‘nondeflate’ has ‘done its job’. In contrast, with events such as thrombosis, other sources of information are critically important to assess the relative degree of risk. The latter point has once again been recently highlighted in the case of late-term thrombosis (> 6 months poststenting) [53,54]. Although well-founded case reports reveal the problem, its potential public health impact can only be assessed through large-scale, systematic long-term follow-up studies. Adverse events related to DES use submitted to the FDA during the initial 6 months following approval showed that in routine clinical practice, more than half of adverse events reported to the FDA occurred when DESs were used for an indication for which the stent was not approved. While these findings raise awareness of the importance of reporting adverse events and product problems, they also inform the need for systematic premarket evaluation of other potential DES indications, as well as gathering of more systematic postmarket data on DES use in broad clinical settings. With regard to the latter, postmarket follow-up of trial cohorts and industry- and professional societysponsored registries should provide more insight to device-related outcomes and device utilization. The lack of information on these widely used medical devices underlines the importance of postmarket surveillance, including surveillance and epidemiology studies to better understand the patterns of use, as well as the long term safety issues and benefits of this new technology.

References 1. Kandzari DE, Tcheng JE, Zidar JP. Coronary artery stents: evaluating new designs for contemporary percutaneous intervention. Cathet Cardiovasc Intervent 2002; 56: 562–576. 2. Sousa JE, Costa MA, Abiziad A et al. Lack of neointimal proliferation after implantation of sirolimus-coated stents in human coronary arteries. Circulation 2001; 103: 192–195. 3. Sousa JE, Costa MA, Abizaid AC et al. Sustained suppression of neointimal proliferation by sirolimus-eluting stents: one-year angiographic and intravascular ultrasound follow-up. Circulation 2001; 104: 2007–2011. 4. Sousa JE, Costa MA, Abizaid A et al. Four-year angiographic and intravascular ultrasound follow-up of patients treated with sirolimus-eluting stents. Circulation 2005; 111: 2326–2329. 5. Morice MC, Serruys PW, Sousa EJ et al. A randomized comparison of a sirolimus-eluting stent with a standard stent for coronary revascularization. N Engl J Med 2002; 346: 1173–1780. 6. Fajadet J, Morice MC, Bode C et al. Maintenance of long-term clinical benefit with sirolimus-eluting coronary stents: three-year results of the RAVEL trial. Circulation 2005; 111: 1040–1044. 7. Moses JW, Leon MB, Popma JJ et al. for the SIRIUS Investigators. Sirolimus-eluting stents vs. standard stents in patients with stenosis in a native coronary artery. N Engl J Med 2003; 349: 1315–1323. 8. Holmes DR, Leon MB, Moses JW et al. Analysis of 1-year clinical outcomes in the SIRIUS trial: a randomized trial of a sirolimus-eluting stent vs. a standard stent in patients at high risk for coronary restenosis. Circulation 2004; 109: 634–640. 9. Schofer J, Schluter M, Gershlick AH et al. for the E-SIRIUS Investigators. Sirolimus-eluting stents for treatment of patients with long atherosclerotic lesions in small coronary arteries: double-blind, randomised controlled trial (E-SIRIUS). Lancet 2003; 362: 1093–1099.

REFERENCES

351

10. Schampaert E, Cohen EA, Schluter M et al. for the C-SIRIUS Investigators. The Canadian study of the sirolimus-eluting stent in the treatment of patients with long de novo lesions in small native coronary arteries (C-SIRIUS). J Am Coll Cardiol 2004; 43: 1110–1115. 11. Grube E, Lansky A, Hauptmann KE et al. High-dose 7-hexanoyltaxol-eluting stent with polymer sleeves for coronary revascularization: one-year results from the SCORE randomized trial. J Am Coll Cardiol 2004; 44: 1368–1372. 12. Farb A, Heller PF, Shroff S et al. Pathological analysis of local delivery of paclitaxel via a polymer-coated stent. Circulation 2001; 104: 472–479. 13. Gershlick AH, Descheerder I, Chevalier B et al. Inhibition of restenosis with a paclitaxel-eluting, polymer-free coronary stent: the European evaluation of paclitaxel eluting stent (ELUTES) trial. Circulation 2004; 109: 487–493. 14. Lansky AJ, Costa RA, Mintz GS et al. for the DELIVER clinical trial investigators. Nonpolymerbased paclitaxel-coated coronary stents for the treatment of patients with de novo coronary lesions: angiographic follow-up of the DELIVER clinical trial. Circulation 2004; 109: 1948–1954. 15. Grube E, Silber SM, Hauptmann KE. Taxus I: prospective, randomized, double-blind comparison of NIRx stents coated with paclitaxel in a polymer carrier in de novo coronary lesions compared with uncoated controls. Circulation 2001; 104: II-463 [abstr]. 16. Colombo A, Drzewiecki J, Banning A et al. Randomized study to assess the effectiveness of slow- and moderate-release polymer-based paclitaxel eluting stents for coronary artery lesions. Circulation 2003; 109: 788–794. 17. Stone GW, Ellis SG, Cox DA et al. A polymer-based, paclitaxel-eluting stent in patients with coronary artery disease. N Engl J Med 2004; 350: 221–231. 18. Grube E. TAXUS VI trial. Presented at European Society of Cardiology Scientific Sessions, August 2004, Munich, Germany. 19. Stone GW. TAXUS V trial. Presented at American College of Cardiology Scientific Sessions, March 2005, Orlando, FL. 20. Meredith I. ENDEAVOR I trial results. Presented at Transcatheter Therapeutics Scientific Sessions, September 2003, Washington, DC. 21. Wijns W. ENDEAVOR II trial results. Presented at American College of Cardiology Scientific Sessions, March 2005, Orlando, FL. 22. Grube E. FUTURE I and II trial results. Presented at Transcatheter Therapeutics Scientific Sessions, September 2003, Washington, DC. 23. Serruys PW. SPIRIT FIRST trial. Presented at Transcatheter Therapeutics Scientific Sessions, September 2004, Washington, DC. 24. Grube E. STEALTH First in Man trial results. Presented at EuroPCR Scientific Sessions, May 2004, Paris, France. 25. van der Giessen WJ. PISCES trial. Presented at EuroPCR Scientific Sessions, May 2004, Paris, France. 26. Morice MC. REALITY trial results. Presented at American College of Cardiology Scientific Sessions, March 2005, Orlando, FL. 27. Goy JJ, Stauffer JC, Siegenthaler M, Benoit A, Seydoux C. A prospective randomized comparison between paclitaxel and sirolimus stents in the real world of interventional cardiology: the TAXi trial. J Am Coll Cardiol; 45: 308–311. 28. Windecker S. SIRTAX trial results. Presented at American College of Cardiology Scientific Sessions, March 2005, Orlando, FL. 29. Kastrati A. ISAR-DIABETES trial results. Presented at American College of Cardiology Scientific Sessions, March 2005, Orlando, FL.

352

CH22

DRUG-ELUTING CORONARY STENTS

30. Kandzari DE, Roe MT, Ohman EM et al. Frequency, patterns and predictors of drug eluting stent utilization in patients with high-risk non-ST-segment elevation acute coronary syndromes: insights from the CRUSADE quality improvement initiative. Am J Cardiol (in press). 31. Ardissino D, Cavallini C, Bramucci E et al. Sirolimus-eluting vs. uncoated stents for prevention of restenosis in small coronary arteries: a randomized trial. J Am Med Assoc 2004; 292: 2727–2734. 32. Sousa JE. SVELTE trial results. Presented at EuroPCR 2004 Scientific Sessions, June 2004, Paris, France. 33. Park SJ. Multicenter prospective registry for DES implantation in long coronary lesions. Presented at Transcatheter Therapeutics 2004 Scientific Sessions, September 2004, Washington, DC. 34. Colombo A, Moses JW, Morice MC et al. Randomized study to evaluate sirolimus-eluting stents implanted in coronary bifurcation lesions. Circulation 2004; 109: 1244–1249. 35. Pan M, Suarez de Lezo J, Medina A et al. Rapamycin-eluting stents for the treatment of bifurcated coronary lesions: a randomized comparison of a simple vs. complex strategy. Am Heart J 2004; 148: 857–864. 36. Nakamura S, Muthusamy TS, Bae JH, Cahyadi YH et al. Impact of the sirolimus-eluting stent on the outcome of patients with chronic total occlusions. Am J Cardiol 2005; 95: 161–166. 37. Werner GS, Krack A, Schwarz G, Prachnau D et al. Prevention of lesion recurrence in chronic total coronary occlusions by paclitaxel-eluting stents. J Am Coll Cardiol 2004; 44: 2301–2306. 38. Hoye A, Tanabe K, Lemos PA et al. Significant reduction of restenosis after the use of sirolimuseluting stents in the treatment of chronic total occlusions. J Am Coll Cardiol 2004; 43: 1954–1958. 39. Kastrati A, Mehilli J, von Beckerath N et al. Sirolimus-eluting stent or paclitaxel-eluting stent vs. balloon angioplasty for prevention of recurrences in patients with coronary in-stent restenosis: a randomized controlled trial. J Am Med Assoc 2005; 293: 165–171. 40. Tanabe K, Serruys PW, Grube W et al. TAXUS III trial: in-stent restenosis treated with stentbased delivery of paclitaxel incorporated in a slow-release polymer formulation. Circulation 2003; 107: 559–564. 41. Ge L, Iakovou I, Sangiorgi GM et al. Treatment of saphenous vein graft lesions with drug-eluting stents: immediate and midterm outcome. J Am Coll Cardiol 2005; 45: 989–994. 42. McFadden EP, Stabile E, Regar E et al. Late thrombosis in drug-eluting coronary stents after discontinuation of antiplatelet therapy. Lancet 2004; 23: 1519–1521. 43. Waters RE, Kandzari DE, Phillips HR, Crawford LE, Sketch MH Jr. Late thrombosis following treatment of in-stent restenosis with drug-eluting stents after discontinuation of antiplatelet therapy. Cathet Cardiovasc Intervent (in press). 44. Mehran R, Leon MB, Feigal DA, Jefferys D et al. Post-market approval surveillance: a call for a more integrated and comprehensive approach. Circulation 2004; 109(25): 3073–3077. 45. Cordis Dear Colleague Letter, July 7 2003: www.fda.gov/bbs/topics/news/cordis [accessed July 27 2005]. 46. Gross TP, Kessler LG. Surveillance for medical devices – USA. Pharmacovigilance, Mann EB, Andrews RD (eds). New York, NY: Wiley, 2002; 411–422. 47. FDA public health web notification: information for physicians on sub-acute thromboses (SAT) and hypersensitivity reactions with use of the Cordis CYPHERTM coronary stent, October 29 2003: www.fda.gov/cdrh/safety/cypher.html [accessed July 27 2005]. 48. FDA public health web notification: updated information for physicians on sub-acute thromboses (SAT) and hypersensitivity reactions with use of the Cordis CYPHERTM sirolimus-eluting coronary stent, November 25 2003: www.fda.gov/cdrh/safety/cypher2.html [accessed July 27 2005].

REFERENCES

353

49. Rao SV, Shaw RE, Peterson ED, Klein LW et al. On- vs. off-label use of drug-eluting stents in clinical practice: data from the American College of Cardiology–National Cardiovascular Disease Registry (ACC-NCDR). American Heart Association Scientific Sessions, New Orleans, LA, November 7–10 2004. 50. Boston Scientific announces preliminary information on US TAXUS sales, Natick, MA, May 4 2004, Boston Scientific Corporation: www.bostonscientific.com [accessed February 9 2005]. 51. Boston Scientific recalls 200 TAXUSTM stent systems, Natick, MA, July 22 2004, Boston Scientific Corporation: www.bostonscientific.com [accessed February 9 2005]. 52. Muni NI, Gross TP. Problems with drug-eluting coronary stents – the FDA perspective. N Engl J Med 2004; 351(16): 1593–1595. 53. McFadden EP, Stabile E, Regar E et al. Late thrombosis in drug-eluting coronary stents after discontinuation of antiplatelet therapy. Lancet 2004; 364(9444): 1519–1521. 54. Eisenberg MJ. Drug-eluting stents: some bare facts. Lancet 2004; 364(9444): 1466–1467. 55. Kandzari DE. Clinical outcomes with zotarolimus-eluting stent and sirolimus-eluting stent: results from the ENDEAVOR III Trial. Presented at Transcatheter Therapeutics Scientific Sessions, September 2005, Washington, DC.

23 The treatment of abdominal aortic aneurysms Dale R. Tavris US Food and Drug Administration, Rockville, MD, USA

Louis M. Messina Division of Vascular Surgery, University of California, San Francisco, CA, USA

Natural history and indications for treatment Abdominal aortic aneurysm (AAA) is a localized bulge in the abdominal aorta. Left untreated, aneurysms dilate slowly, usually producing no symptoms until they burst. Death from aneurysm rupture is often rapid and at least 62% of patients die before reaching the hospital [1]. Overall mortality for ruptured AAA, including hospital deaths, may approach 90% [2]. The greatest potential for a reduction in aneurysm-related mortality therefore lies in the identification, and prophylactic repair, of aneurysms that have not yet ruptured. National Center for Health Statistics (NCHS) data indicate that there are approximately 15 000 deaths resulting from aneurysm rupture each year in the USA [3]. Risk factors include advanced age, male gender, smoking, and a family history of aneurysms. They may affect 1–2% of men older than 50 years [4]. A diagnosis of AAA presents three options: observation, surgery, or endovascular repair. This choice is influenced by the diameter and shape of the aneurysm and by the age, gender, and health of the patient. Although aneurysm diameter is a strong predictor of rupture risk, patient size is also a factor, so that for any given diameter, the risk of rupture is higher in a women than in a man. Gender also influences the decision to Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

356

CH23

THE TREATMENT OF ABDOMINAL AORTIC ANEURYSMS

intervene through its effects on the risk of treatment: women fare less well than men following aneurysm repair, by open or endovascular means [5]. Life expectancy is important in assessing the value of elective AAA repair, because the patient has to live long enough for the cumulative risk of aneurysm rupture to exceed the immediate risks of intervention [6]. Data on rupture risk in patients who are followed with conservative (i.e. nonsurgical) management should be interpreted cautiously, especially because the factors on which patients are selected for nonsurgical management may strongly affect the prognosis. For example, severe comorbidity will increase the likelihood of nonsurgical management (while increasing the risk of rupture), whereas an aneurysm that appears to be expanding rapidly (which also increases the risk of rupture) will be more likely to be treated surgically. That being said, a review of the literature of the natural history of AAAs revealed that 14 studies involving 393 patients with aneurysms < 5 cm in diameter were associated with a rupture rate of 1.0% at 1 year, whereas four studies involving 131 patients with aneurysms > 5 cm in diameter were associated with a 1 year rupture rate of 8.5% [7]. The same review also demonstrated that large aneurysms expand at a greater rate than small aneurysms (0.2–0.4 cm/year for aneurysms < 54 cm, 0.3–0.7 cm/year for aneurysms > 5 cm).

Surgery vs. endovascular treatment – general considerations The traditional surgical treatment of AAA involves direct exposure of the aneurysm through a long abdominal incision, interruption of aortic flow, and the creation of sutured connections (anastomoses) between the ends of a fabric tube (graft) and the nondilated blood vessels above and below the aneurysm. Endovascular repair substitutes transfemoral arterial access for trans-abdominal access, expanding metal cylinders (stents) for anastomotic suture lines, and fluoroscopic imaging for palpation and inspection. The endovascular prosthesis serves as an internal bypass, isolating the aneurysm from the circulation, and producing reductions in aneurysm pressure [8], size, and risk of rupture. Failure to exclude the aneurysm is evidenced by contrast-enhanced blood between the stent-graft and the wall of the aneurysm, i.e. endoleak [9,10]. Some stent-grafts also produce a phenomenon known as endotension, in which the aneurysm dilates in the absence of endoleak [11]. These endograft failures (endoleak and endotension), by leading to aneurysm dilation, substantially increase the risk of aneurysm rupture. Compared to open surgery, endovascular repair causes less physiological derangement [12,13], perioperative mortality, perioperative morbidity, pain, and debility. These benefits are clearest in a patient whose large aneurysm precludes observation, and whose poor health precludes open surgery. The role of endovascular repair is less clear in a patient with a small aneurysm and good health, for whom the benefits depend on long life expectancy, durable protection from risk of rupture, and a low rate of late complications. As endovascular technique and technology have evolved, these factors have changed, and so have attitudes towards the respective roles of surgical and endovascular aneurysm repair [14,15].

NATURAL HISTORY AND INDICATIONS FOR TREATMENT

357

Outcomes associated with open surgical AAA repair Most studies of operative mortality rates associated with elective open surgical repair of AAA demonstrate rates of 5–6%. A review of 64 studies on this subject finds an average mortality rate of 5.5% [7]. The largest study available is an analysis of data from the National Hospital Discharge Survey of 358 521 cases during 1979–1997 [16], which demonstrated a mortality rate of 5.6%. Several additional large studies, with sample sizes of 275 or more subjects, demonstrated mortality rates of 4.5–6.6% [17–23]. In addition, several smaller studies have demonstrated a combined operative mortality rate of 5.2% [24–29]. Many studies have noted that certain categories of patients have much lower operative mortality rates associated with open repair than those noted above. Examples of this include: age < 80, 3.6% [16]; age < 66, 1.7% [17]; no renal failure, congestive heart failure or chronic obstructive respiratory disease, < 2% [30]. Numerous sources note operative mortality rates of 1–2% in high-volume institutions [30–35], with rates of about 1% found for these institutions in the past decade [30]. In addition to operative mortality, late mortality associated with open surgical repair should be considered. A review of late mortality associated with open surgical repair of AAA estimated a mortality rate of 0.1% per year for pseudo-aneurysms developing at the aortic suture line, and less than 0.1% per year for infection-related mortality [36]. This is consistent with several individual studies that looked at late mortality associated with open surgical repair of AAA, which demonstrated late mortality rates of 0–0.3% per year, with an average rate of about 0.18% per year [35,37–43]. As with operative mortality, discussed above, it seems likely that these risks would be reduced when performed in high-volume institutions, in non-high-risk patients, or in more recent times.

Outcomes associated with endovascular stent-grafts One of the most comprehensive sources for the assessment of long-term outcomes associated with endovascular repair is the Eurostar (European Collaborators on Stentgraft Techniques for aortic aneurysm repair) registry [44]. This registry was established in 1996 to assess treatment outcomes associated with endovascular grafts for AAA. Eighty-eight European centers of vascular surgery participated in this study. As of March 2000, 2464 patients had been registered and followed for an average of 12.2 months. The perioperative mortality rate was 2.0%. The cumulative risk of aneurysm rupture in this group was 1% per year, 62.5% of whom died. The risk of late conversions to open surgical repair was 2.1% per year, 24% of whom died at surgery. This study involved several different devices, some of which have been removed from the market, and the clinical outcomes with these devices have been generally less favorable than that seen with devices approved for use in the USA. The US Food and Drug Administration has approved five endovascular stent-grafts for the treatment of AAA [45]: Guidant’s Ancure, Medtronic’s AneuRx, Gore’s Excluder,

358

CH23

THE TREATMENT OF ABDOMINAL AORTIC ANEURYSMS

Cook’s Zenith, and the Endologix Powerlink. Perioperative (i.e. 30-day) mortality and morbidity associated with the implantation of these grafts is typically less than that seen with open surgery. Specifically, of 1545 premarket clinical trial subjects implanted with the above-noted five endografts and described in the five respective ‘Summaries of Safety and Effectiveness Data’, there were 20 perioperative (within 30 days) deaths [46– 50]. Perioperative death rates were in the range 0–1.9% in these studies, with a mean for the whole group of 1.3%. There were no statistically significant differences between the five devices in these studies. However, they have not been used for a long enough time to provide a great amount of confidence with respect to long-term outcomes. Of the five approved products, Guidant’s Ancure is no longer marketed.

FDA experience with the postmarket assessment of an endovascular graft A thorough understanding of long-term outcomes associated with the use of endovascular stent-grafts is crucial for physicians to have in order to make valid decisions on when to use these devices. That is the reason that FDA undertook a study of long-term outcomes associated with the AneuRx stent-graft, for which the greatest amount of intermediate-term data was available.

Methods used for postmarket assessment One of the first major decisions made by the FDAwas to narrow the focus of the analysis to aneurysm-related mortality. The reason for this is that aneurysm-related mortality is the primary outcome which treatment is supposed to prevent. Consequently, it is an excellent indicator of both effectiveness and safety. For this purpose, a death was presumed to be AAA-related if it met one of the three following criteria: 1. The underlying cause of death was noted to be ‘ruptured abdominal aortic aneurysm’, ‘abdominal aortic aneurysm’ or ‘aortic aneurysm’. 2. The patient died within 1 month of implantation or attempted implantation of the AneuRx stent-graft, or of any attempt to repair the aneurysm, or as a result of complications suffered from these procedures. 3. The patient died within 1 month of surgery to remove the AneuRx stent-graft and convert to open aneurysm repair, or other intervention related to AAA, or as a result of complications suffered from these procedures. ‘Presumed’ cases of AAA-related deaths were then followed up with additional investigation in order to determine whether or not evidence to the contrary could be found. In the

FDA EXPERIENCE WITH THE POSTMARKET ASSESSMENT OF AN ENDOVASCULAR GRAFT

359

absence of evidence to the contrary, cases meeting the above definition were considered to be AAA-related. The plan was to compare AAA-related mortality in subjects implanted with AneuRx with AAA-related mortality in similar patients who underwent the alternative treatment for AAA, open surgical repair of the aneurysm. That comparison would serve as the main evidence of the postmarket safety of the AneuRx stent-graft. There were eventually 942 AneuRx subjects identified who met the study criteria, and 66 controls (who underwent open surgical repair of their AAA). There were no aneurysm-related deaths among the controls. But 66 controls was too small a sample size to perform a meaningful comparison with the AneuRx subjects. Consequently, it was decided that the FDAwould have to use the medical literature in order to compare the safety of AneuRx with that of open surgical repair of AAA. In order to calculate the AAA-related death rate, the number of AAA-related deaths was to be divided by the length of follow-up of cohort subjects. The length of follow-up for each individual patient was calculated as the length of time between implant (or attempted implant) and the date on which follow-up ceased. Follow-up was considered to have ceased on the date of death for patients who died prior to a pre-specified freeze date. For patients whose implant was converted to open surgical aneurysm repair, but who did not die, follow-up was considered to have ceased on the day of the conversion.

Results The analysis showed that the perioperative (within 30 days) aneurysm-related death rate associated with the AneuRx Stent-graft was 1.5% (14/942). Following implantation, an additional eight AAA-related deaths were identified during the subsequent 3 years of follow-up, covering 2080 patient-years, for an annualized late mortality rate of 0.40% per year. This FDA analysis estimates an AAA-related death rate of 1.9% at 1 year postimplant, 2.2% at 2 years postimplant, and 2.7% at 3 years postimplant [52].

Discussion – general considerations in the comparison of AneuRx outcomes with literature controls The studies noted in the above section on open surgical AAA repair involve general populations of AAA patients, and are therefore not directly comparable to the study population used in Medtronic’s study, for two main reasons: 1. The sites participating in the study were institutions that see a high volume of patients with AAA.

360

CH23

THE TREATMENT OF ABDOMINAL AORTIC ANEURYSMS

2. Also, the Medtronic study excluded the highest-risk patients. In particular, the AneuRx study excluded the following high-risk categories [51]: American Society of Anesthesiology grade above IV (for some phases of the study, and grade IV and above for other phases); morbid obesity; acute renal failure or chronic dialysis; active systemic infection; < 1 year of life expectancy; leaking aneurysm; aortic dissection; aorto-iliac occlusive disease; and extension of the aneurysm into the iliac arteries.

Recommendations Based on the findings of the above noted study, FDA recommended in a Public Health Notification that the AneuRx stent-graft be used only in patients who meet the appropriate risk:benefit profile and who can be treated in accordance with the instructions for use [52]. In determining the risk:benefit profile for patients with AAA disease and the appropriate treatment option, FDA recommended that the following factors be considered: Long term AAA-related mortality, especially due to AAA rupture. The information above suggests that the risk of late AAA-related mortality associated with AneuRx may exceed that associated with open surgery in some institutions. Because of this, the overall AAA-associated mortality from the AneuRx Stentgraft is likely to cross-over and exceed the AAA-associated mortality from open surgery at some point in time. The experience of the institution or the physician. If open or endovascular surgery is performed in institutions or by physicians with little experience with open or endovascular AAA repair, the mortality rate may be considerably higher than average. Surgical risk factors for the individual patient. In patients who have substantial surgical risk factors, such as age [18] and comorbidities (e.g. cardiac, renal and pulmonary), the mortality rate for open resection of AAA may be considerably higher than average. For example, the predicted operative mortality for a 70 year-old could range from 2% if no risk factors were present to over 40% if multiple comorbidities were present [19]. The patient’s life expectancy. Treatment with an endovascular graft may be preferable for patients with a shortened life expectancy. The patient’s willingness to comply with the follow-up schedule for the endovascular graft.

ADVANCES IN STENT-GRAFT DESIGN AND PERFORMANCE

361

Advances in stent-graft design and performance In order to address many of the serious problems described above, manufacturers of these endovascular grafts have developed several improvements to them. This section describes some of the main efforts on this front. Early clinical experience soon showed that effective aneurysm protection depended on secure hemostatic attachment of the stent-graft to nondilated arteries proximal and distal to the aneurysm. The frequent lack of a nondilated segment of distal aorta [53] necessitated iliac implantation. Tapered stent-grafts achieved this goal by channeling all aortic flow into one iliac artery, but the uni-iliac approach necessitated femoro-femoral bypass and contralateral common iliac artery occlusion, both of which have disadvantages [54], hence the development of bifurcated (Y-shaped) stent-grafts, with a single trunk for infrarenal aortic implantation and two limbs for bilateral iliac implantation. The first bifurcated stent-grafts were inserted whole (unibody) and positioned using a system of catheters and wires [55]. These were soon followed by modular bifurcated stent-grafts, which were assembled in situ from two or more components. Each form of bifurcation has its problems. Unibody stent-grafts tend to be complex and limiting, while modular stent-grafts are prone to component separation. One unibody stent-graft was bulky, difficult to insert, and usable only in patients with small diameter implantation sites. Another, although smaller and more versatile, proved unstable in the long term; the components kinked, broke, occluded, separated, or migrated into the aneurysm [56]. Accumulating complications, re-interventions, aneurysm ruptures and deaths soon outweighed the short-term complications of open surgery for all but the sickest, shortest-lived patients. However, these poor outcomes were not intrinsic failures of the endovascular approach, but failures of the technology of the day. Device design has advanced over the past 10 years as successful features have been preserved and unsuccessful features eliminated. The result has been a steady improvement in long-term performance, as demonstrated by stratified analysis of the Eurostar registry comparing current and obsolete devices. In general, short-term problems have resulted from failures of stentgraft insertion, while long-term problems have resulted from failures of stent-graft attachment and structure.

Stent-graft delivery Early devices were too large, too rigid and too blunt to traverse iliac arteries that were either small or tortuous [57]. Failed insertion and iliac artery rupture were common. Small caliber (< 20 French sheath size), trackability, flexibility, and torsional rigidity enable modern devices to perform with predictable success, even in the presence of severe iliac arterial disease. It is even becoming possible, with the advent of smaller delivery systems and arterial closure devices, to use an entirely percutaneous approach [58].

362

CH23

THE TREATMENT OF ABDOMINAL AORTIC ANEURYSMS

Stent-graft attachment Stent-graft attachment fails, leading to migration and endoleak, when hemodynamic forces overwhelm the stabilizing effects of friction, barb penetration, tissue ingrowth, and column strength. Mathematical modeling suggests that the stent-graft is subject to intermittent pulsatile forces of 7–12 N [59]. At the time of implantation, current self-expanding stent-grafts generate stabilizing frictional forces less than 5 N, depending on the length of the implantation site, the compliance of the stent, and the degree of oversizing. As subsequent neck expansion decreases the degree of oversizing, the frictional forces probably decline, reaching zero when the neck diameter and stent-graft diameter are equal. Barb-mediated attachment is more durable. Multiply barbed stent-grafts can resist forces of up to 20 N [54], and attachment remains secure as long as the barbs remain intact. Only the uncovered portions of balloon-expanded stents induce sufficient ingrowth for secure long-term attachment. Explanted specimens show that uncovered, or externally mounted, self-expanding stents may also generate a hyperplastic response in the surrounding aorta, but this response fails to provide predictable protection from risk of migration with any of the current stent-graft designs. Bifurcated stent-grafts with long, flexible limbs probably derive little support from below (column strength). Short stiff limbs may be more stable, because the source of support, the aortic bifurcation, is closer to the site of the displacement force, the stentgraft bifurcation. The differing migration rates of different devices reflect the relative contributions of the above design features. In general, those that rely on friction and column strength migrate often, while those that rely on suprarenal stents and barbs migrate rarely.

Stent-graft structure Stent fracture has been reported after long-term implantation of all current stent-grafts, but clinically significant consequences have been rare, because cyclical strain is actually very low (< 0.1% of resting diameter) within a month of implantation, and stent function is redundant in the center of the graft, where most fractures occur. Graft erosion has been more of a problem. Micro-movement between the apex of a stent and overlying graft disrupts the fabric, leading to type III endoleak, aneurysm dilatation, and rupture. Most manufacturers have solved this problem by attaching the stents securely to the fabric. However, the sutures themselves can also generate holes by inducing fiber separation in loosely woven fabrics. These small holes rarely cause frank endoleak or rupture, but may be a factor in cases of aneurysm pressurization and dilatation [11,60]. The main disadvantage of the modular approach is a potential for component separation. Long, flexible, redundant limbs and a short intercomponent overlap zone are risk factors. The problem has been largely eliminated by increasing the length of the overlap zone. Some devices derive additional stability from a long body–short leg combination.

REFERENCES

363

Patency Unlike conventional surgical grafts, endovascular stent-grafts are affected by the angulation and narrowing of the native arteries. The unsupported limbs of some early endovascular grafts developed folds and kinks, leading to thrombosis. The fully stented limbs of most modular designs derive support from a continuous chain of stents. Consequently, the thrombosis rates of fully-stented modular stent-grafts are generally < 5%.

Conclusion Large AAAs usually require treatment, either surgical or endovascular, in order to reduce the risk of rupture and consequent death. Although endovascular treatment of these aneurysms is generally associated with reduced morbidity and mortality in the short term, long-term comparisons of these two treatment approaches are much more difficult to conduct and interpret, and many factors need to be weighed when considering the best approach to individual patients. The FDA must remainvigilant in its surveillance of longterm outcomes associated with FDA-approved endovascular grafts. The extent to which improvements in the design of endovascular grafts will alter the relative risk:benefit balance in the comparison of endovascular vs. surgical approaches to AAA is an ongoing question.

References 1. Ingoldby CJ, Wujanto R, Mitchell JE. Impact of vascular surgery on community mortality from ruptured aortic aneurysms. Br J Surg 1986; 73: 551–553. 2. Johansson G, Swedenborg J. Ruptured abdominal aortic aneurysms: a study of incidence and mortality. Br J Surg 1986; 73: 101–103. 3. National Center for Health Statistics. Vital Statistics of the United States, 1988, vol 2, Mortality, part A. DHS Publication No. PHS 91–11010. Washington, DC: Government Printing Office, 1991. 4. Dzau VJ, Creager MA. Diseases of the aorta. In Harrison’s Principles of Internal Medicine, 15th edn, Braunwald E (ed.). New York: McGraw Hill, 2001. 5. Wolf YG, Arko FR, Hill BB, Olcott C et al. Gender differences in endovascular abdominal aortic aneurysm repair with the AneuRx stent-graft. J Vasc Surg 2002; 35: 882–886. 6. Chuter TA, Reilly LM, Faruqi RM, Kerlan RB et al. Endovascular aneurysm repair in high-risk patients. J Vasc Surg 2000; 31: 122–133. 7. Hallin A, Bergqvist D, Holmberg L. Literature review of surgical management of abdominal aortic aneurysm. Eur J Vasc Endovasc Surg 2001; 22: 197–204. 8. Dias NV, Ivancev K, Malina M et al. Intra-aneurysm sac pressure measurements after endovascular aneurysm repair: differences between shrinking, unchanged, and expanding aneurysms with and without endoleaks. J Vasc Surg 2004; 39: 1229–1235. 9. White GH, May J, Waugh RC, Chaufour X, Yu W. Type III and type IV endoleak: toward a complete definition of blood flow in the sac after endoluminal AAA repair. J Endovasc Surg 1998; 5: 305–309.

364

CH23

THE TREATMENT OF ABDOMINAL AORTIC ANEURYSMS

10. van Marrewijk CJ, Fransen G, Laheij RJ, Harris PL, Buth J; EUROSTAR Collaborators. Is a type II endoleak after EVAR a harbinger of risk? Causes and outcome of open conversion and aneurysm rupture during follow-up. Eur J Vasc Endovasc Surg 2004; 27(2): 128–137. 11. Cho JS, Dillavou ED, Rhee RY, Makaroun MS. Late abdominal aortic aneurysm enlargement after endovascular repair with the excluder device. J Vasc Surg 2004; 39: 1236–1240. 12. Baxendale BR, Baker DM, Hutchinson A, Chuter TA et al. Haemodynamic and metabolic response to endovascular repair of infra-renal aortic aneurysms. Br J Anaesth 1996; 77(5): 581–585. 13. Lee WA, Carter JW, Upchurch G, Seeger JM, Huber TS. Perioperative outcomes after open and endovascular repair of intact abdominal aortic aneurysms in the United States during 2001. J Vasc Surg 2004; 39: 491–498. 14. Sternbergh WC, Nordness PJ, York JW, Conners MS et al. Endo-exuberance to endo-reality: trends in the management of 431 AAA repairs between 1996 and 2002. J Endovasc Ther 2003; 10: 418–423. 15. Collin J, Murie JA. Endovascular treatment of abdominal aortic aneurysm: a failed experiment. Br J Surg 2001; 88(10): 1281–1282. 16. Heller JA, Weinberg A, Arons R, Krishnasastry KV et al. Two decades of abdominal aortic aneurysm repair: have we made any progress? J Vasc Surg 2000; 32: 1091–1100. 17. Irvine CD, Shaw E, Poskitt KR, Whyman MR et al. A comparison of the mortality rate after elective repair of aortic aneurysms detected either by screening or incidentally. Eur J Vasc Endovasc Surg 2000; 20: 374–378. 18. Humphreys WV, Byrne J, James W. Elective abdominal aortic aneurysm operations – the results of a single surgeon series of 243 consecutive operations from a district general hospital. Ann R Coll Surg Engl 2000; 82: 64–68. 19. Aune S. Risk factors and operative results of patients aged less than 66 years operated on for asymptomatic abdominal aortic aneurysm. Eur J Vasc Endovasc Surg 2001; 22: 240–243. 20. May J, White GH, Waugh R, Ly CN et al. Improved survival after endoluminal repair with second-generation prostheses compared with open repair in the treatment of abdominal aortic aneurysms: a 5-year concurrent comparison using life table method. J Vasc Surg 2001; 33: S21–26. 21. Olsen PS, Schreoder T, Agerskov K et al. Surgery for abdominal aortic aneurysms. J Cardiovasc Surg 1991; 32: 636–642. 22. Galland RB. Mortality following elective infrarenal aortic reconstruction: a joint vascular research group study. Br J Surg 1998; 85: 633–636. 23. Johnston KW. Muticenter prospective study of nonruptured abdominal aortic aneurysm, II: variables predicting morbidity and mortality. J Vasc Surg 1989; 9: 27–447. 24. Kazmers A, Striplin D, Jacobs LA, Welsh DE, Perkins A. Outcomes after abdominal aortic aneurysm repair: comparison of mortality defined by centralized VA patient treatment file data vs. hospital-based chart review. J Surg Res 2000; 88: 42–46. 25. Ballotta E, Da Giau G, Bottio T, Toniato A. Elective surgery for small abdominal aortic aneurysms. Cardiovasc Surg 1999; 7(5): 495–502. 26. Sandison AJP, Panayiotopoulos Y, Edmondson RC, Tyrrell MR, Taylor PR. A 4-year prospective audit of the cause of death after infrarenal aortic aneurysm surgery. Br J Surg 1996; 83: 1386–1389. 27. Cohnert TU, Oelert F, Wahlers T, Gohrbandt B et al. Matched-pair analysis of conventional vs. endoluminal AAA treatment outcomes during the initial phase of an aortic endografting program. J Endovasc Ther 2000; 7(2): 94–100. 28. Bosch JL, Beinfeld MT, Halpern EF, Lester JS, Gazelle GS. Endovascular vs. open surgical elective repair of infrarenal abdominal aortic aneurysm: predictors of patient discharge destination. Radiology 2001; 220: 576–580.

REFERENCES

365

29. Bertrand M, Godet G, Koskas F, Cluzel P et al. Endovascular treatment of abdominal aortic aneurysms: is there a benefit regarding postoperative outcome? Eur J Anesthesiol 2001; 18: 245–250. 30. Cao P, De Rango P. Abdominal aortic aneurysms: current management. Cardiologia 1999; 44(8): 711–717. 31. Golden MA, Whittemore AD, Donaldson MC, Mannick JA. Selective evaluation and management of coronary artery disease in patients undergoing repair of abdominal aortic aneurysms. Ann Surg 1990; 212(4): 415–423. 32. Perry MO, Calcagno D. Abdominal aortic aneurysm surgery: the basic evaluation of cardiac risk. Ann Surg 1988; 208(6): 738–742. 33. Cronenwett JL, Johnston KW. The United Kingdom Small Aneurysm Trial: implications for surgical treatment of abdominal aortic aneurysms. J Vasc Surg 1999; 29(1): 191–194. 34. Braunwald E, Fauci AS, Kasper DL, Hauser SL et al (eds). Harrison’s Principles of Internal Medicine, 15th edn. New York, NY: McGraw-Hill, 2001; 1431. 35. Hertzer NR, Mascha EJ, Karafa MT, H’Hara PJ et al. Open infrarenal abdominal aortic aneurysm repair: the Cleveland Clinic Experience from 1989 to 1998. J Vasc Surg 2002; 35: 1145–1154. 36. Limet R, Creemers E. Comparison between open and closed repair for abdominal aortic aneurysms: a word of caution. Acta Chir Belg 2000; 100: 12–15. 37. Plate G, Hollier LA, O’Brien P, Pairolero PC et al. Recurrent aneurysms and late vascular complications following repair of abdominal aortic aneurysms. Arch Surg 1985; 120: 590–594. 38. Williamson WK, Nicoloff AD, Taylor LM, Moneta GL et al. Functional outcome after open repair of abdominal aortic aneurysm. J Vasc Surg 2001; 33: 913–920. 39. Biancari F, Ylonen K, Anttila V, Juvonen J et al. Durability of open repair of infrarenal abdominal aortic aneurysm: a 15-year follow-up study. J Vasc Surg 2002; 35: 87–93. 40. Hallet JW, Marshall DM, Petterson TM, Gray DT et al. Graft-related complications after abdominal aortic aneurysm repair: reassurance from a 36-year population-based experience. J Vasc Surg 1997; 25: 277–286. 41. Johnston KW. Nonruptured abdominal aortic aneurysm: six-year follow-up results from the Multicenter Prospective Canadian Aneurysm Study. J Vasc Surg 1994; 20: 163–170. 42. Lederle FA, Johnson GR, Wilson SE, Ballard DJ et al. Immediate repair compared with surveillance of small abdominal aneurysms. N Eng J Med 2002; 346(19): 1437–1444. 43. Arko FR, Lee WA, Hill BB, Olcott C et al. Aneurysm-related death: primary endpoint analysis for comparison of open and endovascular repair. J Vasc Surg 2002; 36(2): 297–304. 44. Towne JB. Endovascular treatment of abdominal aortic aneurysms. Am J Surg 2004; 189: 140–149. 45. Faries PL, Dayal R, Rhee J, Trocciola S, Kent CK. Stent-graft treatment for abdominal aortic aneurysm repair: recent development in therapy. Curr Opin Cardiol 2004; 19: 551–557. 46. US Food and Drug Administration. Summary of safety and effectiveness data: the AneuRx stent-graft system. US Food and Drug Administration, Office of Device Evaluation, Rockville, MD, September 1999: http://www.fda.gov/cdrh/pdf/P990020b.pdf 47. US Food and Drug Administration. Summary of safety and effectiveness data: tube and bifurcated AncureTM. US Food and Drug Administration, Office of Device Evaluation, Rockville, MD, September 1999: http://www.fda.gov/cdrh/pdf/P990017b.pdf 48. US Food and Drug Administration. Summary of safety and effectiveness data: Zenith AAA endovascular graft. US Food and Drug Administration, Office of Device Evaluation, Rockville, MD, May 2003: http://www.fda.gov/cdrh/pdf2/P020018b.pdf 49. US Food and Drug Administration. Summary of safety and effectiveness data: ExcluderTM bifurcated endoprosthesis. US Food and Drug Administration, Office of Device Evaluation, Rockville, MD, November 2002: http://www.fda.gov/cdrh/pdf2/P020004b.pdf

366

CH23

THE TREATMENT OF ABDOMINAL AORTIC ANEURYSMS

50. US Food and Drug Administration. Summar2by of safety and effectiveness data: Powerlink endovascular graft and delivery system. US Food and Drug Administration, Office of Device Evaluation, Rockville, MD, October 2004: http://www.fda.gov/cdrh/pdf4/p040002b.pdf 51. Medtronic Corporation. Investigational plan: Medtronic AneuRx endovascular prosthesis treatment of abdominal aortic aneurysms. Medtronic Corp. 52. FDA public health notification; updated data on mortality associated with Medtronic AVE AneuRx stent-graft system, 2003. 53. Chuter TAM, Green RM, Ouriel K, DeWeese JA. Infrarenal aortic aneurysm morphology: Implications for transfemoral repair. J Vasc Surg 1993; 17: 1120–1121. 54. Yusuf SW, Baker DM, Hind RE, Chuter TA et al. Endoluminal transfemoral abdominal aortic aneurysm repair with aorto-uni-iliac graft and femorofemoral bypass. Br J Surg 1995; 82: 916. 55. Scott RAP, Chuter TAM. Clinical endovascular placement of bifurcated graft in abdominal aortic aneurysm without laparotomy. Lancet 1994; 343: 413. 56. Beebe HG, Cronenwett JL, Katzen BT et al. Results of an aortic endograft trial: impact of device failure beyond 12 months. J Vasc Surg 2001; 33: S55–63. 57. Moore WS. The EVT tube and bifurcated endograft systems: technical considerations and clinical summary. J Endovasc Surg 1997; 4: 182–194. 58. Morasch MD, Kibbe MR, Evans ME et al. Percutaneous repair of abdominal aortic aneurysm. J Vasc Surg 2004; 40: 12–16. 59. Liffman K, Lawrence-Brown MM, Semmens JB, Bui A et al. Analytical modeling and numerical simulation of forces in an endoluminal graft. J Endovasc Ther 2001; 8: 358–371. 60. Resch T, Malina M, Lindblad B, Malina J et al. The Impact of stent design on proximal stentgraft fixation in the abdominal aorta: an experimental study. Eur J Vasc Endovasc Surg 2000; 20: 190–195.

24 Cardiovascular devices: aortic valves Ronald G. Kaczmarek and Chih-Hsin K. Liu US Food and Drug Administration, Rockville, MD, USA

Introduction [1–3] A prosthetic heart valve is a replacement for a malfunctioning or diseased heart valve. A malfunctioning/diseased heart valve does not open and close properly, due to stenosis or insufficiency. The efficient blood flow of the heart decreases and theworkload of the heart increases. If the pathology of the heart valve is not corrected over time, it can lead to heart failure. There are various kinds of prosthetic heart valves, which are categorized into two types, mechanical heart valves and tissue heart valves.

Mechanical heart valves The mechanical heart valve was first implanted in the early 1950s. Over the years, there have been various mechanical designs. The mechanical heart valves have progressed from caged-ball valves in the 1960s, disk valves in the 1970s, to bileaflet valves in the 1980s: Caged-ball valve. The caged-ball valve has a small ball held by a metal cage in order to limit the blood flow to a single direction. The Starr–Edwards caged-ball heart valve was the first commercially available heart valve in the 1960s (see Figure 24.1). To date, there are patients still implanted with this valve.

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

368

CH24

CARDIOVASCULAR DEVICES: AORTIC VALVES

Figure 24.1 Two Starr–Edwards artificial cardiac valves, 1978–1979. Two heart valves of ball and cage design, made by Edwards Laboratories in the USA, of plastic and metal with sterile cases. Reproduced with permission from the Science Museum/Science and Society Library

Disk valve. The disk valve was introduced in the late 1960s and the early 1970s. The tilting disc valve has a circular disk held in place by two struts. The disc opens when blood begins to travel forward and closes when the blood begins to travel backward. The Bjork–Shiley heart valve and the Medtronic Hall valve are two well-known disc heart valves. A refined model of the Shiley heart valve, the convexo–concave model, was withdrawn from the market in the 1980s due to a structural failure caused by strut fractures. Bileaflet valve. In the late 1970s, the bileaflet valve was introduced by St. Jude Medical Inc. and it became the model of the current modern design of mechanical heart valves. The bileaflet valve consists of two semi-circular leaflets that are connected to orifice housing by a butterfly hinge mechanism. The leaflets swing open parallel to the direction of the blood flow and provide central flow similar to that achieved with a natural heart valve. Mechanical heart valves have the advantage of greatest durability. They are suitable for young patients or for patients who do not want additional valve surgery in the future. However, the main risk with the mechanical heart valve is blood clot formation, which

INTRODUCTION [1–3]

369

can lead to strokes. To prevent blood clots, patients need life-long anticoagulant medication. Therefore, mechanical heart valves are contraindicated for patients who can not tolerate long-term anticoagulation therapy [4].

Tissue heart valves (bio-prosthetic heart valves) Compared to the mechanical heart valve, the tissue heart valves are closer to the design of the natural heart valve. Patients implanted with tissue heart valves have better hemodynamics and less damage to blood cells [5]. They do not require long-term anticoagulation because they are not at high risk for blood clot formation. However, valve tissue calcification and degeneration are common problems with tissue heart valves. Patients usually require another heart valve replacement in 10–15 years. There are two categories of tissue heart valves, including animal tissue and human tissue heart valves: 1. Animal tissue heart valve. The two most common animal tissue heart valves include the following: Porcine heart valve: a porcine heart valve tissue is harvested from a pig. The valve is treated with glutaraldehyde to cross-link collagen fibers and reduce antigenicity. Some of the porcine valves are stented valves, which are sewn to rigid or flexible stents and cloth sewing cuffs. Other valves are stentless valves which have excellent hemodynamics. However, they are difficult to implant. The durability of a porcine valve is about 10 years. Bovine heart valve: a bovine heart valve is made out of a cow’s pericardium. Similar to the porcine heart valve design, the bovine pericardium is sewn into a valvular frame. Pericardial valves open more completely and have excellent hemodynamics. The durability of bovine valves is similar to or better than porcine valves. 2. Human tissue heart valve. There are two types of human tissue heart valve: Autograft heart valve: an autograft heart valve is a valve transplanted from one location to another within the same patient. For instance, a cardiac surgical procedure called the Ross procedure involves removing a patient’s diseased aortic heart valve and implanting the pulmonary valve from the same patient to the aortic position. A homograft or animal tissue heart valve is then used to replace the pulmonary valve. Homograft heart valve (also known as allograft heart valve): a homograft heart valve is a valve taken from a deceased donor and transplanted into a patient. The donor’s valve is harvested and treated with antimicrobial solution and cryo-preserved at a very low temperature until it is used. Homograft valves have good hemodynamics

370

CH24

CARDIOVASCULAR DEVICES: AORTIC VALVES

and low thrombogenicity. Patients do not need life-long anticoagulation. However, the homograft valve is difficult to implant and the source of supply is limited.

Operative mortality The operative mortality risk, death related to the surgical procedure, from aortic valve replacement surgery is appreciable. Astor et al. [6] conducted a nationally representative study that employed data from the Nationwide Inpatient Sample, a massive nationally representative database that received data from over 900 hospitals, and found that the overall risk of in-hospital mortality following aortic valve replacement was 6.4%. The in-hospital mortality rate encompasses all deaths during the hospitalization admission for aortic valve replacement surgery and can capture operative mortality that may occur long after surgery. In-hospital mortality varies by various patient factors and concurrent surgeries. The in-hospital mortality rate was lower (4.3%) for first-time aortic valve replacements without concomitant other cardiac surgery, such as coronary artery bypass graft surgery or replacement of other heart valves. The in-hospital mortality rate was higher for replacement of an existing prosthetic aortic heart valve without concomitant other cardiac surgery (8.2%.) Multiple valve replacement also led to higher observed in-hospital mortality rates (8.8%, 95% CI ¼ 6.6–11.5). When existing prosthetic heart valves were replaced during the multiple valve replacement surgery, the in-hospital mortality rate was higher (15.2%, 95% CI ¼ 8.1–22.4). In-hospital mortality is increased when valve replacement is associated with cardiac bypass surgery, with or without multiple valve replacement. The estimates of operative mortality following aortic valve implantation obtained in the Astor et al. study were in close agreement with estimates obtained from the Society of Thoracic Surgeons (STS) National Cardiac Surgery Database. The STS system is a voluntary system. The 30 day mortality rate, deaths within 30 days of the aortic valve replacement surgery, without concomitant cardiac surgery, was 4.3% [7], compared to 4.5% for all aortic valve replacements in the Astor et al. study. The increase in operative mortality when other cardiac procedures were performed simultaneously with aortic valve replacement that was observed in the Astor et al. study was also noted in the STS study. The studies of Astoret al. and STS are among the largest American studies of operative mortality following aortic valve replacement. Roques et al. [8] collected and utilized data on cardiac surgery in Europe from 128 different centers in eight nations: Germany, France, the UK, Italy, Spain, Finland, Sweden, and Switzerland. The operative mortality results of 19 030 patients were analyzed, and data for an extensive array of potential risk factors for operative mortality were collected. The overall mortality rate for aortic valve replacement in the Roques et al. study (6%, 95% CI ¼ 5.27–6.78) was similar to the findings of Astor et al. The UK Heart Valve Registry is a national registry that includes data from over 30 clinical centers [9]. Data collection commenced in 1986 and approximately 50 000

OPERATIVE MORTALITY

371

patients were entered within the next decade. The observed 30 day mortality rate for 29 360 patients who underwent first-time aortic valve replacement was 4.3%. Numerous risk factors for operative mortality from aortic valve replacement have been identified, in addition to the pre-existence of an existing artificial heart valve or the performance of concurrent surgery noted above. These factors encompass both patient specific risk factors as well as risk factors associated with the healthcare facility and surgeon. For example, although aortic valve replacement may be performed successfully even in very elderly patients, age remains an important risk factor for operative mortality, as shown in the Astor et al. study [6]. A far lower operative mortality rate was observed in the Roques et al. study [8] for patients without any risk factors for operative mortality. This rate was only 1.1% (95% CI ¼ 0.36–2.56). Comorbid conditions comprise another group of operative risk factors. Risk factors for in-hospital mortality include New York Heart Association class status IVand the urgent nature of the aortic valve procedure. According to a study by Langanay [10], risk factors for in-hospital mortality included New York Heart Association class IV status (p < 0:01), renal insufficiency (p < 0:001), left ventricular ejection fraction < 40% (p < 0:01), and emergency surgery (p < 0:001). Renal insufficiency is a risk factor of particular importance. Astor et al. [6] found that patients with chronic renal failure were significantly more likely to experience in-hospital mortality (adjusted odds ratio ¼ 2.57, 95% CI ¼ 1.61–4.10). Jamieson et al. [7] observed that in the STS dataset, for aortic valve replacement without other concomitant cardiac surgery, the presence of dialysisdependent renal failure was significantly associated with operative mortality (odds ratio ¼ 4.32, 95% CI ¼ 2.83–6.43). Operative mortality following aortic valve implantation is affected by hospital and surgeon procedural volume. Astor et al. [6] found significantly lower operative mortality for patients who received their aortic valves at the healthcare facilities with the highest quartile of aortic valve implantations, compared to the healthcare facilities in the lowest quartile for such procedures (adjusted odds ratio ¼ 0.58, 95% CI ¼ 0.42–0.81). A study by Birkmeyer et al. [11] that employed Medicare claims data reported that for aortic valve replacement procedures, surgeon volume was inversely related to operative mortality (p < 0:001). Facilities and surgeons with higher procedural volumes may benefit patients through their greater experience with the procedures, particularly with unusual circumstances that may arise during a lengthy operation. Aortic valve replacement may be performed successfully in elderly patients. Given longer lifespans, the age required to fulfill the definition of very elderly continues to increase. An example of a study in octogenarians is that of Langanay et al. [10], who studied 771 patients aged 80 years and older who received aortic valve replacement. Not surprisingly, 99% of these patients received a bioprosthetic valve. Overall operative mortality was 10.1%. Kolh et al. [12] conducted a study of 83 consecutive patients aged 80 years or more who received aortic heart valve replacement. The operative mortality rate without concomitant cardiac surgery was 9%. A substantially greater operative mortality rate (24%) was observed when coronary artery bypass graft surgery was performed concomitantly. The studies of Langanay et al. and Kolh

372

CH24

CARDIOVASCULAR DEVICES: AORTIC VALVES

et al. argue strongly that aortic valve replacement when indicated is warranted in very elderly patients. Furthermore, their studies suggest that operative risk factors for very elderly patients are similar to the risk factors for the general population of aortic valve recipients.

The FDA guidance for replacement valves and the objective performance criteria (OPC) The FDA issued a guidance document for replacement heart valves in October 1994 [4]. The purpose of the guidance is to suggest to heart valve manufacturers tests that must be performed to generate data that will provide reasonable assurance of the safety and effectiveness of the new heart valves for their intended use. The premarket clinical evaluation for heart valves proposed in the guidance, the Objective Performance Criteria (OPCs), was based on the work of Gersh et al. [13]. The OPCs are based on a literature review of replacement heart valve articles published in peer-reviewed journals. The review was updated by Grunkemeir to include all articles which met the specific criteria and were published between 1985 and mid-1993 [4]. The OPCs presented in Table 24.1 are a set of acceptable linearized rates for complication of replacement heart valves marketed in the USA. The complications include thromboembolism, valve thrombosis, hemorrhage, perivalvular leak, endocarditis and non-structural dysfunction [4]. The OPCs do not include structural failures for mechanical valves, since the clinical studies on which they are based are too small and of insufficient duration to identify the rare but serious structural fractures of mechanical heart valve failures. Therefore, structural failures are more appropriately evaluated by premarket in vitro studies and monitored by postmarket surveillance. Structural failures such as leaflet breakage in mechanical heart valves can result in catastrophic events.

Table 24.1 Objective performance criteria (OPC) for complication rates (%/year) of heart valve studies Complications Thromboembolism Valve thrombosis Hemorrhage (major) All perivalvular leak (major) Endocarditis

Mechanical heart valve (%/year)

Tissue heart valve(%/year)

3.0 0.8 3.5 (1.5) 1.2 (0.6) 1.2

2.5 0.2 1.4 (0.9) 1.2 (0.6) 1.2

Adapted from Replacement Heart Valve Guidance, FDA, 1994, Appendix K for OPC values and Appendix L for definitions of complications.

THE NEED FOR EPIDEMIOLOGICAL STUDIES

373

Adverse event reports Adverse events reported to the FDA which involve heart valves include mechanical heart valves (38%) and biological heart valves (62%). Reported patient problems include endocarditis/infection, valve regurgitation/stenosis, thrombus, pannus formation, bleeding, embolism/stroke, and hemolysis. The reported device problems include peri- or paravalvular leaks, valvular structural deterioration, valve calcification, incorrect sizing, obstruction, and leaflet breakage. We will focus on the example of leaflet breakage. Between January 1994 and December 1998 the FDA received 33 adverse event reports (15 manufacturer, 11 user facility and 7 voluntary reports) involving 31 events of leaflet breakage for three different brands of mechanical heart valves [14]. These events were associated with 11 deaths and 20 injuries in which additional heart valve replacements were required. The FDA initiated follow-up for each of the 31 events. All 31 adverse events were analyzed to determine when leaflet breakage occurred (Figure 24.2). The analysis of the reported leaflet breakage events revealed 95% of manufacturer A and 100% of manufacturer B heart valves broke during heart valve surgery. This finding suggests that procedure-related factors associated with implantation or explantation might be the problem, rather than an intrinsic structural defect. Intraoperative leaflet breakage may be caused by the inappropriate use of clamps, retractors or rotators during surgery. The risk of leaflet breakage associated with procedural factors is addressed in the manufacturers’ instructions for use. In contrast, all seven reported leaflet breakages with heart valves of manufacturer C occurred 10–12 years after implantation. Patients experienced abrupt breathing difficulties followed by cardiac arrest. The leaflet breakages were discovered either by medical examinations or autopsies. The breakages might have been related to carbon coating defects and the impact of cavitation bubbles on the carbon surfaces of the heart valve [15]. These heart valves were withdrawn from the market in the 1980s because of the leaflet breakage problems.

The need for epidemiological studies Epidemiologic studies of valve performance in the postmarket environment are of great importance. Adverse event reports are a valuable component of the postmarket surveillance of aortic prosthetic valves. However, adverse event reports cannot provide incidence rates because the denominator (number of patients at risk ) is not provided by the adverse event reporting system, and underreporting of adverse events may adversely affect the accuracy of the estimate for the numerator. The inability of adverse event reports to provide incidence rates of various complications, such as thrombo-embolic events and paravalvular leak, is an important limitation on the value of these reports, since finding an adequate way to address the problem depends largely on knowledge of

374

CH24

CARDIOVASCULAR DEVICES: AORTIC VALVES

Number of events

25 20 15

Manufacturer A Manufacturer B

10

Manufacturer C

5 0 ge

Du

rin

g

ry

r su

1

da

1 y–

ye

ar 0 –1

1

ye

ar

ye

s ar

s

10

2 –1

ye

ar

Time to occurrence of leaflet breakage

Figure 24.2 Time to reported occurrence of leaflet breakage with mechanical heart valves (n ¼ 31). Data source: adverse event reports in MAUDE database (1994–1998)

the incidence rate. Epidemiologic studies, not surveillance, would be required to reliably provide this vital data. In addition to providing incidence data for valve-related complication rates, epidemiologic studies can provide evidence of risk factors for such complications. These risk factors may be intrinsic to the patient (e.g. age or comorbidities) or extrinsic to the patient (e.g. relating to the valve manufacturer, the healthcare facility, and/or the healthcare provider). Knowledge of risk factors is essential to efforts to reduce complications associated with heart valves. Epidemiologic studies of postmarket aortic valve performance face a diverse array of challenges. First, the valves have very lengthy survival expectancies, thus ideally requiring prospective cohort studies that would continue for a decade and more. Loss to follow-up in prospective cohort studies of this length is an important concern. The geographic mobility of patients, particularly within the USA, contributes to loss to follow-up among study participants. Second, the elderly nature of many study participants leads to a decrease in the study population over time from nonvalve-related mortality. It is crucial that sample size calculations account for this so that study power will be adequate. Third, in the postmarket environment in general, patients can receive a given heart valve without providing their consent to participate in a study. This is in sharp contrast to the premarket environment, where access to a new valve can only be obtained through participation in a study. For this reason, the recruitment of patients for studies in the postmarket environment may be more difficult than in the premarket environment. The ability of epidemiologic studies to ascertain the existence of risk factors for valverelated complications is of special importance when considering a product withdrawal of an aortic valve. For the purpose of this discussion, product withdrawal is defined as the voluntary or involuntary removal from the marketplace of a previously marketed heart valve. Patients implanted with thevalve subject to thewithdrawal and their clinicians will

THE NEED FOR EPIDEMIOLOGICAL STUDIES

375

benefit greatly from risk factor data. A drug withdrawal from the marketplace merely requires patients to cease using the withdrawn medication. Drugs can be eliminated by the hepatic and/or renal routes; devices cannot. As noted previously, the operative mortality from aortic valve implantation is appreciable. The risk from the withdrawn valve to the patient must be greater than the operative mortality risk for that patient for a recommendation of explant and replacement of the implanted valve to be warranted. Operative mortality may vary widely with the presence or absence of risk factors, such as advanced age and renal insufficiency. Similarly, epidemiologic studies may demonstrate that the complication risk from the withdrawn aortic valve may vary widely as well. Recommendations for the clinical management of patients implanted with a withdrawn aortic valve are not likely to be ‘one size fits all’. A complex matrix of recommendations that account for operative risk and valve-related complication risk is essential. These recommendations may evolve over time as new risk factor data become available. The management of patients with a withdrawn aortic valve may be an extraordinarily protracted process. Given the appreciable mortality rate for aortic valve explant and replacement, many patients will be advised not to undergo valve replacement surgery. Some patients who are advised to have their valves explanted and replaced may balk at undergoing the required surgery. The withdrawal of the Bjork–Shiley convexo–concave heart valve is particularly illustrative, as the management of the cohort of patients who received the withdrawn valve has extended over decades [16].

Patient tracking The location of patients with a withdrawn heart valve may be a significant challenge, given the mobility of the population of the USA. Tracking requirements may greatly facilitate patient location in the event of a device recall. Selected devices subject to tracking requirements are designed to facilitate the notification of device users and product recalls in matters that require prompt action [17]. A study by Kaczmarek et al. [18] involved a detailed analysis of an example of the utilization of the tracking system. Case reports revealed a fatal tachycardia caused by the malfunction of an implantable cardioverter defibrillator. This device is subject to the tracking regulations. Over 5000 patients required reprogramming of their implantable cardioverter defibrillators to prevent the tachycardia. The effort to locate these patients and reprogram their implantable cardioverter defibrillators was successful. Of the patients believed to be at risk for fatal tachycardia, more than 99.8% were located and their implantable cardioverter defibrillators reprogrammed. More than 98.7% of the patients had their devices reprogrammed in the first 60 days after the initiation of the notification and reprogramming effort. After 4 weeks of the effort, more than 94.5% of the patients had their implantable cardioverter defibrillators successfully reprogrammed. The analysis conducted by Kaczmarek et al. revealed risk factors for delayed patient location and treatment. Patients who changed their physician were significantly less likely to have their devices reprogrammed in the first week (odds ratio ¼ 0.73, 95%

376

CH24

CARDIOVASCULAR DEVICES: AORTIC VALVES

CI ¼ 0.63–0.86, p < 0:001) and significantly less likely within 4 weeks (odds ratio ¼ 0.61, 95% CI ¼ 0.45–0.82, p ¼ 0:001). Patient age was not related to device reprogramming in the first week, although older patients were significantly more likely than their younger counterparts to have their implantable cardioverter defibrillators reprogrammed within 4 weeks. The number of patients with the withdrawn device followed by a given physician proved to be an important variable. Patients whose physicians followed five or more patients with the withdrawn implantable cardioverter defibrillator were more than twice as likely (odds ratio ¼ 2.02, 95% CI ¼ 1.56–2.62) than their counterparts, whose physicians followed less than five such patients, to have their devices reprogrammed within 4 weeks. The study by Kaczmarek et al. provides powerful evidence that the recipients of tracked medical devices can be rapidly located and receive appropriate medical intervention in the case of a withdrawal of a device from the marketplace. Furthermore, the identification of risk factors for delayed patient location and treatment highlights subgroups where additional targeted efforts to locate and treat these individuals may further increase the rapidity of the overall response. The importance of locating patients implanted with a withdrawn medical device must not be underestimated. Patients who cannot be located cannot be treated. The failure to locate such patients will deny them the considerable benefits of optimal clinical management recommendations.

Access issues Equal access to medical care is a noteworthy governmental and societal goal. Concerns regarding racial inequality in access to important medical devices have been raised. For example, an epidemiologic study by Sharkness et al. [19] that employed data from the National Health Interview Survey, a nationally representative study that included 47 485 households and 122 310 persons, found that the prevalence of intraocular lenses was more than three-fold greater in whites than blacks. In contrast, Garver et al. [20] examined National Health Interview Survey data and failed to find evidence of racial inequality in access to heart valves, including aortic valves. Aortic valve implantation is a less discretionary procedure than the use of many other medical devices. The patient’s compelling clinical indication for aortic valve replacement may contribute to overcoming barriers to receiving the procedure.

Conclusions Prosthetic aortic valves are effective in the treatment of aortic stenosis and aortic regurgitation. The operative mortality risk is appreciable, and numerous risk factors have been identified for an unfavorable operative outcome. All prosthetic heart valves entail the risk of life-threatening complications, most notably thromboembolic events for mechanical valves and structural failure for bioprosthetic valves. The FDA has established objective performance criteria for the long-term performance of prosthetic

REFERENCES

377

heart valves. The long life expectancies of prosthetic heart valves pose challenges for their postmarket surveillance and management. Well-designed epidemiologic studies can be of great value in meeting those challenges.

References 1. Bonow RO, Carabello B, de Leon AC Jr et al. American College of Cardiology/American Heart Association guidelines for the management of patients with vascular heart disease. J Am Coll Cardiol 1998; 32: 1486–1588. 2. Grunkemeier G, Starr A, Rahimtoola SH. Clinical performance of prosthetic heart valves. In Alexander W, Schlant R et al., Hurst’s The Heart, 9th edn. New York: McGraw-Hill, 1998; 1851–1866. 3. Gott VL, Alejo DE, Cameron DE: Mechanical heart valves: 50 years of evolution. Ann Thorac Surg 2003; 76: SS2230–S2239. 4. Draft replacement heart valve guidance, Division of Cardiovascular Respiratory and Neurological Devices, Office of Device Evaluation, Center for Devices and Radiological Health, Food and Drug Administration, October 14 1994. 5. Schoen FJ. Future directions in tissue heart valves: impacts of recent insight from biology and pathology. J Heart Valve Dis 1999; 8: 350–359. 6. Astor BC, Kaczmarek RG, Hefflin B, Daley WR. Mortality after aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000; 70: 1939–1945. 7. Jamieson WR, Edwards FH, Schwartz M et al. Risk stratification for cardiac valve replacement. National Cardiac Surgery Database; Database Committee of the Society of Thoracic Surgeons. Ann Thorac Surg 1999; 67: 943–951. 8. Roques F, Nashef SA, Michel P et al. Risk factors and outcome in European cardiac surgery: analysis of the EuroSCORE multinational database of 19 030 patients. Eur J Cardiothorac Surg 1999; 15: 816–823. 9. Asimakopoulos G, Edwards MB, Taylor KM. Aortic valve replacement in patients 89 years of age and older: survival and cause of death based on 1110 cases: collective results from the UK Great Valve Registry. Circulation 1997; 96: 3403–3408. 10. Langanay T, DeLatour B, Ligier K et al. Surgery for aortic stenosis in octogenarians: influence of coronary disease and other comordidities on hospital mortality. J Heart Valve Dis 2004; 13: 545– 552. 11. Birkmeyer JD, Stukel TA, Siewers AE. Surgeon volume and operative mortality in the United States. N Engl J Med 2003; 22: 2117–2127. 12. Kolh P, Lahaye L, Gerard P, Limet R. Aortic valve replacement in octogenarians: perioperative outcome and clinical follow-up. Eur J Cardiothorac Surg 1999; 16: 68–73. 13. Gersh BJ, Fisher LD, Schaff HV et al. Issues concerning the clinical evaluation of new prosthetic heart valves. J Thorac Cardiovasc Surg 1986; 91: 460–466. 14. Manufacturer and user facility device experience (MAUDE) database, Center for Devices and Radiological Health (CDRH), Food and Drug Administration: http://www.accessdata.fda.gov/ scripts/cdrh/cfdocs/cfMAUDE/search.CFM [accessed September 30 2005]. 15. Schoen FJ, Padera RF Jr. Cardiac surgical pathology. In Cohn LH, Edmunds LH Jr (eds), Cardiac Surgery in the Adult. New York: McGraw-Hill, 2003: 119–185. 16. van Gorp MJ, Steyerberg EW, Van der Graaf Y. Decision guidelines for prophylactic replacement of Bjork–Shiley convexo-concave heart valves: impact on clinical practice. Circulation 2004; 109: 2092–2096.

378

CH24

CARDIOVASCULAR DEVICES: AORTIC VALVES

17. Medical Device Tracking; Guidance for Industry and FDA Staff. November 17, 2006. www. fda.gov/cdrh/comp/guidance/169.pdf. [Accessed November 30, 2006]. 18. Kaczmarek RG, Beaulieu MD, Kessler LG. Medical device tracking: results of a case study of the implantable cardioverter defibrillator. Am J Cardiol 2000; 85: 588–592. 19. Sharkness CM, Hamburger S, Kaczmarek RG et al. Racial differences in the prevalence of intraocular lens implants in the United States. Am J Ophthalmol 1992; 114: 667–674. 20. Garver D, Kaczmarek RG, Silverman BG, Gross TP, Hamilton PM. The epidemiology of prosthetic heart valves in the United States. Texas Heart Inst J 1995; 22: 86–91.

25 Hemostasis Devices Dale R. Tavris and Beverly Gallauresi US Food and Drug Administration, Rockville, MD, USA

Ralph G. Brindis Northern California Kaiser Permanente, Oakland, CA, USA

Manual compression has been the mainstay for attaining postprocedure hemostasis of the femoral artery following cardiac catheterization since Seldinger’s advance [1] of obtaining percutaneous vascular access through needle puncture, rather than requiring direct surgical cutdown and closure for performing catheterization. Although generally quite effective, manual compression has some significant disadvantages. Two of the most important disadvantages are patient discomfort and inconvenience. Manual sheath removal is usually delayed 3–4 hours after percutaneous coronary intervention (PCI) is performed, while awaiting the normalization of activated clotting time (ACT) or prothrombin time (PTT) levels. After sheath removal, an additional 6 or so hours of strict bed rest are needed. Delayed manual sheath removal after PCI is not infrequently accompanied by patient discomfort and at times vagally mediated hypotension. The use of hemostasis devices applied at the end of a PCI procedure obviates this problem and allows better use of valuable floor nursing staff, with the elimination of the intensive time requirements associated with manual ‘sheath pulling’. As a consequence of the above, when hemostasis is achieved by manual compression there are often high postprocedural management costs related to groin care oversight by healthcare personnel and overnight hospital stays. These problems, along with patient discomfort associated with manual or mechanical groin compression, have led to an increasing popularity of the use of various arteriotomy closure (hemostasis) devices. Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

380

CH25

HEMOSTASIS DEVICES

The ideal vascular hemostasis device Hemostasis devices are used as a sealant after an invasive procedure that involves the puncture (needle stick) of the femoral artery. The most common use of the hemostasis device is after a diagnostic or interventional cardiac catheterization, to prevent bleeding from the puncture site, as described in a current textbook [2]. The ideal closure device would be one that could be deployed simply and rapidly. The device would be well suited for patients, regardless of any anticoagulation regimens utilized, applicable for all of the varying sizes of groin sheaths, and suitable for variations in groin vascular anatomy, such as arterial vessel size or presence of peripheral arterial disease. In addition to being efficacious in sealing the groin puncture site, the ideal device would facilitate patient comfort and allow early and rapid postprocedural ambulation. Potential complications, including groin bleeding, hematoma, vascular injury resulting in arterio-venous fistulas or pseudo-aneurysms, arterial thrombosis, and infection would be less than or equal to that of manual compression.

Two types of vascular hemostasis devices There were two main types of vascular hemostasis devices marketed by the end of 2003 (only devices marketed by the end of 2003 are discussed in this chapter because there is currently insufficient postmarket data on devices marketed after that time): suture devices and collagen plug sealant devices. In addition, although topical sealant devices may also be used, it should be noted that none of these devices have been approved by the FDA for femoral artery closure, and it is recommended in the manufacturer’s instructions for use that they be used following cardiac catheterization only in conjunction with manual compression.

Suture devices The suture devices function by performing a surgical ligature closure at the arterial puncture site. Perclose was the first suture type device used for the femoral artery hemostasis, manufactured by Abbott Vascular Devices, a division of Abbott Laboratories. There have been several changes to the first generation of the Perclose devices, which have simplified the use of the device [3]. In July 2003, an enhanced version of the Perclose A-TTM suture-mediated closure system was introduced to further improve ease of use and potentially reduce the vessel closure procedure time. Features added to the Perclose A-T included a new suture trimmer and numbered procedure deployment steps printed on the device. The new trimmer facilitates both knot advancement and suture trimming, eliminating a potentially time-consuming step in the vessel closure procedure for many operators. Human factors engineering, the scientific and systematic application of knowledge about human behavior into the design of products, was incorporated into the design of the trimmer to create a device that is more comfortable and intuitive to use.

THE IDEAL VASCULAR HEMOSTASIS DEVICE

381

In May 2004, Abbot announced the launch of the next generation Perclose1 suturemediated vessel closure system, Perclose ProGlideTM, featuring the polypropylene monofilament suture. The device received approval from the FDA during the first quarter of 2004. With monofilament suture, the ProGlide knot advances to the opening in the artery, allowing a single operator to deliver a pre-tied knot to close the access site in the femoral artery following a catheterization procedure. Monofilament suture is a preferred suture for vascular surgeons due to its high knotted tensile strength and minimized tissue reaction. Suture devices such as Perclose can be used for anticoagulated patients, but are contraindicated for use with significant underlying peripheral arterial disease or an arterial puncture distal to the common femoral artery bifurcation. Confirmation of absence of peripheral arterial disease and the presence of an anatomically suitable arterial puncture site is achieved by performing femoral angiography, typically through the groin sheath prior to anticipated device deployment.

Collagen plug sealant devices Sealant devices include Vasoseal [4], Angio-Seal [5], the Duett devices [6], and the QuickSeal CCS Arterial Closure System [7]. Developed by Datascope Corp. in May 1995, Vasoseal was the first device to rapidly stop bleeding after diagnostic and interventional cardiology and radiology procedures. The device permits delivery of the collagen into the tissue tract created by removal of a sheath device and onto the exterior surface of the artery. The collagen interacts with the blood in order to create a hemostatic seal directly over the puncture wound in the artery. The collagen is sterile, nonpyrogenic and absorbable. The presence of peripheral arterial disease is not a contraindication for its use. The Duett sealing device, manufactured by Vascular Solutions and approved by FDA in June of 2000, is a two-component system – a familiar balloon catheter combined with a procoagulant of bovine microfibrillar collagen and thrombin. While the balloon catheter temporarily seals the arteriotomy, the procoagulant is delivered to the entire arterial access site. The procoagulant initiates the body’s own clotting mechanisms to form a seal of both the arteriotomy and tissue tract. Of some concern is the risk of intraarterial deposition of the procoagulant material, causing vessel thrombosis if meticulous attention to deployment technique is not followed. The Angio-Seal Hemostatic Puncture Closure Device, by St. Jude Medical, was approved in September 1996. It is composed of an absorbable bovine collagen sponge and a specially-designed absorbable polymer anchor that are connected by an absorbable self-tightening suture. The device seals and sandwiches the arteriotomy between its two primary members, the anchor and collagen sponge. Hemostasis is achieved primarily by the mechanical means of the anchor–arteriotomy sandwich. The collagen sponge then dissolves within 60–90 days. Since Angio-Seal has an intra-arterial component, significant peripheral arterial disease and an arterial puncture distal to the common femoral artery are contraindications for its use (Angio-Seal is the only collagen plug device for which peripheral arterial disease is a contraindication for use).

382

CH25

HEMOSTASIS DEVICES

Approved in March 2002, the QuickSeal Arterial Closure System, manufactured by Sub-Q, Inc., helps the body make a clot at the place the catheter enters the blood vessel. QuickSeal is made of two parts: a sponge-like material called Gelfoam and a rigid plastic tube used to place the Gelfoam at the puncture site. After the catheter is removed from the artery, the device is slipped into the same hole, stopping just short of the artery. The Gelfoam passes through the tube so that it covers the hole in the artery. The tube is then removed and the Gelfoam stays in place, where it helps a clot to form. Manual compression is still necessary with this device, but it can be applied for a shorter time. The labeling notes that QuickSeal should not be used in patients who have sensitivity or allergy to materials made from pigs, blood clotting problems, or uncontrollable high blood pressure.

Non-invasive patches The topical ‘non-invasive’ patches, Chito-Seal and the Syvek patch, are used adjunctively to manual compression and work by accelerating the natural clotting cascade when in contact with blood. As these topical pads are not invasive, identification of peripheral arterial disease and localization of the arterial puncture is not required for their use. They are not approved by the FDA for vascular hemostasis. All of the suture and sealant devices have an inherent ‘learning curve’ before the operator is truly competent in their use. Differing clinical situations may favor one device over another, for example, the use of an intra-arterial anchor such as Angio-Seal or a suture device in the anticoagulated patient. If early repeat access is anticipated, the Perclose or Duett device might be viewed preferentially to the Angio-Seal device, where 30 days is required for the absorption of the anchor and the plug. In the majority of clinical situations, however, the choice of the hemostasis device is mostly a matter of operator preference.

Origin of FDA concern with the postmarket performance of hemostasis devices Hemostasis devices were first marketed in the USA in 1996 [8]. Between the beginning of 1996 and the end of the year 2000, the Center for Devices and Radiological Health (CDRH) of the FDA received 1880 medical device adverse event reports of serious injuries and 36 reports of death associated with the use of these devices, in accordance with the federal Medical Device Reporting (MDR) regulation [9]. A ‘serious injury’ is defined in the Code of Federal Regulations by the regulation as an injury or illness that is life-threatening, results in permanent impairment of a body function, or permanent damage to a body structure, or necessitates medical or surgical intervention to preclude any of the above. More details about the reporting regulations are presented in Chapter 2 of this book.

ORIGIN OF FDA CONCERN WITH THE POSTMARKET PERFORMANCE

383

Three of the most frequently reported serious injuries associated with hemostasis devices were hemorrhage from the femoral artery, hematoma, and infection. Reported serious injuries during the above-noted time frame included 503 reports of hemorrhage, 331 reports of hematoma, and 190 reports of infection. These reports of serious injuries and deaths were of particular concern to the FDA because the main stated rationale for the use of hemostasis devices is patient discomfort and convenience, rather than safety or effectiveness. As noted above, the alternative to the use of hemostasis devices, for the purpose of stopping femoral artery bleeding following cardiac catheterization, is manual compression of the femoral artery. Compared to manual compression, the use of hemostasis devices are less labor-intensive, and they offer the advantages of quicker discharge of patients, resulting in potential cost savings. However, if these were the only advantages the hemostasis devices have to offer, it would be difficult to justify their use in the face of large numbers of excess serious injuries and deaths. Premarket testing of the hemostasis devices had demonstrated the same types of adverse events as were being reported to FDA in the postmarket period. But the number of reports being received was of concern and suggested the need for additional investigation in order to ascertain whether or not use of hemostasis devices in the postmarket period was associated with a greater incidence of serious adverse events than had been shown with premarket testing. It should be noted that the existence of reports of serious injuries and deaths associated with the use of hemostasis devices did not prove that these devices were less safe than their alternative, i.e. manual compression of the femoral artery. Serious injuries associated with manual compression would not be likely to be reported to the FDA. Therefore, the FDA had no basis for comparison of serious injury incidence between hemostasis device users versus non-users, and little basis for causal attribution of the serious injuries and deaths to the hemostasis devices. Another reason for FDA concern over the hemostasis devices was the fact that females appeared to be at much greater risk than males (that information had not been available from premarket data analysis). There were many more reports of serious injuries and deaths in females than in males, even though males are known to undergo cardiac catheterization much more frequently than females. In summary, our initial analysis of reports to CDRH of serious adverse events associated with the use of hemostasis devices left us with two interrelated concerns: the safety of these devices in general; and their apparent high risk to women, compared to men.

Gender differences The demonstration of a higher risk in women than in men for local adverse events following cardiac catheterization has been noted by several researchers, either associated with the use of hemostasis devices [10–13] or irrespective of the use of hemostasis devices [14–17]. Possible reasons for this include smaller vessel size, differences in pelvic anatomy, and hormonal characteristics of women, although there is no general agreement on this issue.

384

CH25

HEMOSTASIS DEVICES

FDA first used readily available data to test the hypothesis that females were at higher risk than males for the occurrence of serious injuries associated with the use of hemostasis devices. Population at risk was estimated as the number of percutaneous trans-vascular coronary angioplasty (PTCA) procedures performed in 1996, as estimated by the National Center for Health Statistics (NCHS). The NCHS estimated 214 000 of these procedures in females and 452 000 in males [18]. Rates of reported serious injuries by gender were then calculated as the number of reports to FDA, divided by the above NCHS estimates of patients at risk. These estimates do not portray actual risks associated with hemostasis devices, but rather only relative risks by gender. In the first place, it is widely recognized that reports of adverse events to the FDA represent only a small fraction of those that actually occur [19]. Second, the NCHS estimates may not be representative of all cardiac catheterizations during the applicable time period. They included data only from 1996 (the last year for which this data was available at the time this study was undertaken), and only interventional, not diagnostic, cardiac catheterizations. The rationale for this was that serious adverse events were much more likely to occur in association with interventional, as compared with diagnostic, cardiac catheterizations [20]. Furthermore, only a fraction (probably about 20%) of the cases of cardiac catheterization were associated with the use of hemostasis devices. Nevertheless, if one assumes that these obstacles to an accurate determination of absolute risks of serious adverse events apply approximately equally to men and women, then the resulting calculation should provide an accurate estimate of the relative risk for women, as compared to men. Using these methods, FDA showed that the relative risk for women (compared to men) of serious injuries associated with hemostasis device use varied from 1.92 to 2.54 between 1996 and the end of 2000, with a mean of 2.30 [21] (Tabe 25.1). Subanalysis by type of hemostasis device and type of serious injury provided additional interesting comparisons. By type of serious adverse event, females exhibited high relative risks for hemorrhage (RR ¼ 2.67, p < 0:001) and hematoma (RR ¼ 2.61,

Table 25.1 Relative risks by gender of reported hemostasis device-associated serious injuries by year, 1996–2000 (n ¼ 1024) [21] Rate per 100 000 coronary angioplasties Year of report

Male

Female

Female:male risk ratio

1996 1997 1998 1999 2000

2.4 25.2 31.2 23.0 26.8

4.6 51.9 77.6 58.4 56.5

1.92 2.06 2.49 2.54 2.11

Mean

21.7

49.8

2.30

ORIGIN OF FDA CONCERN WITH THE POSTMARKET PERFORMANCE

385

p < 0:001), but not for infection (RR ¼ 1.21, p > 0:05). The lack of a high female relative risk for infection tends to support the likelihood that the high female relative risk for hemorrhage and hematoma was not simply the result of a reporting bias. Further subanalysis by type of device revealed that for hemorrhage, the female relative risks for the two collagen plug devices (RR ¼ 3.10, p < 0:001 and RR ¼ 3.28, p < 0:001, respectively) were much greater than the relative risk for the arterial suture device (RR ¼ 1.50, p ¼ 0:05). The relative risks for hematoma demonstrated a similar pattern. These findings, although suggestive that females are at higher risk than males for serious injuries associated with the use of hemostasis devices, do little to explain the reasons for this high female relative risk. Neither do they prove that these high relative risks are specifically due to the use of hemostasis devices. In order to assess these gender differences in more detail, while avoiding some of the pitfalls noted above, FDA collaborated with the American College of Cardiology in a study that made use of their new National Cardiovascular Data RegistryTM (ACCNCCDRTM) [22]. By utilizing a prospective registry, such as the ACC-NCCDRTM, FDA was able to assess actual rates of adverse events following cardiac catheterization, rather than merely relative rates (based on passive reporting), as in the above-noted study. Nevertheless, the relative risk for women of local vascular adverse events following cardiac catheterization was very similar in this study to that noted above, both in the univariate analysis (RR ¼ 2.13) and the multivariate analysis (OR ¼ 2.30, 95% CI ¼ 2.12–2.50) [23]. To further evaluate these gender differences we conducted another study on the same ACC registry, but with some new variables added [24]. In this new analysis we were able to distinguish between the different types of collagen plug hemostasis devices and also assess the effect of sheath size. We found in this analysis that the higher rate of local vascular adverse events in women as compared to men applied to both collagen plug devices (Vasoseal and Angio-Seal), and to subjects who were given manual compression as well. The consistency with FDA’s previous analysis, which looked only at relative rates and utilized NCHS data [18], is worth noting. Both studies showed high relative risks for women associated with the collagen plug devices, but not with the suture device, Perclose. The size of the sheath through which the needle was inserted was also important. The rate of vascular adverse events increased progressively with sheath sizes of 4–8, going from 1.30% to 5.18% in men and from 1.82% to 9.20% in women. However, the increasing adverse event rate with increasing sheath size was more pronounced in women, as evidenced by an increase in relative risk from sheath size 6 (RR ¼ 1.38) to 7 (RR ¼ 1.73). These findings may be significant for prevention. If the high relative risk in women is largely due to small vessel size, using a smaller hemostasis device may help to ameliorate this problem [13]. If this is true, this could be important information in the consideration of what sheath size to use, especially in women who will be receiving collagen plug devices.

386

CH25

HEMOSTASIS DEVICES

Safety issues More important than the issue of relative risk by gender is the question of the safety in general of the hemostasis devices. Adverse events reported to the FDA provided no basis for assessing this quantitatively. Studies that have assessed complications following the use of hemostasis devices, compared with manual compression controls, have demonstrated widely variable results. Some studies have shown hemostasis devices to be safer, and others have shown manual compression to be safer. Most studies have shown no difference [25–36]. This difference in study results has little to do with the type of hemostasis device studied, as these studies have shown a positive association between adverse events and the use of hemostasis devices when Angio-Seal [37–39], Vasoseal [39,40], and Perclose [37,41] have been used, as well as a negative association between all three of these devices [42] and adverse events. The wide difference in results from these studies is perplexing and deserves comment. Those studies that failed to demonstrate any significant differences were generally quite small, and therefore were lacking in power. The differences between those studies that demonstrated favorable results for hemostasis devices and those that demonstrated unfavorable results, as compared to manual compression controls, could have been related to the fact that these studies were generally performed in a single institution. Some of these institutions may have been characterized by a higher skill level than others in the use of these devices. Another reason for different conclusions regarding the safety of hemostasis devices in different studies is that these studies assess different outcomes. For example, one study found a very high rate of a relatively minor adverse events (prolonged bleeding) in controls (13%) compared to device users (1.3%), but no difference in the rate of more serious adverse events [43]. Another study showed no difference in the rate of major complications (probably due to insufficient power) but significantly more minor adverse events in device users than in controls [33]. And another study showed significantly more infectious complications with the use of Perclose, but significantly less hemorrhagic complications, as compared with manual compression controls, yet significantly more complications requiring surgical correction [44]. Procedural variables can also be important, as shown by a study which demonstrated a higher rate of adverse events in Angio-Seal compared to controls when heparin was not used, but a higher rate in controls when heparin was used [25]. Two recent large meta-analyses have been performed, which assess the results of randomized controlled clinical trials that have assessed the performance of hemostasis devices [45,46], and which overlapped with each other with respect to the studies that they included to a substantial degree. Koreny et al.[45] conclude their meta-analysis of 30 clinical trials, which include a total of about 4000 patients, by saying that when analysis was limited to studies that used intention-to-treat analysis, hemostasis devices were associated with a higher risk of hematoma and pseudo-aneurysm [38]. Intention-to-treat is a key issue when comparing safety and efficacy of hemostasis devices with that of manual compression. Multiple groin punctures, ‘back wall’ arterial

FDA STUDY TO EVALUATE THE RISK ASSOCIATED WITH HEMOSTASIS DEVICE USE

387

punctures, the demonstration of peripheral vascular disease on femoral angiography, or the development of a groin hematoma during the catheterization all lead the physician to often choose the strategy of manual compression rather than deployment of a hemostasis device. These clinical events themselves are clearly associated with high risk of groin complications, and therefore would bias studies against manual compression if intention-to-treat analysis was not the strategy invoked in data analysis. Although Koreny et al. note that their analysis was not influenced by the type of device, using the data from their tables shows that Vasoseal performed substantially worse against controls than either Perclose or Angio-Seal with respect to the three most frequently reported outcomes, hematoma, bleeding, and pseudo-aneurysm. Vasoseal was characterized by an 18.5% rate of hematoma, compared to 14.0% for controls (p ¼ 0:05), 3.4% vs. 2.0% for bleeding (p ¼ 0:19) and 4.9% vs. 2.9% for pseudoaneurysm (p ¼ 0:06); Angio-Seal and Perclose, on the other hand, performed slightly better than controls with respect to these outcomes. The meta-analysis performed by Vaitkus [46] looked at 16 clinical trials, which included 5048 patients, and concluded that, although when the analysis of several safety outcomes included all types of hemostasis devices together the devices performed better than controls (OR ¼ 0.89), when looked at separately, Angio-Seal performed better than controls (OR ¼ 0.51), Vasoseal performed worse than controls (OR ¼ 1.18), and Perclose performed the same as controls (although Perclose performed better than controls following diagnostic cardiac catheterization). The results of these meta-analyses must be interpreted with caution. First, even the metaanalysis that contained the larger number of patients included only 5048, and the one that contained the larger number of institutions included only slightly more than 30 institutions. Furthermore, the results of the various studies varied greatly. For example, although Vaitkus’ analysis demonstrated significantly more adverse events in patients receiving Vasoseal than in controls, this analysis included only eight studies, two of which demonstrated significantly fewer averse events in Vasoseal users than in controls [43,47], twowhich demonstrated about the same rate for Vasoseal as for controls [27,48], and four which demonstrated a nonstatistically significant higher rate in Vasoseal users than in controls [28,33,40,49]. One may conclude from all of this that the performance of these devices is greatly dependent upon the institution in which they are used and the skill of the individual users.

FDA study to evaluate the risk associated with hemostasis device use In summary, the literature on the safety of the hemostasis devices demonstrates widely varying results, depending upon the institution where the research takes place, although there is some suggestion that Vasoseal may fair poorly compared with the other hemostasis devices, or in comparison with controls. If the safety of hemostasis devices is highly variable, and dependent upon where the research is performed, then reporting bias is likely to substantially bias any review of the medical literature on this subject, including meta-analyses of that literature. The only way to obviate that problem is to

388

CH25

HEMOSTASIS DEVICES

conduct research on a sample of institutions that is more broadly representative of the medical community than would be expected of any study that involved only a single or very small number of institutions. For these reasons, FDA conducted a study that utilized data obtained from the American College of Cardiology, National Cardiovascular Data RegistryTM (ACCNCDRTM) [23]. That study [23], which utilized ACC-NCDRTM data from the year 2001, included 166 680 cardiac catheterizations performed in 214 different institutions throughout the USA. Seven local complications associated with the femoral artery catheterization site were assessed, including bleeding, occlusion, loss of distal pulse, artery dissection, pseudoaneurysm, A-V fistula, and death associated with any of these complications. Because follow-up of patients after hospital discharge was not planned for, analysis of the incidence of infectious complications could not be fully assessed, since many infections would present clinically after hospital discharge. Rates were calculated for each of these complications, for each of the three most prevalent types of hemostasis devices in use at the time (the two collagen plug devices, and the arterial suture device). Outpatients were excluded from the analysis, as were any patients for whom critical data was lacking. Following a univariate analysis that assessed complication rates by gender, type of procedure (i.e. diagnostic vs. therapeutic cardiac catheterization) and type of hemostasis (i.e. collagen plug device vs. arterial suture device vs. manual compression controls), a multivariate analysis was performed in order to control for potentially confounding variables. Step-wise backwards multiple logistic regression analysis was performed, using each of the seven outcome variables individually as the dependent variables, and using age, gender, race, type of procedure, type of hemostasis, body mass index, and several indices of comorbidity (New York Heart Association classification, presence of diabetes, hypertension, peripheral vascular disease, acute myocardial infarction, left main coronary artery stenosis, shock, history of congestive heart failure and acute renal failure, and emergency vs. elective status of the procedure) as independent variables. The sample size for the univariate analysis was 166 680, including 113 025 controls, 25 495 uses of the suture device, and 28 160uses of the collagen plug devices.Serious adverse events were reported in 1.56% of patients, the most common being bleeding (1.13%). They were more frequent in patients who had interventional cardiac catheterization (RR ¼ 2.26, p < 0:0001), and they were less frequent in patients who used the collagen plug devices (RR ¼ 0.62, p < 0:001) or the suture device (RR ¼ 0.87, p ¼ 0:02). The sample size for the multivariate analysis was 156 853, and included 2418 serious adverse events (1.54%). In addition to the risk factors noted above, several comorbid conditions were also significantly associated with serious adverse events (OR ¼ 2.15 for left main coronary artery stenosis; 1.08 for New York Heart Association Classification; 1.29 for peripheral vascular disease; 1.31 for myocardial infarction; 1.48 for history of renal failure; 1.10 for hypertension; 1.84 for shock; and 1.60 for emergency indication for the procedure). The hemostasis devices, as was the case in the univariate analysis, continued to demonstrate a negative association with serious adverse events in the multivariate

FOLLOW-UP FDA STUDY TO ASSESS THE SAFETY OF HEMOSTASIS DEVICES

389

analysis (for the collagen plug devices, OR ¼ 0.79, 95% CI ¼ 0.70–0.90; for the arterial suture device, OR ¼ 0.86, 95% CI ¼ 0.76–0.97). Thus, the control for confounding variables did not change the association between the arterial suture device and serious adverse events, whereas it somewhat attenuated, but did not erase, the association between the collagen plug devices and serious adverse events. However, analysis of data collected from the revised registry, which differentiated between the two different types of collagen plug devices, demonstrated a substantial difference between the two. In this analysis, Vasoseal was shown to be positively associated with adverse events [24].

Possible reasons for FDA findings of apparent protective effects of hemostasis devices Why did the FDA study demonstrate an apparent protective effect for some hemostasis devices? One possibility is that there are one or more confounding variables that account for this effect. The results of this study do indicate that patients with various serious coexisting cardiovascular problems were at relatively high risk for complications, and that patients were more likely to receive hemostasis devices if they did not have these coexisting problems. Nevertheless, the protective effect remained substantial and highly statistically significant (p ¼ 0:0004), even after controlling for numerous comorbid conditions. Given that the difference in risk associated with the hemostasis devices and the controls was not very large, there could have been unmeasured factors, connected with both patient risk and the decision of whether or not to use hemostasis devices, which accounted for that difference. In other words, physicians could have, on average, selectively chosen relatively low-risk patients to receive hemostasis devices. As previously discussed, decisions to not use a hemostasis device in situations where an injury occurred to the vessel wall during the procedure, when a groin hematoma occurred during the catheterization itself, or when a ‘predeployment’ femoral angiogram demonstrated the puncture site to present a risk for the use of a hemostasis device, would all bias the study results against manual compression. This type of situation probably accounted, at least in part, for the apparent protective effect of hemostasis devices. Another unmeasured confounding variable (for interventional cardiac catheterizations) could have been sheath size. This could have been associated with both an increase in adverse events and in the decision not to use hemostasis devices, thus confounding the results of this analysis.

Follow-up FDA study to assess the safety of hemostasis devices One of the biggest problems with the first FDA study was the fact that the two collagen plug devices were not differentiated. Consequently, a second study was undertaken, where a smaller number of institutions were recruited to revise their data collection so that, among other things, the two collagen plug devices, Vasoseal and Angio-Seal, could be differentiated.

390

CH25

HEMOSTASIS DEVICES

This follow-up study* [50] resulted in two new and surprising findings: Compared to manual compression, none of the hemostasis methods demonstrated an odds ratio significantly different from one with respect to ‘any vascular complication’, except for Vasoseal, which was characterized by an odds ratio of 2.38 (p ¼ 0:0004). Vasoseal also exhibited a significantly higher rate of entry site bleeding (OR ¼ 3. 97, p ¼< 0:0001) and hematoma (OR ¼ 1.96, p ¼ 0:04). Thus, not only was the protective effect demonstrated in the first study no longer apparent, but one of the two collagen plug devices, Vasoseal, was demonstrated to be associated with a significantly greater rate of serious adverse events than any of the other devices or the controls. We can only speculate on the apparently relatively worse performance of Angio-Seal in this study, as compared with the original (Phase I) study. In our Phase I study we speculated that unmeasured confounding variables might account for at least some of the apparent protective effects. In other words, physicians might be less likely to utilize these devices in situations where the local vascular situation gives them cause to worry about potential local adverse events. If that was the reason for some of the apparent protective effect shown in the Phase I study, it might be that, as more experience was gained with these devices over time, physicians became less likely to avoid using them in high-risk situations, thus resulting in an amelioration of the apparent protective effects. With regard to the new Vasoseal findings, we believe that our findings of greater risk of local vascular complications associated with Vasoseal (compared to either manual compression or the other most commonly used hemostasis devices) are attributable to the device itself rather than some confounding factor. This is suggested by the relatively high relative risk, the consistency with meta-analyses of clinical trials showing that the risk of Vasoseal is significantly greater than the other devices, and the fact that the association remains high after controlling for numerous potential confounding variables. Nevertheless, the possibility remains that the high Vasoseal risk could have been due to unidentified confounding variables, given that Angio-Seal and Perclose are contraindicated in the presence of significant peripheral vascular disease – although this was controlled for in the FDA analysis.

Conclusion The main reason for use of vascular hemostasis devices following cardiac catheterization is to reduce patient discomfort and inconvenience, as well as the cost of medical care. Therefore, the receipt by the FDA of almost 2000 reports of serious injuries or deaths associated with these devices caused much concern and led to the conducting of two FDA studies to further assess the safety of these devices. The first study demonstrated a small apparent protective effect of these devices, compared to manual compression controls, although unidentified confounding variables as the cause of this apparent protective effect could not be ruled out. The fact that the second study failed to confirm the apparent protective effect shown in the first study *This study was funded by the FDA’s Office of Women’s Health.

REFERENCES

391

strengthened the view that unidentified confounding variables in the first study were indeed the cause of the apparent protective effect of the hemostasis devices. Apparently, as physicians became more comfortable in using hemostasis devices, they became more inclined to use them in the same kind of high-risk situations in which manual compression is used – thus eliminating an important source of confounding. The other new major finding of the second study was that one of the collagen plug devices, Vasoseal, was associated with a significantly higher risk of serious adverse events than any of the other devices and the manual compression controls. The first study was unable to demonstrate this because it did not differentiate the two collagen plug devices, but rather evaluated them as a group. This should serve as a reminder of the need for postmarket surveillance systems to evaluate different devices individually rather than considering them as a group.

References 1. Seldinger SI. Catheter replacement of the needle in percutaneous arteriograph: a new technique. Acta Radiol 1953; 39(5): 368–376. 2. Topol EJ (ed.). Textbook of Interventional Cardiology, 4th edn. Philadelphia: Saunders, 2003; 795–798. 3. Abbott Laboratories, Perclose website: http://abbottvasculardevices.com/products/index. php?category ¼ 3 [accessed July 20 2005]. 4. Datascope, VasoSeal website: http://www.datascope.com/ip/ [accessed July 20 2005]. 5. St. Jude Medical, AngioSeal website: http://www.sjm.com/devices/device.aspx?name ¼ AngioSeal%26%23153%3BþVascularþClosureþDevice&location ¼ in&type ¼ 16 [accessed July 20 2005]. 6. Vascular Solutions, Duett website: http://www.vascularsolutions.com/products/duetttech.php [accessed July 20 2005]. 7. SUB-Q, Quick-Seal CCS website: http://www.fda.gov/cdrh/pdf/P010049.html [accessed July 20 2005]. 8. New Device Stops Bleeding Faster: FDA Consumer 1996; 30(1). 9. United States Food and Drug Administration (US FDA). Manufacturer and user facility device experience; MAUDE Database, 2001: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/ cfMAUDE. 10. Carere RG, Webb JG, Miyagishima R, Djurdev O et al.. Groin complications associated with collagen plug closure of femoral arterial puncture sites in anticoagulated patients. Cathet Cardiovasc Diagn 1998; 43: 124–129. 11. Sesana M, Vaghetti M, Albiero R, Corvaja N et al.. Effectiveness and complications of vascular access closure devices after interventional procedures. J Invas Cardiol 2000; 12: 395–399. 12. Eggebrecht H, von Birgelen C, Naber C, Kroger K et al. Impact of gender on femoral access complications secondary to application of a collagen-based vascular closure device. J Invas Cardiol 2004; 16(5): 247–250. 13. Aggarwal K. Vascular closure device complications: the case is not closed yet [commentary]. J Invas Cardiol May 2004; 16(5): 251. 14. Juergens CP, Semsarian C, Keech AC. Beller EM, Harris PJ. Hemorrhagic complications of intravenous heparin use. Am J Cardiol 1997; 80: 150–154.

392

CH25

HEMOSTASIS DEVICES

15. Motwant JG, Topol EJ. Aorto-coronary saphenous vein graft disease: pathogenesis, predisposition and prevention. Am Heart Assoc Inc 1998: 97(9): 916–931. 16. Johnson LW, Esente P, Giambartolomel A, Grant WD et al.. Peripheral vascular complications of coronary angioplasty by the femoral and brachial techniques. Catheteriz Cardiovasc Dx 1994; 31: 165–172. 17. Dauerman HL, Andreou C, Perras MA, Spinner JS et al.. Predictors of bleeding complications after rescue coronary interventions. J Thromb Thrombolysis 2000; 10: 83–88. 18. United States Centers for Disease Control and Prevention (US CDC)/National Center for Health Statistics (NCHS). National Hospital Discharge Survey, 1996. Series 13, Nos 139–140. 19. United States General Accounting Office (US GAO). Medical devices: early warning of problems is hampered by severe underreporting (GAO/PEMD-87-1, December 19 1986). 20. Silba S. Review: rapid hemostasis of arterial puncture sites with collagen in patients undergoing diagnostic and interventional cardiac catheterization. Clin Cardiol 1997; 20: 981–992. 21. Tavris DR, Gallauresi B, Rich SE, Bell C. Relative rates of serious injury and death associated with the use of hemostasis devices by gender. Pharmacoepidemiol Drug Safety 2003; 12: 237–241. 22. Brindis RG, Fitzgerald S, Anderson V, Shaw RE et al.. The American College of Cardiology – National Cardiovascular Data RegistryTM (ACC-NCDRTM): building a national clinical data repository. J Am Coll Cardiol 2001; 37(8): 2240–2245. 23. Tavris DR, Gallauresi BA, Lin B, Rich SE et al.. Risk of local adverse events following cardiac catheterization by hemostasis device use and gender. J Invas Cardiol 2004; 16(9): 459–464. 24. Tavris DR, Dey S, Albrecht-Gallauresi B, Brindis R et al.. Risk of adverse events following cardiac catheterization by hemostasis device use – Phase II. J Invas Cardiol (in press). 25. Kussmaul WG III, Buchbinder M, Whitlow PL, Aker UT et al.. Rapid arterial hemostasis and decreased access site complications after cardiac catheterization and angioplasty: results of a randomized trial of a novel hemostatic device. Am J Cardiol 1995; 25(7): 1685–1692. 26. Gerckens U, Cattelaens N, Lampe E, Grube E. Management of arterial puncture site after catheterization procedures: evaluating a suture-mediated closure device. Am J Cardiol 1999; 83: 1658–1663. 27. Sanborn TA, Gibbs HH, Brinker JA, Knopf WD et al.. A multicenter randomized trial comparing a percutaneous collagen hemostasis device with conventional manual compression after diagnostic angiography and angioplasty. J Am Coll Cardiol 1993; 22: 1273–1279. 28. Camenzind E, Grossholz M, Urban P, Dorsaz PA et al.. Collagen application vs. manual compression: a prospective randomized trial for arterial puncture closure. J Am Coll Cardiol 1994; 24: 655–662. 29. Legrand V, Doneux P, Tilman CS. Comparison of puncture site closure with collagen plug (Vasoseal) or by early manual compression following PTCA. Circulation 1993; 88: 1–72. 30. Nagtegaal EM, Schalij MJ, Buis B. Routine use of collagen to seal the femoral artery; puncture site after percutaneous transluminal coronary angioplasty in fully anticoagulated patients: a clinical evaluation. J Am Coll Cardiol 1993; 21: 231A. 31. Silber S, Bjorvik A, Rosch A, Muhling H. Advantages of sealing arterial puncture sites after PTCA with a single collagen plug: a randomized, prospective trial. J Am Coll Cardiol 1995; 23: 262A. 32. Kiemeneij F, Laarman GJ. Improved anticoagulation management after Palmaz–Schatz coronary stent implantation by sealing the arterial puncture site with a vascular hemostasis device. Cathet Cardiovasc Diagn 1993; 30: 317–322. 33. Slaughter PM, Chetty R, Flintoft VF, Lewis S et al.. A single center randomized trial assessing use of a vascular hemostasis device vs. conventional manual compression following PTCA: what are the potential resource savings? Cathet Cardiovasc Diagn 1995; 34(3): 210–214.

REFERENCES

393

34. Kapadia SR, Raymond R, Knopf W, Jenkins S et al.. The 6 FR angio-seal arterial closure device: results from a multimember prospective registry. Am J Cardiol 2001; 87(3): 789–791. 35. Rickli H, Uterweger M, Sutsch G, Prunner-La Rocca HP et al.. Comparison of costs and safety of a suture-mediated closure device with conventional manual compression after coronary artery interventions. Catheteriz Cardiovasc Intervent 2002; 57: 297–302. 36. Cura F, Kapadia SR, L’Allier L, Schneider JP et al.. Safety of femoral closure devices after percutaneous coronary interventions in the era of glycoprotein IIB/IIIa platelet blockade. Am J Cardiol 2000; 86(10): 780–782. 37. Shrake KL. Comparison of major complication rates associated with four methods of arterial closure. Am J Cardiol 2000; 85: 1024–1025. 38. Dangas G, Mehran R, Kokolis S et al. Vascular complications after percutaneous coronary interventions following hemostasis with manual compression vs. arteriotomy closure devices. Am J Cardiol 2001; 38: 638–641. 39. Carey D, Martin JR, Moore CA, Valentine MC, Nygaard TW. Complications of femoral artery closure devices. Catheteriz Cardiovasc Intervent 2001; 52: 3–7. 40. Von Hoch F, Neumann FJ, Theiss W, Kastrati A, Schomig A. Efficacy and safety of collagen implants for haemostasis of the vascular access site after coronary balloon angioplasty and coronary stent implantation: a randomized study. Eur Heart J 1995; 16: 640–646. 41. Kahn ZM, Kumar M, Hallander G, Shani J, Frankel R. Safety and efficacy of the Perclose suture mediated closure device after diagnostic and interventional catheterizations in a large consecutive population. Cathet Cardiovasc Intervent 2002; 55: 8–13. 42. Resnic FS, Blake GJ, Ohno-Machado L, Selwyn AP et al.. Vascular closure devices and the risk of vascular complications after percutaneous coronary intervention in patients receiving glycoprotein IIb-IIIa inhibitors. Am J Cardiol 2001; 88: 493–496. 43. Chevalier B, Lancelin B, Koning R, Henry M et al.. Effect of a closure device on complication rates in high local risk patients: results of a randomized multicenter trial. Catheteriz Cardiovasc Intervent 2003; 58: 285–289. 44. Smith TP, Cruz CP, Moursi MM, Eidt JF. Infectious complications resulting from use of hemostatic puncture closure devices. Am J Surg 2001; 182: 658–662. 45. Koreny M, Riedmuller E, Nikfardjam M, Siostrzonek P, Mullner M. Arterial puncture closing devices compared with standard manual compression after cardiac catherization – systematic review and meta-analysis. J Am Med Assoc 2004; 291(3): 350–357. 46. Vaitkus PT. A meta-analysis of percutaneous vascular closure devices after diagnostic catheterization and percutaneous coronary intervention. J Invas Cardiol 2004; 16(5): 243–246. 47. Schrader R, Steinbacher S, Burger W, Kadel C et al.. Collagen application for sealing of arterial puncture sites in comparison to pressure dressing: a randomized trial. Cathet Cardiovasc Diagn 1992; 27: 298–302. 48. Gwechenberger M, Katzenschlager R, Heinz G, Gottsauner-Wolf M, Probst P. Use of a collagen plug vs. manual compression for sealing arterial puncture site after cardiac catheterization. Angiology 1997; 48: 121–126. 49. Silber S, Bjorvik A, Muhling H, Rosch A. Usefulness of collagen plugging with VasoSeal after PTCA as compared to manual compression with identical sheath dwell times. Cathet Cardiovasc Diagn 1998; 43: 421–427. 50. Tavris DR, Dey S, Gallauresi B, Brindis RG, Shaw R, Weintraub W, Mitchel K. Risk of local adverse events following cardiac catheterization by hemostasis devices use - Phase II. Journal of Invasive Cardiology 17: 644–650, 2005.

26 ENT devices: cochlear implants James K. Kane and Eric A. Mann US Food and Drug Administration, Rockville, MD, USA

Cochlear implant description Cochlear implants are electronic device systems designed to provide functional hearing to those who have severe to profound hearing loss. The cochlear implant has both external and implanted components (see Figure 26.1). The external components include a speech processor (body-worn or ear-level), headset, and cables. The speech processor receives sound waves through a microphone and converts the acoustic signal into a radio frequency signal, which is transmitted across the skin to the implanted component of the device. The implanted component is placed under the skin in a shallow ‘bed’ made in the bone behind the ear. The receiver/stimulator portion of the implant receives and decodes the radiofrequency signal from the speech processor and converts it to an electrical signal, which is delivered to the stimulating electrode array. The electrode array is inserted into the cochlea through a surgically drilled opening, or cochleostomy. Thus, the implant system converts sound energy into electrical impulses which stimulate the auditory nerve endings within the cochlea. The resultant neural information is transmitted to the auditory cortex and is processed and perceived by the brain as sound.

Intended use The cochlear implant is intended to provide auditory sensation via the electrical stimulation of the auditory nerve. These devices contain many complex, technologically advanced microprocessors, along with software and firmware algorithms. Cochlear implants have been approved for use in postlinguistically deafened adults, as well as Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

396

CH26

ENT DEVICES: COCHLEAR IMPLANTS

Figure 26.1 Model showing cochlear implant, 1999. Reproduced with permission from the Science Museum/Science and Society Library

in prelinguistically and perilinguistically deafened adults and children ( 12 months of age). Cochlear implants are prescription devices that require a team of trained professionals working with implant recipients to produce optimal auditory performance. The team should include ear, nose, and throat surgeons (otolaryngologists), audiologists, speech-language pathologists, and educators skilled in teaching this unique population. Cochlear implants are regulated as Class III medical devices which require an approved Premarket Approval Application (PMA) prior to marketing within the USA. Thus, these devices are subject to the agency’s highest level of regulatory scrutiny and require extensive clinical and nonclinical data to support regulatory approval for their proposed intended use.

Epidemiological investigations involving meningitis associated with cochlear implants Initial reports of meningitis associated with cochlear implants In early June 2002, the Advanced Bionics Corporation, Sylmar, CA, informed US Food and Drug Administration (FDA) that 15 patients implanted with the CLARION1

EPIDEMIOLOGICAL INVESTIGATIONS INVOLVING MENINGITIS

397

cochlear implant between 2000 and 2002 had reportedly developed bacterial meningitis. Five of these reported cases occurred in the USA, in four pediatric patients aged 2–3 years (one child experienced two episodes of meningitis); two of these children died from their infection. The remaining cases from Europe included nine pediatric and two adult patients, with six reported deaths. Of note, 10 of the 11 cases were clustered at two cochlear implant centers in Spain and Germany. Only limited data were available from these early reports to the company, but at least four of the patients had lumbar puncture cultures which identified Streptococcus pneumoniae as the infecting organism. In all cases, meningitis had developed at least 3 months after implantation. This suggested that the infection was not related to inadequate sterilization of the implanted device, because such infections would have occurred soon after implantation. The infections did appear to be associated with CLARION models, which contained a silastic ‘positioner’ component that was wedged next to the implant electrode array within the inner ear to facilitate the transmission of electrical signals to the nerve endings in the cochlea. This feature was unique in the cochlear implant industry to Advanced Bionics devices and raised concerns that this design somehow predisposed patients to meningitis. Also, because the European cases were clustered at two implant Centers, the issue of surgical technique was also a consideration. Advanced Bionics Corporation informed FDA that these initial cases of meningitis were not reported to the Agency because, based on their interpretation of the Medical Device Reporting (MDR) regulation (21 CFR §803), and following consultations with their medical advisory board, they did not believe that the cochlear implant caused or contributed to the occurrence of meningitis. In support of this early determination, cochlear implant devices had never been shown to have a causal relationship with meningitis. However, as the number of reported cases accumulated at an alarming rate, and as concern mounted in the clinical community, the company reconsidered this determination and notified FDA of the cases. As a result of preliminary data from these case reports, and following discussions with FDA and European regulatory authorities, Advanced Bionics elected to recall all unimplanted devices with the positioner component in July 2002. As part of the recall, the manufacturer issued letters to physicians, audiologists, and patients, informing them of the apparent association between meningitis and cochlear implants. These communications stressed the importance of up-to-date vaccinations to reduce the risk of meningitis in implanted patients. However, the risks of surgery to remove the implant were believed to outweigh the risks of meningitis with the implanted device.

FDA Inter-office and medical community collaboration and initial investigation Although meningitis had long been recognized as a theoretical risk with cochlear implant surgery, as with any surgery which opens the inner ear, FDA had never observed any such cases in the hundreds of adult and pediatric subjects enrolled in clinical trials since the inception of this class of devices. Upon receiving the reports from Advanced Bionics, the

398

CH26

ENT DEVICES: COCHLEAR IMPLANTS

Office of Device Evaluation (ODE) immediately sought consultation from the Office of Surveillance and Biometrics (OSB) to investigate postmarket reports of meningitis across all cochlear implant manufactures in the Manufacturer’s and User Facility Device Experience (MAUDE) Database. A search of the entire database yielded only two meningitis reports, both from Advanced Bionics, submitted in the two preceding months. Based on this finding, there was concern that there might be industry-wide under-reporting of this serious adverse event. In order to rapidly assess the extent of meningitis cases in the US cochlear implant population, numerous case ascertainment strategies were employed: A top priority ‘for cause’ inspection of the Advanced Bionics manufacturing facility was scheduled by the Office of Compliance to fully investigate the reported cases, possible contributing factors related to manufacturing and/or design issues, and Medical Device Reporting issues. FDA issued letters to the three approved cochlear implant manufacturers instructing them to submit a report summarizing all known and suspected cases of meningitis that occurred in patients following implantation with their device, regardless of whether or not the company believed they were device-related. This request was made in reference to the Premarket Approval Application (PMA) Postapproval Requirements described in 21 CFR §814.80 and in reference to the Medical Device Reporting requirements described in 21 CFR §803. More specifically, as part of the ‘Conditions of Approval’ for a PMA, manufacturers are advised that they must notify FDA about any adverse event occurring with unexpected severity or frequency within 10 days after the manufacturer receives or has knowledge of the event, if available information reasonably suggests that the device has or may have caused or contributed to a serious injury or death. FDA worked closely with two leading cochlear implant surgeons, Dr Noel Cohen (NYU) and Dr Thomas Balkany (University of Miami), who, out of concern over the recent cluster of meningitis cases, conducted an independent survey of the 401 cochlear implant centers in North America during June and July of 2002. The survey indicated an increase in meningitis cases since 2000 and that the majority of cases were associated with the Advanced Bionics device with the positioner component. The results of the survey were useful to FDA for case ascertainment purposes during the early investigation of this issue. Based on information derived from these three sources, FDA issued a Public Health Web Notification on July 24 2002, describing preliminary reports of at least 25 cases of meningitis worldwide in cochlear implant recipients, including 10 fatalities associated with two of the three FDA-approved cochlear implants [1]. Healthcare providers were encouraged to be vigilant for the signs and symptoms of meningitis in this population, to promptly diagnose and treat otitis media (which preceded the meningitis episode in numerous cases), and to report all cases of known or suspected meningitis to the FDA MedWatch system.

EPIDEMIOLOGICAL INVESTIGATIONS INVOLVING MENINGITIS

399

FDA and CDC collaboration on epidemiologic investigations The epidemiological interpretation of the available data was complicated by several predisposing factors for meningitis in the cochlear implant population. First, certain congenital inner ear malformations, such as the Mondini malformation, are associated with cerebrospinal fluid (CSF) fistulae and recurrent episodes of meningitis, even without cochlear implantation or other otologic surgery [2]. Second, meningitis is the primary etiology of deafness for a significant number of adult and pediatric implant recipients, and it is well established that the risk for contracting meningitis is greater for an individual who has a prior history of this serious infection [3]. Third, there has been a strong trend toward implanting children at an earlier age, based on data showing improved long-term functional outcomes [4]. By the year 2000, one manufacturer had received FDA approval for use in children as young as 12 months. The incidence of meningitis is very high in this very young age group [5]; thus, age was considered as another possible nondevice-related factor relevant to the observed increase in meningitis reports. Finally, the positioner component for the Advanced Bionics Corporation device was introduced into the US market in 1999, and this design modification could have been a predisposing factor causing the sudden increase in meningitis cases. Recognizing the complexity in conducting and interpreting an epidemiological study of this problem, FDA embarked on a collaborative investigation with the Centers for Disease Control and Prevention (CDC), 36 state health departments, and the local health departments of Chicago, New York City, and Washington, DC. Due to the urgent public health nature of the issue, a retrospective cohort study design was used to determine the incidence of bacterial meningitis among a study population of children receiving cochlear implants during the preceding 5.5 years who were less than 6 years old at the time of implant surgery [6]. Using the warranty lists from the three manufacturers, a study population of 4264 children was defined. A variety of meningitis case-ascertainment methods were used for the study. Nineteen meningitis cases were identified within the study population from reports to the company, the MAUDE database, and the surveillance systems of the CDC, state and local health departments. An additional seven cases were identified by a survey that was mailed out to all families on the warranty lists, providing a total of 26 subjects for the study. The incidence of Streptococcus pneumoniae meningitis in this population was determined to be 138.2 cases per 100 000 patient-years, which was more than 30 times the incidence reported in that age group for the general US population (data from the CDC Active Bacterial Core Surveillance Program) [5]. Unfortunately, the study was not designed to determine the incidence of meningitis in an unimplanted cohort of deaf children. Deaf children may have a higher incidence of meningitis, even without a cochlear implant. No data on this incidence are available in the published literature. A nested case-control study was also conducted to assess risk factors for meningitis in this cohort. Twenty-six meningitis cases were compared to a randomly selected control population of 200 children from the cohort who did not develop meningitis. Information on risk factors for the two groups was obtained through parental interviews and from

400

CH26

ENT DEVICES: COCHLEAR IMPLANTS

abstraction of medical records from cochlear implant surgeries, meningitis hospitalizations, and primary care providers. Postimplant meningitis was strongly associated with the use of the positioner component (OR ¼ 4.5, 95% CI ¼ 1.3–17.9) and the incidence of meningitis in patients with the positioner remained elevated for at least 24 months from the time of implantation compared to patients implanted without the positioner. Radiographic evidence of an inner ear malformation associated with a CSF leak was another strong risk factor for the development of meningitis (adjusted OR ¼ 9.3, 95% CI ¼ 1.2–94.5). The ability to assess the protective effects of pneumococcal vaccination in the case-control study was limited because few of the children had received the vaccine. The vaccine was not licensed in the USA until 2000, more than half-way through the study period. Of note, this focused epidemiological study was designed, conducted, analyzed, and published in just over 1 year from FDA’s initial notification of the 11 meningitis cases by Advanced Bionics. This accomplishment reflects the effective, coordinated working relationships among FDA, CDC, state and local health departments, cochlear implant manufacturers and the clinical community.

Recommendations based on investigation findings As a result of these investigations, the cochlear implant community has become highly aware of the association of bacterial meningitis with cochlear implantation. Although the numbers of reported cases remains relatively small when considering a worldwide implant population now greater than 60 000 individuals, FDA has strongly recommended measures to mitigate risk factors identified by these studies, including the following: 1. Specific pneumococcal vaccination recommendations for the cochlear implant population were developed by the CDC during the cohort study and have been posted on both CDC and FDA websites [1]. 2. FDA has worked with manufacturers to update their labeling to reflect the current knowledge of the occurrence of meningitis in the cochlear implant population, as well as measures to reduce the risk of meningitis with their device, e.g. vaccinations, recommendations regarding use of tissue seals around the cochleostomy site. 3. FDA has recommended that implant recipients, their families, and healthcare providers be vigilant regarding early signs and symptoms of bacterial meningitis. 4. As described above, prompt diagnosis and treatment of otitis media are strongly recommended because many of the meningitis cases were preceded by middle ear infections. Finally, the positioner component has been strongly implicated as a risk factor for meningitis, and, while several reasons for the increased risk have been offered [6] (e.g. larger cochleostomy requirement, increased inner ear trauma, inadequate fibrous-tissue

EPIDEMIOLOGICAL INVESTIGATIONS INVOLVING MENINGITIS

401

seal at cochleostomy), the reasons for the actual causal factor(s) remain unknown. Although no longer marketed, there are several thousand adult and pediatric patients currently implanted with positioners, and the long-term risk of the implanted positioner is unknown. Currently, a CDC–FDA collaborative follow-up study on the original pediatric cohort is under way to determine whether the incidence of meningitis remains elevated after 2 years following implantation. Additionally, CDC is reviewing data from the Danish medical registry in an effort to estimate the background incidence of meningitis in the unimplanted deaf population. This estimate may be useful in clarifying the relative contributions of device-related risk factors vs. non-device-related risk factors for meningitis in cochlear implant recipients compared to the general population.

Cochlear implant hermeticity failures During 2003–2004, FDA identified a hermeticity problem with the COMBI 40þ Cochlear Implant System, hereafter referred to as the COMBI 40þ, manufactured by MED-EL Corporation, Innsbruck, Austria. Specifically, the hermetic integrity of the implant case (i.e. its ‘air-tightness’) against intrusion of foreign liquids/gases was failing over time. Consequently, device reliability was being compromised by moisture intrusion, which affected the functionality of the electronic circuitry. The COMBI 40þ has been manufactured since April 1997, with the first implantation in the USA occurring in November 1997, during the clinical trial for the implant system. The company received FDA premarket approval for commercial distribution of the COMBI 40þ in August 2001. Currently, the COMBI 40þ is sold in over 40 countries worldwide and is indicated for use in infants ( 12 months of age), children and adults. Historically, there are two MED-EL cochlear implant device ‘types’ based on manufacturing technology, thick-film and thin-film. These technologies refer to the firm’s proprietary manufacturing processes used to solder the metal feedthrough terminals for the electrodes to the implant’s ceramic case. Devices with thick-film technology represent an older manufacturing technology discontinued in 2002, although thick-film devices continued to be implanted into 2003. The Agency’s awareness of the hermeticity problem associated with the thick-film COMBI 40þ devices developed independently from separate activities by two offices within the Center for Devices and Radiological Health: the Office of Regulatory Affairs (ORA), Division of Field Investigations Branch, and the Office of Device Evaluation (ODE), Division of Ophthalmic and Ear, Nose, and Throat Devices. The activities of the ORA will be described first, followed by the activities of the ODE.

Good manufacturing practices issues identified by ORA inspection (March–April 2004) Subsequent to premarket approval (PMA) for a medical device, manufacturers are routinely inspected, ideally every 2 years, to ensure adherence to Current Good Manufacturing Practice (CGMP) requirements of the Quality System (QS) regulation

402

CH26

ENT DEVICES: COCHLEAR IMPLANTS

described in 21 CFR §820. Such inspections are also scheduled when significant manufacturing changes are reported to the Agency. These inspections are conducted under the auspices of the ORA, Division of Field Investigations branch. Data reviewed during inspections for postapproval manufacturing changes for the Combi 40þ indicated that the majority of device failures for explanted devices resulted from a loss of hermeticity; there was moisture contamination, with corrosion of the electronic printed circuit board, which resulted in the devices being out of electronic specification.

Review of Annual Report by ODE for CY2003 MED-EL’s annual report to the Office of Device Evaluation (ODE) for the reporting period August 2002–August 2003 noted that 25 medical device reports were filed with the FDA. Of the 25 devices reported as having failed, two were still implanted. The remaining 23 explanted devices had been sent to Innsbruck, Austria, for evaluation. Notably, device failure analyses determined that 12 of the 23 explanted devices (52%) failed due to loss of hermeticity, resulting in moisture intrusion. The Division of Ophthalmic, and Ear, Nose, and Throat Devices sent a letter to MED-EL requesting additional information regarding the potential causal factors related to these device failures and to provide insight into the categorization of failures over time. Further, they were asked to describe the reported clinical complaints and their relationship, if any, to device functionality prior to complete device failure (e.g. unusual sound percepts, such as roaring tinnitus, excessive loudness and/or pain when device was turned on, intermittent function, and decreased performance. Such perceptual/behavioral reports frequently are suggestive of device malfunction prior to complete device failure [7] and commonly are referred to as ‘soft’ failures [8,9]. These clinical symptoms often occur even though an implant may pass the manufacturer’s in situ device integrity tests. In contrast, when a device stops working completely (indicated by a failure of the external speech processor to ‘link’ with the implant receiver), such failures are called ‘hard’ failures [8].

Interoffice coordination to evaluate the hazard A meeting between the Office of Device Evaluation, the Office of Compliance (OC), and the Office of Science and Engineering Laboratories (OSEL) was held in April 2004, to discuss the hermeticity problem associated with MED-EL’s cochlear implant. Discussion items included: 1. Whether or not hermeticy was indicated as the cause for device failure. 2. Determining the number of pediatric and adult patients implanted with thick-film devices having the potential to fail from loss of hermeticity. 3. Any known technological problem(s) that would affect the loss of hermeticity.

EPIDEMIOLOGICAL INVESTIGATIONS INVOLVING MENINGITIS

403

It was not possible to determine the number of patients potentially at risk for device failure based on the review of the annual report or from the numbers obtained during the inspection. Part of the discrepancy in the numbers resulted from inclusion of various device designs not implanted in the USA in the reported counts, as well as the different time periods for the various counts. A Health-Risk Assessment (HRA) was done by FDA to determine the seriousness of the hermeticity problem as it related to MED-EL’s implanted patient population. Of particular concern was the infant/child population implanted with this device. Because infants and children, unlike adults, may not be able to complain when experiencing pain or overstimulation due to device malfunction, their suffering potentially could go unnoticed and untreated and any unusual behaviors might be ignored or attributed to some other factor. Further, such device malfunctions may cause decreased performance over time or preclude improvement in speech perception abilities, which would be less likely detected in the younger patient population. An additional concern was the possibility of permanent neural damage resulting from leakage of DC current, which was noted in MED-EL’s hazard analysis for their implant system. Because corrosion from the hermeticity problem could potentially short-circuit the built-in safety features against DC leakage current (e.g. blocking capacitors), permanent neural damage was a theoretical possibility. After overall consideration of the likelihood of occurrence of a potentially hazardous event, the probability of injury occurring as a result of the event, and the severity of injury that might reasonably be expected to occur, the health risk was determined to be ‘high’.

Further investigation Based on the health-hazard risk rating and review of MDR reports for the Combi 40þ device, the Office of Compliance scheduled a complete ‘for cause’ inspection of MED-EL Corporation in June 2004. Significant deviations from the QS regulation and the medical device report reporting requirements were noted and discussed with MED EL management, who agreed to correct all deficiencies. A review of the manufacturer’s Complaint Report Database identified a subset of records that reported device failure and explantation due to a confirmed loss of hermeticity in devices manufactured with thick-film technology. This latter group was reviewed to identify the nature of the complaints from the implant centers regarding device functionality, and perceptual reports from patients prior to device failure. The most frequently identified functional problem was cited as ‘poor coupling’, indicating that the telemetry link from the speech processor to the implanted receiver was failing. In contrast, perceptual/performance complaints did not demonstrate a clear pattern. Reports of ‘pain’, variations of ‘loud’ sounds, ‘noise’, ‘intermittent’ function, and ‘no hearing’ occurred most frequently; these descriptors occurred in isolation and in combination. Notably, none of the reported perceptual/performance complaints were unique to the MED-EL device; other cochlear implant manufacturers’ medical device reports contain similar perceptual reports for their device failures as well. Also, duration of the

404

CH26

ENT DEVICES: COCHLEAR IMPLANTS

implant prior to failure did not have a noticeable effect on the type of percepts reported. Although associations between specific component malfunctions and reported percepts/ sensations/performance changes could not be quantified, reports of ‘pain’ or variants of ‘loud’ sounds were associated, in some patients, with the occurrence of DC leakage currents, as determined by bench testing of explanted devices.

MED-EL proposal to assess the relationship between device failure and patient harm During the inspection, to provide objective data confirming that no harm occurred as a result of device failure (e.g. leakage of DC current), MED-EL proposed reviewing changes in programming parameters (e.g. loudness levels) between the new and old (explanted) devices in patients who had been reimplanted with a second MED-EL device. However, such comparative data can not be used to confirm an absence of neural tissue damage. That is, MED-EL’s reference for ‘no harm’ was the operational parameters of the device just prior to explantation but, because it is possible that damage occurred over time since the initial date of device activation, the latest parameters may simply represent those necessary to compensate for loss of functionality as a result of neural damage. Stated differently, if neural damage did occur over time, there is no reason to assume that device parameters prior to explant should be any different from those subsequent to reimplantation, because the reimplanted device would stimulate the same damaged auditory system as did the explanted device, all other factors (e.g. surgical placement of electrodes) being equal. Furthermore, changes in operational parameters over time, such as increased stimulation levels, may simply result from the patient adapting to electrical stimulation over time and preferring to have increased loudness levels. Consequently, the FDA and MED-EL jointly concluded that there was no valid way at present to separate the two possibilities, neural damage vs. normal adaptation, to account for observed parameter changes.

Recommendations and interventions At the conclusion of the inspection, a FDA-483 (Inspectional Observation Form) was presented and discussed with MED-EL management that detailed the inspectional findings requiring correction. This review and discussion prompted the following voluntary actions by MED-EL: Issue a ‘cease distribution’ memo for thick-film devices to all distributors of the device. Provide FDA with a listing of the current status of all thick-film devices (both implanted and unimplanted). Render unusable all unimplanted thick-film devices. Issue a ‘Product Safety Alert’ letter to send to all user facilities.

REFERENCES

405

Conclusions The above examples show that even ‘state-of-the-art’ medical devices may have unforeseen failure modes resulting from design-related problems that compromise both safety and effectiveness. Importantly, such problems may take years to manifest themselves as clinical or non-clinical problems. Likewise, these examples show that vigilant monitoring of adverse events by the FDA, in conjunction with an ability to quickly recognize, assess, and respond to unexpected device problems, are necessary to help ensure that medical devices continue to meet their intended use while minimizing patient risk. In addition, the examples underscore the need for manufacturers to be alert to potential device problems, to report potential problems quickly and accurately, and to err on the side of caution in their interpretation of potential problems.

References 1. FDA public health web notification: risk of bacterial meningitis in children with cochlear implants: http://www.fda.gov/cdrh/safety/cochlear.html [accessed April 6 2005]. 2. Kline MW. Review of recurrent bacterial meningitis. Pediatr Infect Dis J 1989; 8: 630–634. 3. King MD, Whitney CG, Parekh F, Farley MM. Recurrent invasive pneumococcal disease: a population-based assessment. Clin Infect Dis 2003; 37: 1029–1036. 4. Svirsky MA, Tech SW, Neuburger H. Development of language and speech perception in congenitally, profoundly deaf children as a function of age at cochlear implantation. Audiol Neurootol 2004; 9: 224–233. 5. Schuchat A, Robinson K, Wenger JD, Harrison LH et al. Bacterial meningitis in the United States in 1995. N Engl J Med 1997; 337: 970–976. 6. Reefhuis J, Honein MA, Whitney CG, Chamany S et al. Risk of bacterial meningitis in children with cochlear implants. N Engl J Med 2003; 348: 435–445. 7. Parisier SC, Chute PM, Popp AL. Cochlear implant mechanical failures. Am J Otol 1996; 17: 730– 734. 8. von Wallenberg EL, Brinch JM. Cochlear implant reliability. Ann Otol Rhinol Laryngol 1995; 166 (suppl): 441–443. 9. Buchman CA, Higgins CA, Cullen R, Pillsbury HC. Revision cochlear implant surgery in adult patients with suspected device malfunction. OtolNeurootol 2004; 25: 504–510.

27 Silicone gel-filled breast implants: surveillance and epidemiology S. Lori Brown US Food and Drug Administration, Bothell, WA, USA

Joan Ferlo Todd US Food and Drug Administration, Rockville, MD, USA

Breast implants have been a continuing issue for the regulatory agencies around the world for over two decades. In this chapter, we will discuss types of breast implants, a brief regulatory history of breast implants, US Food and Drug Administration (FDA) studies using passive surveillance (MedWatch/Medical Device Reports), FDA studies on breast implant rupture, formal reviews of epidemiologic studies on breast implants and connective tissue disease, and the emerging issue of suicide and breast implants.

Breast implant types There are two types of breast implants currently available in the USA, silicone gel-filled breast implants and inflatable saline-filled implants. Both types of implants have a silicone elastomer shell (envelope) which is filled with either silicone gel (see e.g. Figure 27.1) or sterile isotonic saline. Some implant models have both a silicone gel-filled lumen and a saline-filled lumen. These bi-lumen implants are classified as silicone gel implants because of their silicone gel component.

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

408

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

Figure 27.1 Silicone gel-filled breast implant. Courtesy of the FDA History Office

Another type of implant which was recently on the market in the European Union (EU), and under study but not approved in the USA, was the Trilucent implant. This implant was a silicone shell filled with processed soy bean oil. This device was removed from the market because of adverse events associated with the device, including breakdown products from aging soybean oil [1], and concerns that peroxidated products could be genotoxic.

A BRIEF REGULATORY HISTORY OF BREAST IMPLANTS

409

Nomenclature for breast implants may be confusing, since in many publications ‘silicone implants’ mean silicone gel-filled implants, and in other publications this term refers to either silicone gel-filled or saline-filled implants (which have a silicone elastomer shell). For the purposes of clarity in this chapter, breast implants will mean all types of breast implants, and silicone gel-filled breast implants will be specified as such.

A brief regulatory history of breast implants Silicone gel-filled breast implants have been on the market in the USA since 1963. In 1976, the Medical Device Amendments were enacted giving the FDA regulatory authority over medical devices, including silicone gel-filled breast implants [2]. At that time, FDA planned to review medical devices based on their regulatory class (see Chapter 2), with a strategy of first reviewing devices with new technology or devices for which there was insufficient information to assure safety and efficacy. Based on the recommendation of the General and Plastic Surgery Devices Advisory Panel in 1978, breast implants were classified as class II devices: those for which general controls alone are insufficient to provide reasonable assurance of safety and effectiveness, but for which there is sufficient information to establish special controls. Advisory committees (panels) are groups of experts commissioned by the FDA to provide advice on questions posed by the agency [3]. In 1988, FDA reclassified all breast implants (silicone gel-filled and saline-filled – see below) into class III: devices for which there may be insufficient information to determine that the application of general controls alone, or general and special controls, are sufficient to provide reasonable assurance of safety and effectiveness (see Chapter 2). After this reclassification, in 1989 FDA announced its intention to require breast implant manufacturers to provide a premarket application (PMA) and produce studies to support the safety and effectiveness of breast implants. As required by law, FDA gave the manufacturers a minimum of 36 months to comply. In the meantime, concerns were raised about the safety of breast implants. In 1989 an unpublished study showed that polyurethane foam, which was used as a coating on some silicone gel-filled breast implants to reduce capsular contracture, would degrade and release 2-toluene diamine (TDA), a chemical known to cause cancer in animals, under conditions of high temperature and alkalinity. The manufacturer removed these from the market in 1991. Based on a study by one of the manufacturers, FDA estimated that the risk of cancer from exposure to TDA is about one in 1 million over a woman’s lifetime [4,5]. By the early 1990s there was growing concern about breast implants because of the potential for implants to break or leak silicone gel or shed silicone particles from the implant shell [6]. It was not known how often implants ruptured or how often gel or silicone particles migrated out of the implant shell into breast tissue or beyond, neither was it known whether there were health consequences to implant rupture. There were also concerns about increased risk for connective tissue disease and cancer in women with implants. Breast implant manufacturers sent information to the FDA to fulfill the requirement for premarket approval. In August 1991, the FDA determined that the information breast

410

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

implant manufacturers had sent was inadequate to warrant a review. However, the issue had become controversial, and FDA sought outside advice from advisory committees (panels). Therefore, the FDA convened the General and Plastic Surgery Devices Panel in November, 1991, to consider whether there was adequate information on silicone gelfilled breast implants to permit continued marketing. Despite the Panel’s opinion (which concurred with that of the FDA) that there was inadequate information to approve these devices, the Panel voted to keep silicone gel-filled breast implants on the market while manufacturers collected additional information to support the continued marketing of their product [2]. Several months later, in early 1992, the FDA called for a voluntary moratorium (not a ban, as sometimes characterized) on the sale and implantation of silicone gel-filled breast implants until the Panel could meet again to consider additional information. This moratorium did not affect saline-filled breast implants. The additional information FDA wanted the Panel to consider included evidence that breast implants leaked or ruptured at an unknown rate, and some studies which had been published in the 1970s and early 1980s on a possible link between breast implants and connective tissue disease, which had not been included in the manufacturers’ earlier submissions to the FDA. There was also growing concern that breast implants might cause an atypical connective tissue disease characterized by muscle and joint pain and fatigue. In addition, there was evidence that the manufacturing process to make breast implants was poorly controlled. The Panel reconvened in February 1992 and in March 1992 the FDA announced its decision. FDA lifted the voluntary moratorium on silicone gel-filled breast implants and announced its decision to allow access to silicone gel-filled breast implants only in controlled clinical studies for reconstruction after mastectomy, correction of congenital deformities, or replacement of ruptured silicone gel-filled implants. Until these clinical studies (called adjunct studies) could be submitted by manufacturers and reviewed by FDA, the Agency authorized temporary limited distribution of silicone gel-filled implants for patients undergoing reconstruction on an urgent need basis, with an informed consent, based on the finding that there was a ‘public health need’ for implants for reconstruction but not for augmentation purposes. While this made silicone gel-filled breast implants unavailable for cosmetic breast augmentation, the FDA allowed that the companies could later conduct clinical trials that would include a limited number of augmentation patients (core studies) as well as reconstruction and revision patients. Some physicians’ professional societies questioned the FDA decision [7,8] and the FDA Commissioner defended the Agency’s decision [9,10]. At that time, there were no epidemiologic studies to either refute or confirm that women with breast implants were at greater risk for debilitating connective tissue disease than women without breast implants, despite the fact that these devices had been marketed for nearly 30 years. In addition, in 1992, while it was known that silicone gel-filled breast implants ruptured and leaked, there was little information on how often this occurred or its consequences. In fact, there was not even the most basic descriptive epidemiologic information on the prevalence of breast implants in the US population, or firm estimates of the proportion used for reconstruction after cancer or for other medical purposes compared to implantation for breast enlargement (augmentation) in healthy women for cosmetic reasons.

FDA SURVEILLANCE STUDIES ON BREAST IMPLANTS

411

Subsequent studies indicated that 0.5–2-million US women had breast implants and estimates indicated that the majority (60–80%) were for cosmetic augmentation, not reconstruction after breast cancer or trauma [11–13].

FDA surveillance studies on breast implants While there was a dearth of epidemiologic studies on breast implants, there was no lack of reports to the FDA’s adverse event reporting systems on breast implants. The FDA receives reports of adverse events on all medical devices through its national reporting program. These adverse events are received from manufacturers, importers, user facilities, and voluntarily from concerned individuals. In order to review all reports on breast implants received by the FDA, three adverse event databases maintained by the FDA were searched. The three databases are: Device Experience Network (DEN). Manufacturer and User Device Experience (MAUDE). Alternative Summary Reporting (ASR). The DEN database contains FDA’s earliest adverse event reports on silicone breast implants. These reports were received under the mandatory Medical Device Reporting (MDR) program during 1984–1997. The reports reviewed were mandatory manufacturer reports on breast implants that may have malfunctioned or caused or contributed to a death or serious injury. Unlike MAUDE (see below), DEN does not contain device or patient problem codes. These codes are from a dictionary of possible device or patient problems [14]. DEN uses causative factor codes instead of manufacturer evaluation and conclusion codes. DEN now serves as an archival database of reports for January 1 1984– December 31 1997. The MAUDE database contains reports of medical device adverse events submitted by user facilities since 1992, consumers, importers, and distributors since 1993, and manufacturers since 1996. Distributor reporting was discontinued in 1997. MAUDE has device and patient problem codes, and manufacturer evaluation and conclusion codes. MAUDE is the predominate database used by FDA for evaluation of individual device-related adverse event reports. The Alternative Summary Reporting (ASR) program was instituted by FDA in 1995 as a means to receive well known and documented adverse events associated with medical devices. ASR allows manufacturers to report these types of events in a concise and condensed report. Under ASR, manufacturers submit a report of injuries and/or malfunctions (depending upon device) to FDA once every 3 months in summarized form, instead of submitting individual reports for each adverse event. In October 1999, new requirements for the ASR program started. With these new requirements, firms now provide patient and device codes as well as evaluation and conclusion codes for each

412

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

Table 27.1 Reports to the FDA related to all types of breast implants received, January 1 1984–December 31 2005 FDA database report source DEN1 MAUDE2 ASR3 Total

Implant type Silicone gel-filled 96 954 13 999 34 794 145 747

Saline-filled 22 9 153 184

698 020 070 788

All types 119 23 187 330

652 019 864 535

1

DEN, Device Experience Network: from a database that was used in 1984–1997. MAUDE, Manufacturer and User Facility Device Experience: reports received January 1 1992– December 31 2005. 3 Alternative Summary Reporting (ASR): these are reports allowed from some manufacturers when large numbers of well-known adverse events or malfunctions are reported. Instead of reporting each eventindividually,theymay senda summaryofreportsfor thatparticular adverse event ormalfunction. 2

adverse event. Summary reports do not contain narrative text. Instead, each event (or line item as it is called in ASR) is line-listed and summarized by codes. By the end of 2005, the FDA had received 330 535 reports on breast implants from the DEN, MAUDE, and ASR databases combined (Table 27.1). Over the years, the FDA has performed a number of analyses based on adverse event reporting to characterize the reported problems with breast implants. These studies are described below.

Silicone gel-filled breast implant reports 1984–1995 By the end of 1995, the FDA had received 98 405 adverse event reports on silicone gel breast implants (this does not include reports for saline breast implants) [15]. Many of these were reported to FDA after the 1992 Panel meeting described above (Figure 27.2). It is not unusual for media coverage to increase awareness and reporting of adverse event reports. Between 1984 and 1991, there were a total of 3479 reports for silicone gel-filled breast implants. In 1992 alone, there were 27 130 reports entered into the database; this rose to 33 982 reports entered in 1994. Thus, there was a > 2000% increase in manufacturers’ reports on silicone gel-filled breast implants from 1991 to 1992. The large increase in reports was attributed to multiple reasons, including: reporting on individual breasts (e.g. if women had two ruptured implants, each rupture was reported as a separate report); multiple reports of the same adverse event when multiple sources reported the same adverse event (e.g. if the same event was reported to multiple manufacturers because the reporter was uncertain of which brand had been implanted); press coverage of breast implants; and growing litigation over breast implants (i.e. reports of alleged injuries). Because of the large number of reports of well-known problems, some manufacturers of implants were allowed to submit adverse event reports as quarterly summaries (ASR) rather than individual reports, starting in 1995 [16].

413

FDA SURVEILLANCE STUDIES ON BREAST IMPLANTS 35000 30000 25000 Adverse 20000 Event Reports 15000 10000 5000 0 1984

1986

1988

1990

1992

1994

Year

Figure 27.2 Adverse event reports for silicone gel-filled breast implants by year, received 1984–1995. Because of a backlog on entering reports, not all reports from 1995 were shown in this report. Additionally, in 1995, manufacturers started reporting through alternative summary reports and these were not available at that time

Because various terms were used to describe implant-related adverse events, the FDA combined terms that appeared to be related to rupture. Silicone gel-filled breast implant rupture or leaking was reported in about 26 661 reports (Table 27.2), accounting for 27% of reports overall. FDA considered this to be the most frequently reported adverse event. The second most commonly reported problem was ‘reaction’, accounting for 26 106 (26%) reports. The term ‘reaction’ was used to code the reported adverse event when the description of the event(s) was non-specific. Reaction was defined as ‘adverse effect, irritation, or swelling’. Adverse event reports for medical devices are coded in two different ways, the problem with the device (e.g. implant rupture) and the problem or consequence to the patient Table 27.2 Implant reports for ruptured or disrupted implants from DEN and MAUDE databases, 1984–1995 [15] Coded problem report

(n)

Burst Leak Deflate Disintegrate Collapse Tear, rip, or hole in device Breaks Crack

16 237 7 030 1 436 791 610 522 27 8

Total

26 661

414

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

(e.g. additional surgery to remove a ruptured implant). Because historically FDA focused on ‘device problems’, when illnesses in women were reported to FDA, the list of available codes was inadequate to allow an unambiguous coding of the adverse event. In addition, many of the reports were complex and involved multiple adverse events that had occurred over a period of years, or even decades. Because of this, the term ‘reaction’ was used by manufacturers to code a wide variety of adverse events, making it very difficult to identify the real problems. New terms were added to the patient problem coding manual, for instance ‘capsular contracture’, because of the many reports of capsular contracture experienced by women with breast implants. The difficulty in summarizing these reports without actually reading each one makes a poignant argument for adopting and using a standardized system of nomenclature to code adverse events along with detailed training for standardizing their use. Many of these reports were also coded as ‘nonspecific’ (23 542 reports, 24%) and had no information on the adverse event other than to say that some unspecified problem with silicone gel-filled breast implants was reported.

Infections related to breast implants Over an 8 month period in 1996, the FDA received two reports of deaths purportedly due to toxic shock syndrome in women after augmentation mammoplasty. There is some risk of infection associated with any surgery, but deaths due to cosmetic surgery are of particular concern because the surgery is elective and usually performed on an otherwise healthy individual. FDA undertook a study to characterize the infection reports related to breast implants. The FDA identified 1971 reports of infections related to breast implants reported during 1977–1997 [17]. The study reports included silicone gel-filled breast implants (62%), saline-filled breast implants (32%), and tissue expanders (6%), which are temporary implants to expand the tissue in preparation for a permanent breast implant. Many of the reports merely stated that there had been an infection without any supporting details (45.5%). The infecting organism(s) was reported in 8% of reports and included Staphylococcus sp., Pseudomonas sp., Bacillus sp., Clostridium sp., Serratia sp., Mycobacteria sp., Scedosporium sp. and Aspergillus sp. FDA also noted a wide range in time from mammoplasty to the reported infection. It was of interest that over half of the infections were reported to occur over 26 weeks after the implantation. This is consistent with the possibility for biofilm formation and may be a risk factor for capsular contracture [18,19]. Biofilms are a surface film of microorganisms which adhere to solid surfaces such as implants. A recent study suggested that infection rates related to initially inserted breast implants are estimated at 2%, with an average cost of $20 000 for medical and surgical treatment [20].

Breast implants and mammography Another issue that was investigated with surveillance data from the FDA’s adverse event reporting system was breast implant issues during mammography. Women with breast

FDA SURVEILLANCE STUDIES ON BREAST IMPLANTS

415

implants are at the same risk for breast cancer as other women [21], and are urged to undergo mammography. In the MAUDE database, the FDA identified 66 reported adverse events that described issues with mammography related to breast implants [22]. The majority of these reports, 41 of 66 (62.1%), described breast implant rupture suspected to have occurred during mammography. Other adverse events reported included crushing implants during mammographic compression, pain during mammography attributed to implants, inability to perform mammography because of capsular contracture or fear of implant rupture, and delayed detection of cancer attributed to implants. It is well known that 22–83% of mammographically visualizable breast tissue may be obscured by breast implants, because radio-dense silicone implants obscure glandular tissue [23], and special positioning of augmented breasts is required to maximize imaging [24]. Because FDA’s study was based on surveillance data, it was not possible to determine the prevalence of problems for women with breast implants during mammography. This information would be important for informing all women of possible risks with breast implants, and especially those at higher risk for cancer because of family history or other risk factors, of the added difficulty of performing mammography with breast implants. Importantly, young women considering breast implants may not have experienced a mammogram and may not consider their future need for mammography when they are deciding to obtain breast implants.

Effects of mother’s implants on offspring Another issue which has been of continuing concern to women with implants has been the possibility that a woman’s implants affect her offspring. Some suggested mechanisms by which this might occur are via placental transmission of siloxanes (the building blocks of silicone), silicone, or other chemicals from breast implants [25]. Another possible source is from breast milk if contaminated with these compounds. Potential maternal–child health issues related to breast implants fall into three main categories: (a) female reproduction, including infertility and spontaneous abortion; (b) potential teratogenic effects of silicone and other chemicals in implants; and (c) lactation quality (purity and safety of milk) and quantity (ability of women with implants to nurse). There is very little published information on any of these issues. FDA searched MAUDE and DEN for reports of maternal–child problems attributed to a woman’s breast implants [26] and identified 339 relevant reports. Nearly half of these reports (46%) described actual problems with breastfeeding or expressed concern that implants would be unsafe or interfere with breastfeeding; 44% of reports (n ¼ 149) described either nonspecific or specific signs, symptoms, or illnesses in children thought to be related to their mother’s implants. For the most part, these reports were vague and did not specify an illness or signs and symptoms, but simply stated that a child was ill due to his/her mother’s implants (106 of 149 reports). In those reports in which illness, signs, or symptoms were specified, the majority reported gastrointestinal problems (24 of 43 reports in which it was specified) or

416

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

connective tissue disease (16 of 43 reports), or symptoms or signs of connective tissue disease (13 of 43 reports). An additional 3.5% of all reports (n ¼ 12) described a congenital anomaly believed by the reporter to be due to breast implants. Without an extensive prenatal history on the mother, it is difficult to evaluate these reports. In a study of health outcomes in a cohort of offspring of women with breast implants, there was no observed increase in overall congenital anomalies in children born to mothers after breast implantation [27]. However, there was a statistically significant increase in children born with digestive organ malformation (RR ¼ 1.9, 1.4–2.4) or pyloric stenosis (RR ¼ 3.0, 1.9–4.7) to mother’s after (but not before) breast implantation. There was also a slight increase in esophageal disorders in children of women with implants (RR ¼ 1.6, 1.1–2.3) but this increase was also observed in offspring born prior to mother’s breast implantation (2.1, 1.5–2.8). Although there are not compelling studies to indicate that offspring of women with implants might be adversely affected, there is inadequate study to rule out that possibility. Because effects could be subtle and occur over the lifetime of an offspring, only large, well-designed longitudinal studies could resolve this issue.

Breast implant rupture Implant rupture was mentioned as an issue with implants in 1992 [9] and emerged as the most frequently reported problem in adverse event reports to the FDA (see above). Manufacturers opined that the rupture rate for breast implants was 0.2–1.1%, but the American Medical Association taskforce suggested that it could be as high as 4–6% [7]. Manufacturers made the argument, in meetings with the FDA, that when the number of ruptured implants reported to them was used as the numerator and the number of implants ever sold was used as the denominator, implant rupture could be seen to be rare. Their analysis was problematic, since it is well established that adverse events are underreported at an unknown but significant rate. Neither did this consider that the number of implants sold was not equal to the number of women exposed (or even breasts exposed), since practices with implants included removing and replacing ruptured implants; stacking multiple implants in a surgical pocket; discarding implants that ruptured during surgery or out of the box; and stocking implants, sometimes in large numbers. In order to take these possibilities into account, FDA undertook its own assessment of breast implant rupture, as discussed below. First, a little background on implant rupture: silicone gel-filled breast implants may rupture without causing any changes, or causing minimal changes, to the implanted breast. This is because the gel may be retained in the implant capsule (scar) that forms around the implant. In other cases, silicone gel-filled breast implant rupture will be recognized because it causes cosmetic changes in the breast, such as changes in the breast shape or the formation of granulomatous tissue (lumps) in the breast that may be visible or detected by palpation. It is not clear whether there is a natural progression from a rupture contained by the fibrous scar (capsule) that forms around the implant to an

FDA SURVEILLANCE STUDIES ON BREAST IMPLANTS

417

extracapsular rupture, which is when silicone extrudes outside of the capsule, because little is known of the natural history of rupture. The phenomenon of ‘silent rupture’ was first described in a series of 350 women with breast implants undergoing screening mammography [28]. Implant ‘leak’ was observed in 16 (5%) of the women in this series. Subsequent studies have indicated that mammography is the least sensitive imaging method for detecting implant rupture (and, as described above, may also contribute to implant rupture) and that magnetic resonance imaging (MRI) in the most sensitive and specific method [29–31]. In contrast to the potential for a silent rupture with silicone gel-filled breast implants, saline breast implants typically quickly deflate when ruptured, with a noticeable change in breast size. In some cases, the deflation may occur over an extended period of time through a pin-hole sized defect, over days to weeks, but it is nonetheless recognized as ruptured as the implant deflates. Several studies reported rates of silicone gel-filled breast implant rupture confirmed by explantation [32–37]. These series included examination of implants explanted for a variety of reasons, including suspected rupture. The definition of rupture was inconsistent between these reports because there were no agreed standards for defining implant rupture. Further complicating the vague definition of implant rupture was the fact that intact implants may ‘bleed’ silicone oil. Despite the lack of a definitive definition of rupture, prevalence rates of other than intact implants were found to be consistently high, 23–63% of explanted implants in these series. It was argued that these studies were not representative of implants in general because results came from a population of women who were having problems with their implants – hence their explantation. The range of breast implant rupture estimates (from 0.2% by industry to up to 63% in some studies of explanted implants) was so wide that it was not possible to even give a reasonable range for silicone gelfilled breast implant rupture rate. Rupture is an important issue for several reasons. First, ruptured implants do not perform as intended. Second, ruptured gel implants may release silicone outside of the shell. Silicone gel may migrate within the breast or outside the breast to the axilla [38–40] or as far as the hands [41]. There have been unusual presentations of migrated gel, such as an abdominal mass [42] and an aggressively expanding silicone granuloma [43]. There are also reports of silicone found in distant organs, such as liver [44]. Third, it is not known whether there are health effects of having silicone gel loose in the breast (including that maintained inside the scar capsule) or whether there may be adverse health effects related to having gel and chemical components of gel migrate to locations outside the breast. A recent study of tissue samples from three women with (explanted) silicone gel-filled breast implants indicated that there were small but measurable amounts of siloxanes in the tissue from the implant capsule (scar) and surrounding fat and muscle, but this was not detected in control women without silicone gel-filled breast implants [45]. There is also concern that, as was seen with TrilucentTM implants, there may be break-down products or changes in the chemical composition of aging silicone [46,47].

418

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

Because of these concerns and the lack of a definitive estimates of either for the incidence rate or prevalence of silicone gel-filled breast implant rupture, FDA worked with an epidemiologist at the US National Cancer Institute (NCI), on a cohort from its ongoing study on breast implants and cancer risks, to examine breast implant rupture. The FDA contracted for an MRI study at two of the sites in this NCI study on women who still had their silicone gel-filled breast implants. We selected two plastic surgery study sites in Birmingham, Alabama, because of the high response rate there to a previous patient questionnaire. Women who still resided in the Birmingham area were called and asked to respond to a questionnaire on their current health status and surgeries related to their breast implants. Women from both Birmingham sites who still had silicone gel-filled breast implants were invited to undergo an MRI examination at the Kirklin Clinic at the University of Alabama at Birmingham, to determine the status of their breast implants with respect to rupture. Three experienced radiologists independently examined and rated all MR images for signs of implant rupture and extracapsular silicone. A consensus reading, based on the majority opinion of these radiologists, was developed for each implant as to whether it was ruptured, indeterminate, or had no evidence of of rupture (intact). There were 344 women with silicone gel-filled breast implants in this study who underwent an MRI evaluation [48]. The average age of women in the study was 51.4 8.4 years and the average age of their implants was 16.5 3.4 years at the time of the evaluation. The findings indicated that of the 687 implants in the study, 378 (55%) were rated as ruptured. Another 50 implants (7.2%) were rated as indeterminate, indicating that there was uncertainty as to their status but reason to be suspicious of rupture. Overall, 265 (77%) women had at least one implant that was rated as ruptured or indeterminate. The radiologists’ agreement on implant status, measured with a weighted kappa statistic for agreement, indicated that agreement among these experienced radiologists was almost perfect. Risk factors for implant rupture included implant age, implant location (with submuscular implants being more likely to rupture than subglandular placement), and implant manufacturer. Radiologists also evaluated whether there was silicone outside of the scar capsule (extracapsular silicone). Extracapsular silicone was observed around 85 of the 687 implants, affecting 73 (21%) of women in the study. The radiologists’ agreement on this was not as high as it had been for implant rupture status [49]. This study was unique in that it was the first study in which an unselected and presumably unbiased population of women with silicone breast implants underwent an MRI examination. This contrasted with other studies, in which implants were evaluated in women who had problems or complaints about their implants. The study had shortcomings, including the fact that it was limited to two study sites in the same state and one particular brand of implants predominated. A similar subsequent study, funded by implant manufacturers, examined 271 women with 533 silicone gel-filled breast implants [50]. In their study, 26% of implants in 36% of women examined were ruptured, and an additional 6% of implants were probably ruptured. In this study, in contrast to the FDA study, there were three facilities at which MRI examinations were performed, and equipment in one of the facilities, which

FDA SURVEILLANCE STUDIES ON BREAST IMPLANTS

419

examined 116 (43%) of the 271 women, was subsequently found to not be working properly, so that silicone-excited sequences could not be generated. Some, but not all, of these cases were subsequently imaged at another site, with a change in diagnosis from no rupture to possible rupture or from possible rupture to ruptured for 9 of 28 evaluations. This indicates that their finding of 36% of women with ruptured implants likely represents a minimum estimate, since more ruptures might have been diagnosed with properly functioning equipment. In this study, 22% of ruptured implants had signs of extracapsular silicone. In the decade after 1992, the prevailing thought on silicone gel-filled breast implant rupture had moved from believing that rupture was rare (0.2–1.1%) to recognizing that rupture was very common, particularly as implants age. A meta-analysis of 35 studies reporting on rupture status of over 8000 breast implants reported that implant failure was 30% at 5 years, 50% at 10 years, and 70% at 17 years [51]. There is still no consensus on how the rupture of silicone breast implants should be managed or even whether intermittent screening for rupture after implantation should be recommended. Clinical (physical) examination for implant rupture is inadequate, and expensive imaging methods would be required for a sensitive and specific method of screening and detection [52,53]. Despite the current recognition that breast implants are not typically intact years after implantation, there remains little information on the potential health effects of ruptured vs. intact implants or of the prudent course for detecting and managing silicone gel-filled breast implant rupture.

Re-operations after mammoplasty Re-operations related to complications and problems with implants have emerged as a major issue with implants. Several studies indicate that women with breast implants are likely to undergo additional surgeries after implantation, in order to address complications or fix unacceptable cosmetic defects due to breast implants [54,55]. Complications and problems include a wide variety of issues, such as breast implant rupture, capsular contracture, infections, gel migration, chronic breast pain, changes in nipple sensation, hematoma, seroma, extrusions, and cosmetic irregularities. Repeated surgeries (re-operations) expose women to all the risks inherent to surgery and also demonstrate that problems with implants persist over a long period of time. The 1999 Institute of Medicine report on the safety of silicone implants stated the following on re-surgeries, due to breast implant rupture and other complications: ‘In general, the frequency of reoperations and local complications is sufficient to be of concern to the committee and to justify the conclusion that this is the primary safety issue with silicone breast implants, and it is certainly sufficient to require very careful and thorough provision of the kind of information contained in this chapter to women considering breast implant surgery. The committee concludes that many of these risks continue to accumulate over the lifetime of a breast implant’ [56].

420

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

In some cases, women may opt to have implants removed and not replaced, because of continuing painful complications and problems. There may be serious cosmetic consequences to this decision [57].

Breast implants and connective tissue disease: formal reviews At the time of the Panel meeting in 1992, there were no epidemiologic studies on the association of connective tissue disease with breast implants. Starting in 1994, there were a series of studies published. The first of these studies raised considerable criticism, since the study was funded by breast implant manufacturers [58,59]. Many subsequent studies were also funded by implant manufacturers. The FDA reviewed these studies on breast implant association with connective tissue disease in several publications [60–62]. The conclusions were similar to those reached by others, such as the Institute of Medicine on the Safety of Silicone Breast Implants and the National Science Panel on Silicone Implants [56,63], who formally reviewed these studies: the studies had some flaws and were not large enough to definitively rule out a small increased risk of connective tissue disease in women with breast implants. However, it can be concluded from these studies that there is not a large association between breast implants and individual connective tissue diseases that were studied or all connective tissue diseases combined. To rule out a small increase in various connective tissue diseases would require a large, well-designed study. It is unlikely that such a massive study would be undertaken, given the mostly negative findings from smaller studies. There were some investigators who believed that women with breast implants were prone to an ‘atypical connective tissue disease’, based on the observation that women with breast implants reported fatigue, myalgias, and arthralgias. However, the search for a silicone-specific syndrome did not reveal a consistent pattern of signs and symptoms that could be attributed to breast implants. A related issue, over which there has been a continuing controversy, is whether there is an increase in other difficult-to-diagnose syndromes in women with implants, specifically fibromyalgia or chronic fatigue syndrome. FDA’s study on breast implant rupture (described above) indicated that women who had extracapsular silicone gel were 3.8 times more likely to self-report doctordiagnosed fibromyalgia than were other women with implants but without extracapsular silicone spread [64,65]. This study differed from other studies in that the implant rupture status for every woman in the study cohort was known at a single moment in time. These women had responded to a questionnaire about their current health status weeks prior to their MRI examination for implant rupture. This study was not definitive because it was essentially a cross-sectional study. It was not known whether implantation occurred before or after the onset of the fibromyalgia. In addition, the FDA study was subject to all the biases of self-reports, since patients may incorrectly report a diagnosis (over- or underreporting may occur). The ideal study would be prospective and longitudinal, would include implant status, medical record review for date of diagnosis (and implantation), and current physician’s diagnosis, as well as a comparable comparison group without implants.

SUMMARY

421

A recent manufacturer’s study in support of approval for silicone gel-filled breast implants reported a statistically significant increase, from baseline (before implantation) to 3 years postimplantation for augmentation patients, of fatigue, exhaustion, joint swelling, frequent muscle cramps, joint pain, combined fatigue, combined pain, and combined (symptoms of) fibromyalgia [66]. Unfortunately, there was no comparable control group to compare the findings with, so it is difficult to interpret these findings and the question of an association of breast implants with fibromyalgia is left unresolved.

Breast implants and suicide Surprising results from a study on mortality among augmentation mammoplasty patients indicated that while overall mortality was slightly lower than the general population, the standardized mortality ratio for suicide was 4.24 (95% CI ¼ 0.9–19.2) for this group compared to similar patients undergoing other plastic surgery procedures [67]. Subsequent studies have been consistent with increased suicide risk among Finnish [68], Swedish [69], and Danish [70] women with breast augmentation. Whether women who are likely to desire and get cosmetic augmentation are in a high-risk group for suicide to begin with is the subject of debate, with some arguing that they may be and others arguing that the increase in suicide is due to depression over serious adverse events associated with breast implants [71–73]. One study suggests that psychotropic drug use, including antidepressants, may be more likely in women with breast implants compared to population controls or women who have had breast reduction surgery [74]. Women with breast implants were particularly more likely to report psychotropic drug use if they had multiple surgical operations due to their implants, suggesting the possibility that multiple breast surgeries because of problems with implants might take a psychological toll on these women. The issue of whether women who desire breast implants are more likely to have a pre-existing psychopathology which increases their risk of suicide, or if breast implants (and/or problems and adverse events associated with implants) cause otherwise healthy women to contemplate and commit suicide, is an issue that will be difficult to resolve.

Summary By the end of 2005, the FDA had received 330 535 reports of problems with breast implants through its adverse event reporting systems. A study of adverse event reports to the FDA by the end of 1995 pointed to breast implant rupture of silicone gel-filled implants as a frequently reported problem. Subsequent studies using the most sensitive method of rupture detection, MRI evaluation, in unselected populations indicates that breast implant rupture is common in older implants. Various committees have studied breast implant issues and have found that, while evidence regarding connective tissue disease and breast implants does not indicate that there is a large excess of specific or combined connective tissue disease, an

422

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

increase too small to have been detected by existing studies cannot be ruled out. Existing studies have found that local complications associated with implants result in frequent re-operations. Local complications, such as implant rupture, gel migration, capsular contracture, breast pain, changes in nipple sensation, hematoma, seroma, infection and extrusions were found to be a continuing source of problems for women with breast implants. While concerns over connective tissue disease and cancer have diminished, there is still concern over interference with mammography, the consequences of implant rupture, and the high rate of re-operation. And questions still arise over whether women with implants are more likely to develop fibromyalgia or other poorly understood syndromes that result in fatigue and chronic aches and pains. Should breast implants return to the market in the USA (they continue to be marketed in Europe), the FDA will require that manufacturers conduct long-term studies that will help to refine current information on both the short-term and potential long-term health effects of these devices. It will also be very important to inform women about these effects. Women should clearly understand that breast implantation, whether for cosmetic or reconstructive purposes, is not without risk.

References 1. Monstrey S, Cristophe A, Delanghe J, De Vriese S et al. What exactly was wrong with the trilucent implants? A unifying hypothesis. Plast Reconstr Surg 2004; 113: 847–856. 2. US FDA breast implant consumer handbook: http://www.US FDA.gov/cdrh/breastimplants/ indexbip.html 3. Sherman LA. Looking through a window of the Food and Drug Administration: US FDA’s advisory committee system. Preclinica 2004; 2:99–102. 4. TDA and polyurethane breast implants: http://www.US FDA.gov/bbs/topics/ANSWERS/ ANS00667.html 5. Hester TR Jr, Ford NF, Gale PJ, Hammett JL et al. Measurement of 2,4-toluendiamine in urine and serum samples from women with Meme or Replicon breast implants. Plast Reconstr Surg 1997; 100: 1291–1298. 6. Silicone in Medical Devices, Conference Proceedings. HHS Publication, US FDA 92–4249, 1992. 7. Council on Scientific Affairs. American Medical Association. Silicone gel breast implants [Council Report]. J Am Med Assoc 1993; 270: 2602–2606. 8. Fisher JC. The silicone controversy: when will science prevail? N Engl J Med 1992; 326: 1696– 1698. 9. Kessler DA, Merkatz RB, Schapiro R. A call for higher standards for breast implants. J Am Med Assoc 1993; 270: 2607–2608. 10. Kessler DA. The basis of the US FDA’s decision on breast implants. N Engl J Med 1993; 326: 1713–1715. 11. Bright RA, Jeng LL, Moore RM. National survey of self-reported breast implants: 1988 Estimates. J Long Term Eff Med Dev 1993; 3: 81–89. 12. Terry MB, Skovron ML, Garber S, Sonnenschein E, Toniolo P. The estimated frequence of cosmetic breast augmentation among US women, 1963–1988. Am J Publ Health 1996; 86: 891–892.

REFERENCES

423

13. Cook RR, Perkins LL. The prevalence of breast implants among women in the United States. Curr Top Microbiol Immunol 1996; 210: 419–425. 14. Device and Patient Problem Codes: http://www.US FDA.gov/cdrh/mdr/373_appdxb.html 15. Brown SL, Parmentier CM, Woo EK, Vishnuvajjala RL, Headrick ML. Silicone gel breast implant adverse event reports to the US FDA, 1984–1995. Publ Health Rep 1998; 113: 535–543. 16. Guidance for industry – medical device reporting – alternative summary reporting (ASR) program: http://www.US FDA.gov/cdrh/osb/guidance/315.html 17. Brown SL, Hefflin B, Woo EK, Parmentier CM. Infections related to breast implants reported to the Food and Drug Administration, 1977–1997. J Long Term Eff Med Implants 2001; 11: 1–12. 18. Pajkos A, Deva AK, Vickery K, Cope C et al. Detection of subclinical infection in significant breast implant capsules. Plast Reconstr Surg 2003; 111: 1605–1611. 19. Pittet B, Montadon D, Pittet D. Infection in breast implants. Lancet Infect Dis 2005; 5: 94–106. 20. Darouiche RO. Treatment of infections associated with surgical implants. N Engl J Med 2004; 350: 1422–1429. 21. Brinton LA, Lubin JH, Burich MC, Colton T et al. Breast cancer following augmentation mammoplasty (United States). Cancer Causes and Control 2000; 11: 819–827. 22. Brown SL, Todd JF, Luu HM. Breast implant adverse events during mammography: reports to the Food and Drug Administration. J Women’s Health 2004; 13: 371–378. 23. Hayes H Jr, Vandergrift J, Diner WC. Mammography and breast implants. Plast Reconstr Surg 1988; 82: 1. 24. Eklund GW, Busby RC, Miller SH, Job JS. Improved imaging of the augmented breast. Am J Roentgenol 1988; 151: 469–473. 25. Bondurant S, Ernster V, Herdman R (eds). Safety of Silicone Implants. Washington, DC: National Academy Press, 1999; 248–263. 26. Brown SL, Todd JF, Cope JU, Sachs HC. Breast implant surveillance to the US Food and Drug Administration: maternal–child health problems. J Long Term Eff Med Impl 2006; 16: 281–290. 27. Kjo¨ller K, Friis S, Signorello LB, McLaughlin JK et al. Health outcomes in offspring of Danish mothers with cosmetic breast implants. Ann Plast Surg 2002; 48: 238–245. 28. Destouet JM, Monsees BS, Oser RF, Nemecek JR et al. Screening mammography in 350 women with breast implants: prevalence and findings of implant complications. Am J Roentgenol 1992; 159: 973–978. 29. Scaranelo AM, Marques AF, Smialowski EB, Lederman HM. Evaluation of the rupture of silicone breast implants by mammography, ultrasonography, and magnetic resonance imaging in asymptomatic patients: correlation with surgical findings. Sao Paulo Med J 2004; 122: 41–47. 30. Goodman CM, Cohen V, Thornby J, Netscher D. The life span of silicone gel breast implants and a comparison of mammography, ultrasonography, and magnetic resonance imaging in detecting implant rupture: a meta-analysis. Ann Plast Surg 1998; 41: 577–578. 31. Middleton MS. Magnetic resonance evaluation of breast implants and soft-tissue silicone. Top Magn Reson Imaging 1998; 9: 92–137. 32. de Camara DL, Sheridan JM, Kammer BA. Ruptures and aging of silicone gel breast implants. Plast Reconstr Surg 1993; 91: 828–834. 33. Rolland C, Guidoin R, Marceau D, Ledoux R. Nondestructive investigations on ninety-seven surgically excised mammary prostheses. J Biomed Mater Res Appl Biomater 1989; 23:285–298. 34. Malata CM, Varma S, Scott M, Liston JC, Sharpe DT. Silicon breast implant rupture: common/ serious complication? Med Prog Technol 1994; 20: 251–260. 35. Peters W, Keystone E, Smith D. Factors affecting the rupture of silicone gel breast implants. Ann Plast Surg 1994; 32: 449–451.

424

CH27

SILICONE GEL-FILLED BREAST IMPLANTS: SURVEILLANCE AND EPIDEMIOLOGY

36. Robinson OG, Bradley EL, Wilson DS. Analysis of explanted silicone implants: a report of 300 patients. Ann Plast Surg 1995; 34: 1–6. 37. Brown SL, Silverman BG, Berg WA. Rupture of silicone gel breast implants: causes, sequelae and diagnosis. Lancet 1997; 350: 1531–1537. 38. Kulber DA, Mackenzie D, Steiner JH, Glassman H et al. Monitoring the axilla in patients with silicone gel implants. Ann Plast Surg 1996; 35: 580–584. 39. Clodius L. Local and regional lymph node response to ruptured gel-filled mammary prostheses. Br J Plast Surg 1978; 31: 349–352. 40. Wintsch W, Smahel J, Hausner RJ, Schoen FJ et al. Migration of silicone gel to axillary lymph nodes after prosthetic mammoplasty. Arch Pathol Lab Med 1981; 105: 371–372. 41. Teuber SS, Reilly DA, Howell L, Oide C, Gershwin ME. Severe migratory granulomatous reaction to silicone gel in three patients. J Rheumatol 1999; 26: 699–704. 42. Puckett MA, DeFriend D, Williams MP, Roobottom CA. A leaking breast prosthesis presenting as an abdominal mass. Br J Radiol 2004; 77: 790–791. 43. Malyon AD, Dunn R, Weiler-Mithoff EM. Expanding silicone granuloma. Br J Plast Surg 2001; 54: 257–259. 44. Pfleiderer B, Garrido L. Migration and accumulation of silicone in the liver of women with silicon. Magn Reson Med 1995; 33: 8–17. 45. Flassbeck D, Pfleiderer B, Klemens P, Heumann KG et al. Determination of siloxanes, silicon, and platinum in tissues of women with silicone gel-filled implants. Anal Bioanal Chem 2003; 375: 356–362. 46. Birkefeld AB, Eckert H, Pfleiderer B. A study of the aging of silicone breast implants using 29Si 1 H relaxation and DSC measurements. Biomaterials 2003; 25: 4405–4413. 47. Pfleiderer B, Ackerman JL, Garrido L. Migration and biodegradation of free silicone from silicone gel-filled implants after long-term implantation. Magn Reson Med 1993; 30: 534–543. 48. Brown SL, Middleton MS, Berg WA, Soo MS, Pennello G. Prevalence of rupture of silicone gel breast implants revealed on MR imaging in a population of women in Birmingham, Alabama. Am J Roentgenol 2000; 175:1057–1064. 49. Berg WA, Nguyen TK, Middleton MS, Soo MS et al. MR imaging of extracapsular silicone from breast implants: diagnostic pitfalls. Am J Roentgenol 2002; 178: 465–472. 50. Hlmich LR, Kjoller K, Vejborg I, Conrad C et al. Prevalence of silicone breast implant rupture among Danish women. Plast Reconstr Surg 2001; 108: 848–858. 51. Marotta JS, Widenhouse CW, Habal MB, Goldberg EP. Silicone gel breast implant failure and frequency of additional surgeries: analysis of 35 studies reporting examination of more than 8000 explants. J Biomed Mater Res Appl Biomater 1999; 48: 354–364. 52. Holmich LR, Fryzek JP, Kjoller K, Breiting VB et al. The diagnosis of silicone breast implant rupture: clinical findings compared with findings at magnetic resonance imaging. Ann Plast Surg 2005; 54: 583–589. 53. Netscher DT, Weizer G, Malone RS, Walker LE et al. Diagnostic value of clinical examination and various imaging techniques for breast implant rupture as determined in 81 patients having implant removal. South Med J 1996; 89: 397–404. 54. Gabriel SE, Woods JE, O’Fallon WM, Beard CM et al. Complications leading to surgery after breast implantation. N Engl J Med 1997; 336: 677–682. 55. Brown SL, Pennello G. Replacement surgery and silicone gel breast implant rupture: self-report by women after mammoplasty. J Women’s Health Gender Based Med 2002; 11: 255–264. 56. Bondurant S, Ernster V, Herdman R (eds). Safety of Silicone Breast Implants. Washington, DC: National Academy Press. 57. Breast implant potential complications and reoperations: http://www.US FDA.gov/cdrh/ breastimplants/breast_implant_risks_brochure.html [accessed August 7 2005].

REFERENCES

425

58. Gabriel SE, O’Fallon WM, Kurland LT, Beard CM et al. Risk of connective-tissue diseases and other disorders after breast implantation. N Engl J Med 1994; 330: 1697–1702. 59. Risk of connective tissue disease and other disorders after breast implantation. [letters]. N Engl J Med 1994; 331:1231–1234. 60. Silverman BG, Brown SL, Bright RA, Kaczmarek RG et al. Reported complications of silicone gel breast implants: an epidemiologic review. Ann Intern Med 1996; 124: 744–756. 61. Brown SL, Langone JJ, Brinton LA. Silicone breast implants and autoimmune disease. J Am Med Assoc 1998; 53: 21–24, 40. 62. Brown SL. Epidemiology of silicone gel breast implants. Epidemiology 2002; 13: S34–S39. 63. Janowsky EC, Kupper LL, Hulka BS. Meta-analysis of the relationship between silicone breast implants and the risk of connective-tissue disease. N Engl J Med 2000; 342: 781–790. 64. Brown SL, Pennello G, Berg WA, Soo MS, Middleton MS. Silicone gel breast implant rupture, extracapsular silicone, and health status of a population of women. J Rheumatol 2001; 28: 996–1003. 65. Brown SL, Duggirala HJ, Pennello G. An association of silicone gel breast implant rupture and fibromyalgia. Curr Rheumatol Rep 2002; 4: 293–298. 66. US FDA Summary Panel memorandum, March 2 2005. General and Plastic Surgery Devices Panel. Mentor Corporation silicone gel-filled breast implants, p. 63: http://www.US FDA.gov/ ohrms/dockets/ac/05/briefing/2005–4101b1.htm [accessed July 15 2005]. 67. Brinton LA, Lubin JH, Burich MC, Colton T, Hoover RN. Mortality among augmentation mammoplasty patients. Epidemiology 2001; 12: 321–326. 68. Pukkala E, Kulmala I, Hovi S-L, Hemminki E et al. Causes of death among Finnish women with cosmetic breast implants, 1971–2001. Ann Plast Surg 2003; 51: 339–342. 69. Koot VC, Peeters PH, Granath F, Grobbee DE, Nyren O. Total and cause specific mortality among Swedish women with cosmetic breast implants: prospective study. Br Med J 2003; 326: 527–528. 70. Jacobsen PH, Ho¨lmich LR, McLaughlin JK, Johansen C et al. Mortality and suicide among Danish women with cosmetic breast implants. Arch Intern Med 2004; 164: 2450–2455. 71. Sarwer DB. Invited discussion: causes of death among Finnish women with cosmetic breast implants, 1971–2001. Ann Plast Surg 2003; 51: 343–344. 72. McLaughlin JK, Wise TN, Lipworth L. Increased risk of suicide among patients with breast implants: do the epidemiologic data support psychiatric consultation? Psychosomatics 2004; 45: 277–280. 73. Zuckerman D. Mortality in Swedish women with cosmetic breast implants [letter]. Br Med J 2003; 326: 1266. 74. Breiting VB, Holmich LR, Brandt B, Fryzek JP et al. Long-term health status of Danish women with silicone breast implants. Plast Reconstr Surg 2004; 114: 217–226.

28 Ophthalmic devices and clinical epidemiology Malvina B. Eydelman, Gene Hilmantel, James Saviola, and Don Calogero US Food and Drug Administration, Rockville, MD, USA

Introduction FDA regulates a wide variety of medical devices intended for diagnosis and treatment of ocular problems. In order to ascertain device safety and effectiveness, clinical epidemiologic studies are frequently utilized. Two of the most widely used ophthalmic devices are intraocular lenses (IOLs) and contact lenses. These devices are highly effective in correcting refractive errors resulting from aphakia, hyperopia and myopia. Clinical epidemiological studies have had an important influence on FDA’s premarket and postmarket evaluation of IOL and contact lens safety. This chapter summarizes some of the studies establishing that IOLs and contact lenses are relatively safe devices, with very low rates of serious adverse events. Worldwide use of these devices over many decades has confirmed FDA’s assessment.

Epidemiological contributions to IOL evaluation Cataract is generally defined as an opacification or loss of transparency in the crystalline lens of the eye. The clinical consequence of such opacification of the lens is reduction of the visual acuity. Cataract is the leading cause of blindness worldwide [1]. It is also the leading cause of vision loss in the USA [2]. An estimated 20.5 million (17.2%) Americans older than 40 years have cataract in either eye [3]. Many risk factors of age-related

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

428

CH28

OPHTHALMIC DEVICES AND CLINICAL EPIDEMIOLOGY

cataract have been identified during the past 20 years [4–6]. To date, however, the only treatment known to eliminate existing cataract is surgical removal of the lens. Historically, cataract patients had an intracapsular cataract extraction followed by aphakic spectacles. The modern era of implantation of an artificial lens at the time of cataract extraction began with Harold Ridley. During World War II, many ophthalmologists had noted that perforating eye injuries from airplane canopies made from acrylic Perspex plastic often resulted in minimal intraocular irritation secondary to the material itself. This observation, in conjunction with complaints from aphakic patients about their poor quality of vision, prompted Harold Ridley to design the first intraocular lens [7]. The first IOL implantation occurred in 1949 in the UK. Subsequently, IOLs have undergone numerous modifications in the design and surgical implantation procedures. In 1981, the first IOL was approved by the US Food and Drug Administration (FDA) for the correction of aphakia following cataract extraction. Since then, implantation of an IOL at the time of cataract extraction surgery has become the standard of care. Today, cataract surgery has reached an extraordinarily high level of quality and performance, and 20/20 uncorrected vision on the first postoperative day is not uncommon [8]. As a result, in the USA, cataract surgery has become the most frequently performed surgical procedure among the 30 million Medicare beneficiaries, with approximately 1.35 million cataract operations performed annually [9]. Since 1981, there have been more than 1000 IOL models approved in the USA. FDA’s unique use of epidemiological data associated with IOLs has been critical in the evaluation of these devices and has allowed the introduction of these new models into the US market, using a ‘least burdensome’ approach. During the early 1980s, results of cataract surgeries with IOL implantations were compared by FDA to the visual results and complications previously reported in the literature for cataract surgery without IOL implantation. In 1983, Stark et al. [10] compiled data for 17 different IOLs from seven manufacturers, with a total of 45 543 cases. They stratified data by the four classes of IOLs available at that time: posterior chamber, anterior chamber, iris fixation, and iridocapsular. Data for each class of IOLs was subsequently analyzed for the key efficacy outcome – final visual acuity of 20/40 or better after IOL implantation. To determine whether the differences observed in visual outcomes were due to the IOL used or due to the patient selection, the authors performed a best-case analysis by excluding those patients with pre-existing pathology. Age appeared to be a factor in reducing visual outcomes and the authors generated efficacy outcomes stratified by age decades for the entire cohort, as well as best-case patients with no preoperative ocular pathology or macular degeneration detected at any time. In addition, an analysis of all adverse events was performed and stratified by the class of IOL. Sight-threatening complications were tabulated in two ways: ‘cumulative’, if the complication occurred at any time during the first year; or ‘persistent’, if the complication was present at the 12–14 month follow-up visit. Outcomes of this large epidemiological study, sometimes referred to as the ‘Stark grid’, proved to be extremely useful. In the early 1980s, FDA began using the summary tables generated by Stark et al. for comparison to the outcomes with new IOLs during their evaluation for marketing

EPIDEMIOLOGICAL CONTRIBUTIONS TO IOL EVALUATION

429

approval. The manufacturers of these devices found the ‘Stark grid’ very useful for internal data audits, prior to data submission to the FDA. Subsequently, FDA personnel created an ‘FDA grid’, which has been updated several times using the clinical outcomes from the most recent IOL marketing approvals but remains very similar in concept and types of analyses to the original ‘Stark grid’. The new grid includes data utilized by IOL manufacturers in their marketing approval process for their respective devices. The latest FDA grid represents statistical analysis of more than 5000 eyes and represents modern surgical techniques and IOL materials. The international ophthalmic community has accepted FDA’s approach of utilizing a historical control for investigations of new IOL models. The ‘FDA Grid’ was incorporated into an international standard for IOL clinical investigations [11], and thus serves as an international reference for safety and efficacy endpoints for new IOL investigations. Another unique use of epidemiological data that influenced the IOL approval process and public policy in the USA was the work conducted by FDA staff in the evaluation of the safety of IOLs in adults younger than 60 years [12]. When the FDA Ophthalmic Devices panel, composed of experts in ophthalmology and optometry, considered the first IOLs to be approved in 1981, the recommendation for approval was limited to adults 60 years and older because of the lack of long term data. Representatives of the American Intra-Ocular Implant Society echoed the concerns of the panel members and recommended a conservative approach. FDA agreed with the recommendations. Thus, labeling for the first IOLs limited use of these devices to adults 60 years of age and older. In subsequent years, none of the IOL sponsors designed clinical investigations to specifically address the safety and efficacy of IOLs in younger adults. Thus, FDA did not have sufficient valid scientific evidence to support a change of indication based solely on submissions to the Agency. Therefore, until 2003, all IOLs in the USA were approved for use in adults 60 years and older only. IOL manufacturers were not allowed to label or advertise their devices for use in patients younger than 60 years. This contradicted common use of IOLs for treatment of younger patients by clinicians as a practice of medicine. With the development of phakic IOLs, designed to correct refractive errors in mostly younger patients, the need for assessing the safety of all IOLs in patients younger than 60 became urgent. In order to establish that the performance of IOLs in adults younger than 60 years was comparable to that in adults older than 60 years, a collaborative effort was undertaken with the American Academy of Ophthalmology (AAO) and Storm Eye Institute to resolve this dilemma. Data from the FDA, the AAO National Eyecare Outcome Network (NEON), and Storm Eye Institute databases were analyzed for safety and efficacy outcomes. In addition, a comprehensive literature review was conducted to identify safety and efficacy outcomes and their relationship to patient age at the time of implantation. The cumulative data available in the ‘FDA grid’ were assessed for short-term (fewer than 18 months) clinical outcomes. Following stratification by IOL class, the 1 year postoperative visual acuities in patients younger than 60 years (n ¼ 418) were compared with the visual acuities in patients 60 years and older (n ¼ 4821). Since FDA’s data on IOLs implanted in younger adults was rather sparse, these data were supplemented with the data from the AAO NEON database, which, as described by Lum et al. [13], contains

430

CH28

OPHTHALMIC DEVICES AND CLINICAL EPIDEMIOLOGY

visual acuity and adverse event data on more than 7000 eyes stratified by age decade. The AAO NEON database short-term efficacy and adverse event data from patients younger than 60 and those 60 and older were stratified by IOL class and compared. In addition, postmarketing reports of IOL adverse events and complications in the FDA’s Medical Device Reporting (MDR) system were evaluated for the long-term clinical outcomes of IOLs implanted in younger adults. This database is designed to capture reports of product malfunctions and failures as well as serious injuries and deaths related to the use of medical devices. The reports submitted to the FDA MAUDE and MDR databases during December 1993–mid-April 2000 were reviewed. When the age of the patient involved was reported (n ¼ 1168), these reports were stratified accordingly (patients younger than 60 years, n ¼ 184; patients 60 years and older, n ¼ 984). The database from the Center for Research on Ocular Therapeutics and Biodevices at the Storm Eye Institute of the Medical University of South Carolina was also utilized for analysis of long-term (more than 18 months of follow-up) clinical outcomes. The Center had analyzed 6790 human eyes implanted with IOLs obtained postmortem during January 1985–March 2000. This database was reviewed for length of implantation and ocular pathology present at the time of death for each patient who received an IOL at a given age; 172 post mortem eyeballs were available from patients who were younger than 60 years at the time of implantation and 6060 post mortem eyeballs from patients who were 60 years and older. The center had also analyzed 6143 explanted IOLs that were submitted in January 1985–March 2000. Clinical reasons for explantation, IOL type, and age of the patient were collected. The rates of complications leading to explantation were stratified by type of IOL and patient age. Analysis of data from FDA databases, the AAO NEON database, post mortem eyeballs and explanted IOLs from the Center for Research on Ocular Therapeutics and Biodevices, and the published literature, allowed FDA personnel to conclude that there was substantial scientific evidence to support the use of IOLs in adults younger than 60 years. Their conclusions were presented in the publication, ‘Retrospective evaluation of intraocular lenses in adults younger than 60 years’ [12]. References to this publication, with its extensive review of epidemiological data, has allowed manufacturers to request that FDA change the indication for their IOLs from ‘use in adults 60 years and older’ to ‘use in all adults’. In summary, utilization of epidemiological data has played a major role in the evaluation of IOLs by FDA. It has allowed FDA to develop policies that helped to facilitate the 25 year improvement in the quality of cataract patients’ treatment unparalleled in medicine today [8].

Epidemiology of contact lens ulcers and public policy Millions of people wear contact lenses safely. However, contact lens wear is not without serious complications; contact lens-related corneal ulcers pose a risk of significant vision loss. A corneal ulcer is a localized area of tissue erosion associated with inflammatory cells in the cornea. Corneal ulcers can be either infectious or noninfectious [14]. When

EPIDEMIOLOGY OF CONTACT LENS ULCERS AND PUBLIC POLICY

431

infectious, the disease is clinically referred to as ‘microbial keratitis’ (MK). Infectious corneal ulcers are more serious than sterile ulcers, and usually cause significant scarring of the cornea. Corneal scarring opacifies the normally clear corneal tissue; this can lead to visual impairment or blindness when the resulting opacity covers the pupil. In the USA, contact lenses are regulated by the FDA. Rigid, plastic, corneal contact lenses made from polymethylmethacrylate began to achieve a significant level of use in the 1950s. However, it was not until the 1970s, after soft lenses were approved for marketing, that contact lenses became a popular alternative to spectacles for the correction of refractive error. The FDA regulations required significant clinical testing of soft contact lenses prior to marketing approval in order to provide a reasonable assurance of safety and effectiveness. In 1981, the FDA approved the first cosmetic extended wear lenses (‘extended wear’ means that the lenses could be worn overnight, during sleep; in this context, ‘cosmetic’ means not restricted to the special use of correcting vision after cataract surgery). In the 1980s, extended wear use for periods of up to 30 continuous days began to increase in the contact lens-wearing population. In the 1970s and early 80s, reports of corneal ulcers in contact lens wearers started appearing in the literature [15,16]. These cases seemed to be associated with soft lenses – particularly extended wear soft lenses. Spurred by FDA’s growing public health concerns, the industry-supported Contact Lens Institute sponsored the first population-based studies to assess the level and significance of the problem. The results were published in 1989 in two landmark papers by Oliver Schein, Eugene Poggio, and their co-workers [17,18]. Their work represented the first significant epidemiologic research that determined the incidence of corneal ulcers and the relative risk of different modes of wear. To estimate the incidence of ulcers in different modes of contact lens wear, investigators had to determine: (a) the number of contact lens ulcers in a given population; and (b) the number of patients in that population using each type of contact lens [17]. Investigators contacted all ophthalmologists in five New England states to collect the total number of lens-related, new ulcer cases over a 4 month period. They determined the proportion of the New England population using each type of contact lens by a telephone survey of about 4000 randomly-selected households. The investigators combined the survey data on contact lens usage with census population data to estimate the number of people in the area who wore each type of contact lens. Thus, annual incidence was calculated using the following formula: Annual incidence ¼

3 ðnumber of cases in 4 monthsÞ population ðproportion of people wearing lensesÞ

Poggio et al. [17] reported that the annualized incidence of MK was 20.9 cases per 10 000 patient-years in users of extended wear soft lenses, and 4.1 per 10 000 patient-years in users of daily wear soft lenses (designed for daytime use only). From these rates, they estimated that annually among the 13 million nationwide contact lens wearers in the late 1980s, there were about 8000 cases of MK in extended wear lens users and about 4000 in daily wear users. Not all patients wore their lenses in the mode appropriate for their

432

CH28

OPHTHALMIC DEVICES AND CLINICAL EPIDEMIOLOGY

lens type. Some slept in ‘daily wear lenses’ and some never slept in ‘extended wear lenses’. The authors estimated MK rates based upon actual lens wear. The incidence for daily wear lenses worn strictly on a daily wear schedule (never worn overnight) was about 2–3 per 10 000 patient-years; the incidence for extended wear lenses worn overnight was about 22 to 32 per 10 000 patient-years [17]. In the companion study, Schein et al. [18] reported on a case-control study of soft lens patients conducted at six university ophthalmology centers. They compared 86 patients with MK to two different control populations: one hospital-based (n ¼ 61) and one population-based (n ¼ 410). Investigators found that extended wear soft lenses were associated with approximately four times the risk of MK as daily wear soft lenses. As discussed above, some patients followed wearing schedules inconsistent with lens type. Patients who used extended wear lenses in an overnight wear mode showed 10–15 times the risk of those who wore daily wear lenses on a strictly daily wear basis. Increasing length of extended wear was directly related to increased risk. The risk for patients who wore extended wear lenses 2–7 days before removal was about 7–10 times the risk for those who wore daily wear lenses on a strictly daily wear basis. By combining this with the Poggio [17] incidence data, we can conclude that for this ‘2–7 day’ continuous wear schedule, the incidence was on the order of about 20 per 10 000 patient-years. Somewhat unexpectedly, investigators found that wearers who smoked were at higher risk than non-smokers. Based upon Schein’s data relating length of extended wear to increased risk of MK, the FDA recommended limiting continuous wear to a maximum of 7 days. In 1989, the FDA asked manufacturers to voluntarily reduce the maximum indicated time for extended wear to 7 days, and sent letters to eyecare practitioners advising them of the situation. In the USA, cosmetic contact lenses were not marketed for continuous wear longer than 7 days for the rest of the twentieth century. Following up on the work of Schein and Poggio, investigators in a number of other countries have conducted similar studies over the past 15 years. An important 1991 study in the UK [19] suggested that contact lens wear had become the primary cause of MK. This case-control study assessed the relative risk (RR) of several predisposing factors associated with MK (compared to ‘no predisposing factor’). These relative risks for the most important factors were as follows: Contact lens wear: RR ¼ 80:1. Trauma: RR ¼ 13:9. Ocular surface disease: RR ¼ 7:4. This study also confirmed earlier findings that extended wear was significantly riskier than daily wear; that extended wear for > 6 days was riskier than for shorter time periods; that contact lens hygiene probably was a significant, but limited, factor in risk; and that rigid lens wear was safer than soft lens wear. Analysis showed that more than half of all cases of MK were associated with contact lens wear.

EPIDEMIOLOGY OF CONTACT LENS ULCERS AND PUBLIC POLICY

433

In a study published in 1994 [20], investigators surveyed all cases of ‘contact lensinduced keratitis’ (epithelial defects with an underlying infiltrate or ulcer) in Sweden over a 3 month period. The maximum length of continuous wear was 14 nights. This study found substantially lower annualized incidences than the US study: about 2/10 000 for daily wear soft lenses, and about 10–13/10 000 for extended wear soft lenses. Cheng et al. [21] reported the incidence of contact lens corneal ulcers in Denmark in the late 1990s to be very similar to the rates in the USA reported in the earlier Poggio [17] study. This survey of all practicing Danish ophthalmologists over a 3 month period found an annualized incidence of 3.5/10 000 in daily wear soft lenses and 20.0/10 000 in extended wear soft lenses (up to 2 weeks of continuous wear). This study also reported on the morbidity of the ulcerative disease. Of the 92 eyes with MK, 5% ended therapy with vision of 20/70 or worse, 13% with 20/30 or worse. A study [22] in Hong Kong in the late 1990s reported the incidence of MK to be about 3/10 000 for daily wear soft lenses, and about 9/10 000 for extended wear soft lenses (maximum of 6 nights of continuous wear). Although the incidence of significant problems was low in most studies, it must be kept in mind that there are currently an estimated 36 million contact lens wearers in the USA [23]. Therefore, it is likely that there are thousands of lens-related cases of MK each year. Additionally, many patients wear contact lenses for a significant portion of their lives. A patient who uses extended wear lenses from age 20 to age 39 will have a cumulative risk of MK of approximately 0.04 (20 years 0.0020 ulcer risk/year), if one assumes a constant risk using the Poggio [17] estimate. For this hypothetical patient, a 1 in 25 chance of getting ulcerative keratitis would represent a significant risk. An early illustration of how epidemiologic research influenced public policy was provided by the investigation of Acanthamoeba contamination of contact lens solutions. The majority of contact lens-related MK cases are bacterial. However, Acanthamoeba, a free-living amoeba commonly found in fresh water and soil, causes a relatively small number of cases, which are difficult to treat [24]. Acanthamoeba keratitis cases related to contact lens wear were initially seen in the early to mid1980s [25,26]. In the 1980s, many patients made their own contact lens saline solution from distilled water and salt tablets (for use in heat disinfection units). This was considerably less expensive than purchasing commercial contact lens saline solution. In an early case-control study of 27 cases and 81 matched controls conducted by the Centers for Disease Control [27], investigators identified the use of home-made nonsterile saline solution as a highly significant risk factor. Based largely upon this study, in 1987 the FDA sent to eye care practitioners a safety alert concerning home-made saline, and requested stronger warnings on salt tablet containers. Since then, patient use of home-made saline has almost become nonexistent. This decline was also related to the gradually decreasing use of home heat disinfection as chemical disinfection products came to dominate the market (note that, except for the early Poggio [17] study, all of the MK incidence rates found in the previously cited studies were determined after the risks of home-made saline were well known.) Microbial keratitis in contact lens wear is thought to be related to factors such as corneal hypoxia, contamination of contact lenses and solutions, changes in the tear film,

434

CH28

OPHTHALMIC DEVICES AND CLINICAL EPIDEMIOLOGY

and microtrauma [28]. Hypoxia (lack of oxygen flow to the cornea under the contact lens) has generally been considered to be the most important factor. With the development of new silicone-containing contact lens materials in the late 1990s, many clinicians believed that safer extended wear of contact lenses might be at hand. The siliconecontaining lenses could pass many times more oxygen to the cornea than lenses made from conventional materials. For the first time since the 1980s, manufacturers requested approval from the FDA of up to 30 days of continuous wear for some of these ‘hyperpermeable’ contact lenses. The FDA realized that conventional premarket studies of 400–800 patients would be insufficient to determine the risk of MK in any new 30 day continuous wear lenses. In order to both safeguard public health and provide a ‘least burdensome’ approach for the introduction of new technology, the FDA offered manufacturers two options. They could either have an unconventionally large preapproval study, or have a more conventionally sized preapproval study, with approval contingent on the requirement to run a large postapproval, population-based study to determine the incidence of ulcerative disease. To date, all manufacturers have chosen the latter option, using significant infiltrative keratitis (including non-infectious and infectious cases) as a surrogate endpoint in premarket studies for these new lenses. An FDA advisory panel met in November of 2000 for discussion of the nature of any postapproval studies. Two different epidemiologic approaches were suggested for these studies. The first was to prospectively follow a large cohort of 30 day lens wearers for at least 1 year to accumulate enough ‘patient-years’ of exposure to make a reasonable estimate of the disease incidence. The second was to do a ‘case-control’ study to assess the relative risk of the new hyperpermeable lenses (worn up to 30 days continuously) compared to the previously approved conventional extended wear lenses (worn up to 7 days). The advisory panel recommended that the new lenses should not have a rate of MK substantially higher than that in the currently approved 7 day wear lenses (generally thought to be in the order of 20 per 10 000 patient-years). In a case-control study of this type, an investigation would collect a sample of ‘cases’ of extended-wear-related MK and would also collect a sample of extended wear contact lens patients without MK as a control. For each group, investigators would determine the number of patients in 7 day wear and the number in 30 day wear. From these numbers, investigators could estimate the relative risk of MK in the two modes of wear. It may not be immediately apparent how the relative risk can be estimated from this type of study. Consider the situation in which the entire population of extended wear patients is available. They could be placed in a table of the following type:

7-Day extended wear 30 day extended wear

MK present

No MK

a c

b d

Here, a, b, c, and d represent the number of patients in each category. The proportion of 7 day wearers with MK is: a/ða þ b). Likewise, the proportion of 30 day wearers with MK

EPIDEMIOLOGY OF CONTACT LENS ULCERS AND PUBLIC POLICY

435

is: c/ðc þ d). The relative risk of 30 day to 7 day wear is: RR ¼

c/ðc þ dÞ a/ða þ bÞ

This cannot be directly estimated from the case and control sample data. However, consider the ‘odds’ that a 7 day wearer will get MK, as opposed to not getting MK. These ‘odds’ are a/b. Similarly the ‘odds’ of a 30 day wearer of getting MK are c/d. The ‘odds ratio’ of 30 day wear to 7 day wear is: OR ¼

c/d a/b

Notice that this ratio is mathematically equivalent to: cb c/a ¼ ad d/b The numerator of the last expression (c/a) can be estimated from a representative sample of extended wear MK cases. Similarly, the denominator (d/b) can be estimated from a representative sample of extended wear patients who do not have MK. Thus, the sample ratio: c/a d/b provides an estimate of the population ‘odds ratio’. Note that when the incidence of a disease is very low (as is the case with MK), the ‘odds’ of getting a disease, e.g. c/d, are approximately equal to the risk of getting the disease, e.g. c/ðc þ dÞ; where c << d. Thus, in this case, the ‘odds ratio’ calculated from the case-control data can provide a good estimate of the ‘relative risk’ [29]. The case-control study design has a number of points in its favor, and is often the method of choice for studying diseases of low incidence. Aside from permitting an assessment of the relative risk of different modes of wear, it also permits the estimation of the risk presented by other co-factors (actual length of continuous wear, hygiene, patient compliance, age, etc.), uses patients in an uncontrolled ‘real world’ environment, and can be conducted with a small sample size at relatively low cost. Disadvantages include the following: Estimates can be prone to several types of selection bias. In particular, if both the exposure factor and the outcome are causes for referral, a type of bias known as Berkson’s bias results [30]. The actual incidence of the disease cannot be ascertained.

436

CH28

OPHTHALMIC DEVICES AND CLINICAL EPIDEMIOLOGY

If the number of patients in any one of the cells is very small, the confidence interval for the relative risk will be very large. The reason for this last factor can be clarified by considering the equation for the variance of the log of the ‘odds ratio’ [31]: varðln ORÞ ¼ 1/a þ 1/b þ 1/c þ 1/d As can be seen, if any one of a, b, c, or d are very small, the variance will be quite large. With regard to 30 day lenses, the FDA believed that it would take quite a few years (at least 3–5) for the new lenses to establish a substantial market share and thereby produce enough ‘30 day ulcers’ to make the ‘c’ number in the above table large enough to produce a reasonable variance. As opposed to the case-control design, prospectively following a large cohort has a number of advantages. It can: Directly assess the actual risk of the of the disease. Achieve results in a timely manner, perhaps within 2–3 years of initiation. However, a prospective study might be subject to some of the same problems that can occur in a controlled premarket study: possible self-selection of more responsible, motivated patients or practitioners, and a relatively controlled follow-up environment. It also requires a larger patient sample with resulting increased cost. After considering all the factors, the FDA recommended the prospective cohort study design, although it is still open to other approaches. Probably the most important consideration was the importance of getting the information in a timely manner, as a case-control design would likely necessitate a longer delay after approval. In the recently initiated postapproval studies, each protocol involved approximately 100 monitoring sites and was designed to collect data on 4500–5000 patient-years of 30 day wear [32]. Data was to be collected from a variety of clinical settings, such as private optometry and ophthalmology practices, and commercial optical chains. Endpoints were defined as cases of presumed microbial keratitis and cases of reduction in visual acuity. Any study of contact lens-related microbial keratitis has the problem of distinguishing noninfectious (sterile) ulcers from infectious ones (MK). Infectious ulcers are more likely than sterile ulcers to be central, large, and associated with significant inflammation within the eye (anterior chamber reaction) and severe pain [14]. The ulcers that provide a less severe clinical impression are often referred to as ‘contact lens peripheral ulcers’ (although lesions given this designation are not always ‘peripheral’) and are often considered to be sterile [33]. Presumed infectious ulcers often do not grow significant bacteria when cultured [34] and ‘contact lens peripheral ulcers’ are often culture-positive [35]. Since the clinical features of the two types of keratitis overlap, both are usually treated vigorously with topical antibiotic drops. It has been argued that some of the population-based studies may have included significant numbers of non-infectious

CONCLUSION

437

ulcers, thereby artificially inflating the estimates of the incidence of MK [36]. The FDA is attempting to ensure that each postapproval study makes the distinction between infectious and sterile ulcers in as consistent and objective a manner as possible. The first postapproval study for a 30 day continuous wear contact lens, a prospective study conducted by CIBAVision [37], was published in December 2005. The study used 131 centers to recruit 6245 participants, who ultimately accumulated 5561 patient-years of exposure (4292 patient-years for typical continuous wear of at least 3 weeks). Questionnaires were used at 3 and 12 months to ascertain typical time for continuous wear, and to determine the occurrence of red or painful eyes. For patients who reported having had red or painful eyes, medical records were obtained. The study used a special Endpoint Adjudication Committee of recognized experts in order to categorize infiltrative cases. For all cases with a corneal infiltrate, this committee reviewed the medical records in detail and made a determination concerning likelihood of etiology, based upon predetermined guidelines. Ten patients in the study were classified as having MK; two of these experienced vision loss and eight did not. Thus, the incidence of patients with presumed MK for the study (all patients, regardless of wearing schedule) was 10 per 5561 patient-years, or 18.0 per 10 000 patient-years (95% CI ¼ 8.5–33.1). Similarly, the incidence of vision loss related to MK was 3.6 per 10 000 patient-years (95% CI ¼ 0.4–12.9). There were a large number (56) of incidents of indeterminate etiology. It is possible that a few of the ‘indeterminate’ cases were infectious, but the Endpoint Adjudication Committee made the infectious/sterile distinction, in ways they could reproducibly identify (e.g. lesion size, anterior chamber reaction, etc. were utilized). We note that the estimated incidence of MK is highly dependent upon the definition used for the clinical condition. Because the definitions of MK used in a number of earlier studies (e.g. [17,18]) were somewhat more restrictive than the one used in the CIBA postapproval study [37], it is difficult to compare the different study rates directly. It is encouraging that the prospectively determined rate of vision loss in this postapproval study was low. However, it should also be noted that in the order of 1% of patients in any given year may be given antibiotic treatment for infiltrative keratitis. In summation, epidemiologic methods have proved to be of great value in the investigation of contact lens-related corneal ulcers. Past studies have had a significant influence upon public policy in the USA and the FDA continues to use population-based methods to help better evaluate the relevant risks. Future studies will further clarify the public health issues related to contact lens wear.

Conclusion Ophthalmic device technology is advancing rapidly. New types of intraocular lenses are being developed to correct astigmatism and near vision problems in patients who have had cataract surgery. Intraocular lenses have been developed for use solely for the correction of myopia, hyperopia, and astigmatism in patients who do not have cataracts. New types of contact lenses may greatly reduce the risk of lens-related infection or minimize discomfort. Additionally, refractive lasers, lasers that treat retinal disease, and

438

CH28

OPHTHALMIC DEVICES AND CLINICAL EPIDEMIOLOGY

glaucoma devices, have made tremendous advances in recent years. There is even work in progress on electronic devices whose aim is to restore vision in patients with poorly functioning or non-functioning retinas. Epidemiology has provided the FDA with a powerful methodology to aid in decisions concerning new device approvability and in the postapproval monitoring of device safety. Epidemiologic methods will continue to play an important role in the FDA’s evaluation of new ophthalmic technology and the protection of public health.

References 1. Thylefors B, Negrel AD, Pararajasegaram R, Dadzie KY. Global data on blindness. Bull World Health Org 1995; 73: 115–121. 2. Rahmani B, Tielsch JM, Katz J, Gottsch J et al. for the Baltimore Eye Survey. The cause-specific prevalence of visual impairment in an urban population. Ophthalmology 1996; 103: 1721–1726. 3. The Eye Diseases Prevalence Research Group. Prevalence of cataract and pseudophakia/aphakia Among Adults in the United States. Arch Ophthalmol 2004; 122: 487–494. 4. Leske MC, Chylak LT, Suh-Wuh W et al. The lens opacities case control study: risk factors for cataract. Arch Ophthalmol 1991; 109: 244. 5. The Italian–American Study Group: Risk factors for age-related cortical, nuclear, and PSC cataracts. Am J Epidemiol 1991; 133: 541. 6. Harding JJ, van Heyningen R. Epidemiology and risk factors for cataract. Eye 1987; 1: 537. 7. Trivedi R, Apple D et al. Sir Nicholas Harold Ridley. He changed the world, so that we might better see it. Indian J Ophthalmol 2003; 51: 211–216. 8. Olson RJ, Mamalis N et al. Cataract treatment in the beginning of the twenty-first century. Am J Ophthalmol 2003; 136: 146–154. 9. Report of the Lens and Cataract Panel, National Eye Institute: http://www.nei.nih.gov/resources/ strategicplans/neiplan/frm_lens.asp [accessed January 27 2006]. 10. Stark WJ, Worthen DM, Holladay JT et al. The FDA report on intraocular lenses. Ophthalmology 1983; 90: 311–317. 11. ISO 11979-7, 2001. Ophthalmic implants – Intraocular lenses – Part 7: clinical investigations. 12. Boam AB, Edelman MB, Lum FC et al. Retrospective evaluation of intraocular lenses in adults younger than 60 years. J Cataract Refract Surg 2003; 29: 575–587. 13. Lum F, Schein O, Chachat AP et al. Initial two years of experience with the AAO National Eyecare Outcome Network (NEON) cataract surgery database. Ophthalmology 2000; 107: 691–697. 14. Asuri MK, Venkata N, Kumar VM. Differential diagnosis of microbial keratitis and contact lensinduced peripheral ulcer. Eye Contact Lens 2993; 29: S60–S62. 15. Krachmer JH, Purcell JJ. Bacterial corneal ulcers in cosmetic soft contact lens wearers. Arch Ophthalmol 1978; 96: 57–61. 16. Adams CP Jr, Cohen EJ, Laibson PR, Galentine P, Arentsen JJ. Corneal ulcers in patients with cosmetic extended-wear contact lenses. Am J Ophthalmol 1983; 96: 705–709. 17. Poggio EC, Glynn RJ, Schein OD, Seddon JM et al. The incidence of ulcerative keratitis among users of daily-wear and extended-wear soft contact lenses. N Eng J Med 1989; 321: 779–783. 18. Schein OD, Glynn RJ, Poggio EC, Seddon JM, Kenyon KR. The microbial keratitis study group. The relative risk of ulcerative keratitis among users of daily-wear and extended-wear soft contact lenses. N Engl J Med 1989; 321: 773–778.

REFERENCES

439

19. Dart JKG, Stapleton F, Minassian D. Contact lenses and other risk factors in microbial keratitis. Lancet 1991; 338: 650–653. 20. Nillson SEG, Montan PG. The annualized incidence of contact lens induced keratitis in Sweden and its relation to lens type and wear schedule: results of a 3-month prospective study. CLAO J 1994; 20: 225–230. 21. Cheng KH, Leung SL, Hoekman HW, Beekhuis WH et al. Incidence of contact lens-associated microbial keratitis and its related morbidity. Lancet 1999; 354: 179–183. 22. Lam DSC, Houang E, Fan DSP, Lyon D et al., the Hong Kong Microbial Keratitis Group. Incidence and risk factors for microbial keratitis in Hong Kong: comparison with Europe and North America. Eye 2002; 16: 608–618. 23. Prepared statement of the Federal Trade Commission before the Committee on Energy and Commerce, United States House of Representatives, September 9 2003. FTC web site: http:// www.ftc.gov/os/2003/09/contactlens.htm#N_4 [accessed November 25 2004]. 24. Seal DV. Acanthamoeba keratitis update – incidence, molecular epidemiology and new drugs for treatment. Eye 2003; 17: 893–905. 25. Witschel H, Sundmacher R, Seitz HM. [Amebic keratitis: clinico-histopathologic case report]. Klin Monatsbl Augenheilkd 1984; 185: 46–49 [in German]. 26. Moore MB, McCulley JP, Luckenbach M, Gelender H et al. Acanthamoeba keratitis associated with soft contact lenses. Am J Ophthalmol 1985; 100: 396–403. 27. Stehr-Green JK. Acanthamoeba keratitis in soft contact lens wearers: a case-control study. J Am Med Assoc 1987; 258: 57–60. 28. Liesegang TJ. Contact lens-related microbial keratitis. Part II: pathophysiology. Cornea 1997; 16: 265–273. 29. Kramer MS. Clinical Epidemiology and Biostatistics. New York: Springer-Verlag, 1988; 9. 30. Kramer MS. Clinical Epidemiology and Biostatistics. New York: Springer-Verlag, 1988; 106. 31. Sokal RR, Rohlf FJ. Biometry. New York: WH Freeman, 1995; 762. 32. Saviola JF, Hilmantel G, Rosenthal AR. The U.S. Food and Drug Administration’s role in contact lens development and safety. Eye Contact Lens 2003; 29: S160–S165. 33. Holden BA, Reddy MK, Saukaridurg PR, Buddi R et al. Contact lens-induced peripheral ulcers with extended wear of disposable hydrogel lenses: histopathologic observations on the nature and type of corneal infiltrate. Cornea 1999; 18: 538–543. 34. Alexandrakis G, Alfonso EC, Miller D. Shifting trends in bacterial keratitis in south Florida and emerging resistance to fluoroquinolones. Ophthalmology 2000; 107: 1497–1502. 35. Donshik PC, Suchecki JK, Ehlers WH. Peripheral corneal infiltrates associated with contact lens wear. Trans Am Ophthalmol Soc 1995; 93: 49–60. 36. Brennan NA. Is there a question of safety with continuous wear? Clin Exp Optom 2002; 85: 127–140. 37. Schein OD, McNally JJ, Katz J, Chalmers RL et al. The incidence of microbial keratitis among wearers of a 30 day silicone hydrogel extended-wear contact lens. Ophthalmology 2005; 112: 2172–2179.

29 Orthopedic devices: epidemiologic considerations Ronald G. Kaczmarek, Michele G. Bonhomme, Stanley A. Brown, Judith U. Cope, and Daniel S. McGunagle US Food and Drug Administration, Rockville, MD, USA

Introduction Medical devices are a vital component of the practice of orthopedics. We begin by reviewing the characteristics and properties of materials used in these devices. The chapter focuses on two major orthopedic devices of particular importance: the hip implant (see Figure 29.1) and the artificial intervertebral disc. The underlying pathologies that lead to the use of the devices, operative risk, device effectiveness, complications, future trends, and cost considerations are discussed. In addition, we examine adverse event reporting for orthopedic devices and methodologic challenges faced by postmarket studies of the safety and effectiveness of orthopedic devices. These methodologic challenges arise from a diverse array of causes, including the very long-term studies required to fully examine the performance of many implanted orthopedic devices over their considerable expected lifespans.

Selected orthopedic materials Metals There are three primary alloy systems used in orthopedics: the iron-based stainless steels, the cobalt–chrome alloys and the titanium alloys. Perhaps the most critical property of Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

442

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

Figure 29.1 Artificial hip prosthesis, mid-twentieth century, made from cobalt alloy. Reproduced with permission from Science Museum/Science and Society Library

these is their high corrosion resistance, which results from the formation of stable metal oxide films on the surface. Both stainless steel and cobalt-based alloys are about 10 times stiffer than bone, and titanium is about five times as stiff as bone. The major alloying elements in the most common stainless steel, ‘316L’, are 17–19% chromium, 13–15% nickel, 2–3% molybdenum, and < 2% manganese. There are newer alloys which have less nickel, but more chromium and manganese. These are typically used in the wrought form, and their strength and ductility, which vary (range 12–40%), is controlled by the degree of cold working. Stainless steels have been used for all sorts of devices for fracture and spinal fixation. The original Charnley total hips were, and still are, made of stainless steel. The cobalt-based alloys are far more complicated, metallurgically, than stainless steel. The classic is a casting alloy of 27–30% chromium, 5–7% molybdenum and < 1% nickel. It has high corrosion resistance and was originally named ‘stellite’, as the star among the alloys. Variations of this casting formulation have resulted in a wrought cobalt– chromium–molybdenum alloy. Both these alloys have extremely high wear resistance, due to the formation of chromium carbides which makes them the material of choice for metal-on-metal joint prostheses. These alloys are also used for virtually all joint replacement prostheses, as well as for fracture fixation where device bending is not necessary. Enhanced attachment between prosthesis stem and bone can be accomplished by sintering cobalt alloy beads to the bone interface surfaces. There are also other cobalt alloys, some of which have significant amounts of nickel, some of which have higher strength than is necessary.

SELECTED ORTHOPEDIC MATERIALS

443

Titanium has been traditionally used as commercially pure (CPTi) or as an alloy with 6% aluminum and 4% vanadium. The CPTi is considered to be one of the most biocompatible metals, but its strength is increased with alloying. Newer alloys are formulated with aluminum and niobium, with niobium and zirconium, or with molybdenum, zirconium and iron, to achieve a variety of mechanical properties. Titanium does not have as good wear resistance as the other alloys, but it is often the material of choice in modular or mixed metal devices as the bone interface component.

Ceramics There are several types of ceramic materials used on orthopedics. They are typically high-strength, hard, but brittle materials. Aluminum oxide (alumina) and zirconium oxide (zirconia) have been used in recent years for the ball heads of total hip replacements. When stabilized with yttria or magnesia, zirconia is stronger than alumina, but not as hard. Alumina is the material of choice for ceramic-on-ceramic bearing combinations. The combination produces a very low-wear-rate prosthesis, but there are lingering concerns about brittle fracture. The modern processing techniques have resulted in fully dense, high-strength products.

Polymers The word ‘polymer’ means many mers, for example, polyethylene (PE) is many ethylene mers. The most common polymer used for total joint bearing is ultra-high molecular weight polyethylene (UHMWPE), which has a molecular weight of 2–4 million Daltons. It is used for articulation against all the metals and ceramics. UHMWPE is synthesized as a powder which can be molded into large blocks or extruded as rods, and then machined. It can also be injection-molded into the final product. UHMWPE melt distorts in an autoclave, so it must be irradiation-sterilized. In recent years it has been recognized that the irradiation can have two adverse effects: it breaks the molecule chains into shorter segments, which can then oxidize causing embrittlement; and it can cross-link the molecules for changes in properties. Shelf life has been improved by sterilization in inert atmosphere packaging. Recently, irradiation has been used under several controlled conditions to deliberately cross-link UHMWPE, resulting in improved wear resistance. Ever since Sir John Charnley implanted his first low-friction arthroplasty, polymethylmethacrylate (PMMA) ‘bone cement’ has been used to fill the space between the implant and bone for total joint replacement. It is not a cement in the adhesive sense, but is a filler like concrete. PMMA is sold as a kit, with liquid monomer in a vial and powdered polymer in a sack. They are mixed together and the chemicals in the two parts trigger the polymerization of the liquid, which glues the mass together in a solid block. Like PE, PMMA is a polymer with a carbon backbone. Unlike PE, which has two hydrogen atoms per carbon, the PMMA-mer has one methyl and one acrylic side group per carbon pair.

444

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

These make the material stronger and stiffer than UHMWPE, but also more brittle. Barium sulfate or zirconia is often added to make the cement radio-opaque. A variety of techniques are used with PMMA. Mixing may be done by hand, in a mixer, under a vacuum or centrifuged prior to use. It may be finger packed into the bone when in the dough stage, or injected under pressure with a large-diameter tube. It is also injected through a needle for vertebroplasty or kyphoplasty.

Artificial hips Risk factors for hip implant Osteoarthritis is the most important underlying condition that creates the need for an artificial hip implant. Pain is the primary symptom of this noninflammatory arthritis of the hip. The pain increases with exercise and is often diminished by rest. Patients may complain of stiffness after prolonged sitting. Radiographic studies may reveal narrowing of the hip joint space caused by loss of cartilage. The prevalence of osteoarthritis increases with age. Other risk factors for osteoarthritis include trauma and repetitive joint use. Obesity is a risk factor for osteoarthritis of the hip, but the magnitude of the risk is much lower than for osteoarthritis of the knee [1]. Body mass index [BMI] is directly associated with the risk of total hip replacement. For example, a study by Karlson et al. [2] found that higher BMI was significantly associated with an increased risk of hip replacement because of osteoarthritis (p for trend ¼ 0.0001). Women with BMI 35 were twice as likely as more lean women with a BMI < 22 to undergo hip replacement. In a study by Flugsrud et al. [3], 50 034 patients were followed for 9 years, and a dose–response relationship was found between BMI and total hip replacement for primary osteoarthritis. The highest quartile of BMI subjects had a relative risk of 2.0 (95% CI ¼ 1.4–2.9) among men, and 3.0 (95% ¼ CI 2.1–4.1) among women, compared in both cases to the lowest quartile of BMI subjects. Aseptic necrosis, also referred to as osteonecrosis or avascular necrosis, of the femoral head is another important underlying pathology that can result in the need for artificial hip replacement. The causes of aseptic necrosis are truly legion, including anemia, corticosteroid use, ethanol abuse, pancreatitis, and trauma. Many cases of aseptic necrosis are idiopathic as well [4]. Pavelka [5] has observed that the femoral head is a common site of osteonecrosis and that almost one-half of all cases of femoral head osteonecrosis require hip arthroplasty. The mean age of patients with aseptic necrosis of the femoral head is far lower than that of their counterparts with severe osteoarthritis of the hip. Aseptic necrosis patients are often near 40 years of age [6]. The life expectancy of such patients is substantially greater than the typical patient who receives an artificial hip secondary to osteoarthritis. Aseptic necrosis patients pose a greater clinical challenge, as they will be more likely to require revision of their artificial hip implant because of their generally younger age at initial implantation. Total hip arthroplasty can yield very substantial clinical dividends. These benefits were well quantified in a study by Bachmeier et al. [7]. A total of 86 patients with

ARTIFICIAL HIPS

445

osteoarthritis of the hip who received artificial hip arthroplasty were followed for 1 year. Dramatic improvements in function and quality of life were documented by the use of the Western Ontario and McMaster Universities Osteoarthritis Index, a well established and recognized tool. One year following hip replacement, the following observations were made: a 71% reduction in pain, a 55% reduction in stiffness, and a 68% increase in physical function. The study also employed the Medical Outcomes Study SF-36 Health Survey to assess the impact of the hip replacement. Again, the observed clinical changes were dramatic. The measures for pain improved 22%, indicating a decrease in pain, and there were improvements in physical function (247%), physical role functioning (402%), general health (110%), vitality (143%), social functioning (275%), and mental health (114%). An examination of the medical literature in the aggregate for evidence of improvement of the patient’s health-related quality of life following total hip arthroplasty was conducted by Ethgen et al. [8]. This examination included studies published in French or English during January 1980–June 2003 that utilized a prospective cohort study design and what the authors considered to be a validated self-reported quality of life measurement tool. The authors included 32 studies of both total hip and knee arthroplasty and 26 studies limited to total hip arthroplasty. They concluded that total hip arthroplasty was highly successful in improving health-related quality of life parameters. The failure of age to present a barrier to favorable total hip arthroplasty outcomes was another noteworthy conclusion.

Operative mortality The operative mortality risk from total hip arthroplasty is low. For example, a study by Parvizi et al. [9] included 30 714 consecutive patients who had received elective hip arthroplasty at the Mayo Clinic during 1969–1997. The 30 day mortality rate was 0.29%. Parvizi et al. also examined risk factors for 30 day mortality. Pre-existing cardiovascular disease (p < 0.0001), male gender (p < 0.001), and age 70 years (p < 0.0001) were all associated with increased 30 day mortality risk. They also reported that the 30 day mortality rate declined over time to only 0.15% by the 1990s. A report by Milleret al. [10] on the results of one surgeon who completed a total of 4967 hip arthroplasties during 1970–1996 further supports the contention that the mortality risk from hip arthroplasty is low. In-hospital mortality was 0.5% in the primary hip replacement group and 0.62% in the revision group. Patient age > 70 years was an identified risk factor for postoperative mortality risk (p < 0.001). Nunley and Lachiewicz [11] reported on the results of a single surgeon who performed 1108 total hip arthroplasties. Only 0.45% of the patients expired within 90 days of their surgery. An important factor affecting hip implant operative mortality is procedural volume (i.e. number of procedures performed), both for the physician and the healthcare facility. The effect of procedural volume on mortality risk has been observed for the implantation of a number of medical devices. For example, a nationally representative study by Astor et al. [12] found a statistically significant trend between hospital volume and the odds of

446

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

in-hospital mortality following aortic valve implantation. Particularly striking in the Astor et al. study was the finding that, for initial aortic valve replacements without other surgery, the highest quartile of hospital volume had an adjusted odds ratio of 0.39 (95% CI ¼ 0.23–0.64) compared to the lowest quartile. Procedural volume has also been found to be of importance in a number of studies of hip surgery. A study that employed Medicare data by Taylor et al. [13] found lower inhouse and 30 day mortality rates for total hip arthroplasty, partial hip arthroplasty, and revision total hip arthroplasty at relatively higher volume hospitals for hip implantation. A study conducted by Kreder et al. [14] of a statewide hospital discharge registry found a statistically significant relationship between low-volume surgeons and an adverse outcome [p < 0.05]. Losina et al. [15] employed Medicare claims data to analyze the experience of 57 488 Medicare beneficiaries who received primary total hip replacement. These patients were followed for a median of just less than 4 years, a total of 47 months. Hip revisions were less common when performed by high-volume surgeons, as opposed to low-volume surgeons. This persisted when controlling for hospital volume.

Device survival A study by Franklin et al. [16] is illustrative of studies of artificial hips that were performed at a solitary medical facility. These studies often extend retrospectively decades in time in order to generate a sufficient sample size. During the course of the study, important variables, such as surgical technique, implant composition, and/or prophylactic anti-thrombotic medical regimen, may change substantially. In addition, findings from single-center studies may not reflect general real-world experience. The hospital records study of Franklin et al. [16] included 654 primary hip replacements that were performed in 548 patients during 1982–1999 at the FSACentral Hospital in Akureyri, Iceland. These patients were followed until the end of 2001. The majority of the patients were elderly, with a mean age of 68.4 years for male patients and 68.8 years for their female counterparts. For the purposes of the study, the exchange of one or both of the prosthetic components or removal of the implant was considered to comprise implant revision. Only 37 of the 654 (5.7%) hip implants underwent revision during the course of the study. The causes of revision were as follows: aseptic loosening, 28 cases; recurrent dislocations, 7 cases; and infection, 2 cases. For total hip replacements where osteoarthritis was the indication, the cumulative revision rate at 10 years was 8%, and 13% at 16 years. The risk of revision was not related to patient gender. Increasing patient age was associated with a decrease in revision risk. A noteworthy strength of the Franklin et al. study was a very high degree of patient follow-up. The relative isolation of Iceland may have contributed to the very high rate of follow-up. The authors claimed that no patients were lost to follow-up during the study, an admirable and rarely achieved accomplishment. Significant changes in both implant composition and clinical procedures occurred during the protracted study. During the course of the study, the surface finish of the stem of the artificial hip implant changed from matte to polished, and the heparin prophylaxis protocol was modified in 1992.

ARTIFICIAL HIPS

447

In addition, medical practice in Europe and other foreign lands may differ from the USA and thus results may differ. In sharp contrast to data from a single clinical center, national registry data can provide data from numerous clinical centers, which is therefore more representative of the general population of patients. The Danish Hip Arthroplasty Register is an example of a national registry [17]. Over 18 000 primary total hip replacements were entered into the registry in only 4 years of operation. The registry data showed 97% hip implant survival after 3 years. Another example of a national registry is the Swedish National Total Hip Arthroplasty Register. Soderman et al. [18] studied a randomly selected cohort of patients who received total hip replacement during 1986–1995 and completed a questionnaire. These patients were also evaluated clinically with the Harris Hip Score. Failure consisted of either implant revision or a Harris Hip Score of < 60 points. The observed 10 year survival of the hip implants (i.e. absence of ‘failure’) was 87%. The authors also used the Western Ontario and McMaster University Osteoarthritis Index (WOMAC) to evaluate hip implant survival. The WOMAC assesses patient pain, stiffness and ability to perform an array of activities. The observed 10 year hip implant survival was 80% with the WOMAC as the clinical measurement tool. Implant wear rates are greater in younger than in elderly patients [19]. Very long-term survival of hip implants is more important for younger recipients of such implants, as they have longer life expectancies. A study by Callaghan et al. [20] addressed this concern. A total of 330 cemented Charnley hip replacements were implanted in 262 patients by the same physician. Fully 51 of the patients with a total of 62 hip implants survived 25 years after their surgery. The majority (77%) of these patients did not require revision of their hip implants. The most common indication for revision was aseptic loosening of the implant; it occurred in 11 (18%) of the hip implants. These results indicate that even an early generation of hip implants could function adequately for very extended periods. These results are encouraging for younger artificial hip implant recipients. Patient activity affects implant survival. The type of activity is an important consideration. Activities without high levels of impact loading, such as walking, swimming and golf, are recommended after hip arthroplasty. Activities that create high levels of impact loading, such as basketball, hockey, and football, are generally not recommended because of their potentially adverse effect on implant survival [21].

Complications Thigh pain is a complication of interest following hip implantation. The severity is often mild and may resolve without any medical intervention. The reported incidence of thigh pain following hip implantation is 0.5–40% [22]. Implant revision is required for thigh pain only when nonsurgical measures, such as nonsteroidal antiinflammatory medications, fail. A crucial complication following hip implantation is thromboembolism. Anticoagulation, often with warfarin, plays a vital role in the prevention of thromboembolic events

448

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

following artificial hip implantation. The use of thromboprophylaxis was quantified from data from the Hip and Knee Registry in a study by Anderson et al. [23] This registry received data from 464 orthopedic surgeons from 319 hospitals. Thromboprophylaxis use was as follows in patients who received primary total hip arthroplasty: elastic stockings, 61%; warfarin, 56%; low-molecular weight heparin, 38%; intermittent pneumatic compression, 35%; aspirin, 4%. The mortality rate from thromboembolism today following hip implantation is low. Wroblewski et al. [24] found that only 1 of 1294 patients who underwent total hip arthroplasty died from a pulmonary embolism. Implant retrieval analysis is valuable for numerous medical devices, including artificial hips, in determining failure mechanisms. Hirakawa et al. [25] observed that particulate wear debris creates the formation of granulomatous tissue. This granulomatous tissue may eventually lead to aseptic loosening of the hip implant. Clinical studies have documented that aseptic loosening is a major cause of revision surgery following total hip arthroplasty. For example, a study by Clohisy et al. [26] analyzed 439 revision hip surgeries. By far the most common reason for the revision surgery was aseptic loosening (55% of the surgeries). Heterotopic ossification, bone growth beyond the normal limits of bone, is another complication of artificial hip implantation. A study by Lieberman et al. [27] included 161 patients who received 184 total hip arthroplasties. Radiographs were obtained periodically for a year following operation. A greater incidence of heterotopic ossification was observed among cemented hip implants (22%) than among non-cemented hip implants (9%; p < 0.05). It was noteworthy that Grade 4 heterotopic ossification was associated with a lower Harris Hip Score (p < 0.05). The Harris Hip Score addresses both pain and function considerations. Concerns have been raised regarding the possible relationship between joint replacement and the subsequent incidence of cancer. Visuri et al. [28] examined the risk of cancer in Nordic recipients of total hip and total knee arthroplasty. Their review was a meta-analysis of studies that included 24 000 total knee arthroplasty recipients and 49 000 total hip arthroplasty recipients. The patients were followed for a mean of 6.8 years. The standardized incidence ratio (SIR) for the number of cancers at all sites was 0.93 (95% CI ¼ 0.91–0.95). Site-specific and overall SIRs in total knee arthroplasty patients and total hip arthroplasty patients were comparable. These data do not support a link between total knee arthroplasty or total hip arthroplasty and the subsequent development of cancer.

Adverse event reports on artificial hips Medical device adverse events are reported to the US Food and Drug Administration through the Medical Device Reporting system. The reports received are commonly referred to as ‘Medical Device Reports’ or ‘MDRs’. A more detailed discussion of the MDR system can be found in Chapter 2. Figure 29.2 displays the number of MDRs for artificial hips received each year during January 1 1992–December 31 2004. The large jump in the number of MDRs received in

449 1549

668 620

781 534

880

793

466

327

03

02

01

00

99

04 20

20

20

20

20

97

96

95

94

98

19

19

19

19

19

93

111 118 119

19

19

92

1800 1600 1400 1200 1000 800 600 400 200 57 0

19

Number of hip MDR reports received by Year

ARTIFICIAL HIPS

Year Figure 29.2 Number of hip MDR reports received, January 1 1992–December 31 2004 (n ¼ 7023)

2001 deserves explanation and provides an example of the limitations of the reporting system. In 2001, 1549 MDRs were received for all artificial hips. Of these, 815 (52%) were reports for Sulzer InterOp acetabular cups. This sudden change in the number of MDRs received is an example of how one might expect the reporting system to work to alert FDA to problems. The manufacturer became aware that there was a problem with the device in the fourth quarter of 2000, and reacted by pulling the product off the market in early 2001. However, the reports database did not fully reflect the scale of the problem until well into the year 2001. The pattern of hip MDR reports over the 2000–2002 time period shows that a lag exists between the onset of a problem with a medical device and FDA becoming aware of the problem through the reporting system. The largest part of the lag is driven by delays in individuals and user facilities reporting adverse events to manufacturers and FDA. The lag is enlarged by delays on the part of manufacturers, in gathering further data, before they in turn report to FDA. A second problem revealed by Figure 29.2 is the small number of hip MDRs received each year, considering the large number of devices reported in the literature as failing every year. Kurtz et al. [29] report that in the years 1997–2002 over 30 000 revisions of total hips were performed in the USA. From Figure 29.2, one can see that in no year has FDA received even one-tenth of this number of reports. Underreporting of adverse events has limited, and continues to limit, the utility of FDA’s adverse event database by making it difficult to assess whether observed events are in excess of what is expected. Figure 29.3 displays the distribution of hip MDRs received during January 1 1992– December 31 2004, based on device characteristics. The most important differences in the design of artificial hips are differing materials that make up the gliding, i.e. ‘bearing’, surface and differences in the method of attaching the total hip components to the bone. As shown in Figure 29.3, 3748 (53%) of the hip MDRs involved metal on polyethylene bearing hip designs, fixed with bone cement. An

450 Number of MDR Reports Jan 1, 1992–Dec 31, 2004

CH29 4000 3500 3000 2500 2000 1500 1000 500 0

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS 3748 2765

Metal on polyethylene bearing, cemented components

Metal on polyethylene bearing, uncemented components

205

34

Ceramic on polyethylene bearing, uncemented or cemented

Ceramic on ceramic bearing, cemented or uncemented

271 Other Hip component configurations

Device description

Figure 29.3 Distribution of hip MDR reports by device type (n ¼ 7023)

additional 2765 (39%) MDRs reports described cementless fixation of metal femoral balls bearing on polyethylene. This second group includes adverse event reports for the Sulzer InterOp Acetabular shell, discussed above. Ceramic femoral head bearing on polyethylene and on ceramic acetabular liners make up a very small proportion of the 7023 total MDRs.

Future trends The incidence of total hip arthroplasty has increased over time. A study by Kurtz et al. [29] employed data from the National Hospital Discharge Summary (NHDS) to examine the rate of hip arthroplasty in the USA. The NHDS receives data from more than 400 hospitals. Based on NHDS data, the annual number of primary total hip arthroplasties in the USA increased from 119 000 in 1990 to 193 000 in 2002, representing a 62% increase in this time span. The incidence of total hip arthroplasty has increased over time. The prevalence of hip implants is far greater in elderly individuals than in younger individuals. Sharkness et al. [30], in a nationally representative study that included 47 485 households and 122 310 individuals, found that the prevalence of hip implants among individuals aged 18–44 years was only 0.3 per 1000. In contrast, the hip implant prevalence rate was 11.3 per 1000 among individuals aged 65–69 years and 23.9 per 1000 among individuals aged 75 years and older. Hip implants can be employed in very elderly patients. A study by Pagnano et al. [31] found a significant increase in Harris Hip Scores postoperatively among 65 patients with a mean age of 92 years and a minimum age of 90 years.The elderly proportion of the population continues to increase on a worldwide basis. These respective facts are a recipe for an increase in total hip arthroplasty rates. A study by Ostendorf et al. [32] found that the number of total hip replacements in Sweden increased by 20% from 1986 to 1997. The corresponding increase in The Netherlands was 68%. Wells et al. [33] noted that in

ARTIFICIAL HIPS

451

Australia the incidence of primary total hip arthroplasty increased by 19.6% from 1994 to 1998. The demand for total hip arthroplasty is expected to increase markedly in the coming decades. Much of the increased demand will be driven by an aging population. For example, Ostendorf et al. [32] have estimated that the number of total hip replacements in Sweden will increase by at least 25% by the year 2020 and by almost 50% in The Netherlands. The increased demand will place a burden on healthcare budgets. The USA can also anticipate an increase in future hip arthroplasty rates. The aging of the baby boom generation will double the number of elderly Americans over the next few decades. In addition to the aging factor, the increasing prevalence of obesity in the USA will likely contribute to a rise in hip arthroplasty rates. The Centers for Disease Control and Prevention (CDC: http://www.cdc.gov/nccdphp/dnpa/obesity/ trend/prev_ char.htm) [34] reports that obesity among US adults increased by 74.2% during 1991–2001. Overall, the CDC has observed that 44 million Americans are classified as obese. Given the relationship between obesity and the risk of artificial hip replacement noted earlier, the rise in obesity among Americans should increase the future need for hip replacement surgery. Equitable access to the considerable benefits of total hip arthroplasty is a worthwhile goal. Sharkness et al. [30] found that the prevalence of hip implants among whites was more than two-fold greater than the corresponding prevalence among blacks, raising concerns that blacks may lack equal access to this procedure. Greater family income was not associated with hip implant prevalence. Near-universal access to Medicare after age 65, coupled with the preponderance of hip implantation incidence occurring in the elderly, may have contributed to this finding. Studies outside the USA have also questioned the equality of access to hip arthroplasty [35]. The annual cost of joint replacement is substantial. Utilizing data from the National Discharge Survey, Kim et al. [36] estimated that expenditures in the USA for total hip replacements and total/partial knee replacement procedures were roughly $9.9 billion. There were 439 833 hip and knee procedures identified. The cost per procedure was in excess of $22 000. An increased demand for hip replacements will further increase the cost burden at a time when the healthcare system is displaying increasing signs of financial strain. Given the relatively greater rate of hip replacement in elderly individuals, the magnitude of the impact on the Medicare system will be particularly great. The Medicare system is believed to have a long-term funding deficit that is greater than the shortfall faced by the Social Security system, and this shortfall will be exacerbated by the increased demand for artificial hip implants. Funding limitations could potentially adversely impact access to hip implants if available resources are inadequate to fully meet the demand for artificial hip implants. In summary, the artificial hip is an extraordinarily valuable medical device that relieves crippling pain and restores mobility for hundreds of thousands of patients worldwide. The device can be implanted with a low operative mortality risk. Important complications include aseptic loosening, postoperative pulmonary embolism, and heterotopic bone ossification. There is strong reason to believe that the utilization of this device will increase sharply over the next two decades.

452

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

Intervertebral disc replacement Risk factors for spinal surgery Most spinal implant surgeries are performed to treat degenerative disease that occurs with aging, namely osteoarthritis. Osteoarthritis of the spine is more complicated than hip osteoarthritis. Very different from the simple ball joint of the hip, the complex structures of the back include the articulating surfaces of the vertebrae (facet joints) and the intervertebral discs. Although the discs are not true joints, they provide cushions between the flat surfaces of the vertebrae, thus allowing motion and creating mechanical strength for the spine to handle compressive forces as the body moves. The degenerative changes of osteoarthritis occur in the vertebral facet joints (spondylosis) as well as in the intervertebral discs (degenerative disc disease). The formation of osteophytes around the vertebral canal results in narrowing of the vertebral foramina with compression of nerve roots, causing neurological signs and symptoms such as muscle weakness and pain. Degenerative disc disease leads to disc herniation and spinal instability. Pain and instability that remain unresponsive to medical therapies are the primary indications for surgery. However, although osteoarthritis is a prevalent disease in older adults it is often asymptomatic, and other causes of back instability, malalignment and pain are also common in the elderly. Therefore, it is important to fully evaluate patients with these problems to determine whether they relate to osteoarthritis or are due to other underlying disease. Back pain itself is complex, affected not only by biomechanical factors but psychological ones as well. More so than with the hip, the timing and clinical indications for spinal surgeries, and determining which patients will have favorable outcomes, is often unclear. The criteria for diagnosis may be unreliable, and some question whether the outcomes of surgery are adequate to justify the risks and costs [37,38,39]. In the USA during the mid-1990s the estimated annual direct medical cost for low back pain treatments was estimated to be $33 billion [40].

Epidemiology of spinal osteoarthritis Persons with degenerative disease of the spine may have some of the same risk factors that are seen with hip osteoarthritis [41,42]. Although few epidemiologic studies have been conducted to study spinal osteoarthritis per se, major factors appear to be older age, genetic inheritance and trauma [43]. A British cohort study of spinal osteoarthritis in women examined radiological findings of anterior vertebral osteophytosis and disc space narrowing separately and found them to be associated with age, back pain, and radiographic evidence of hip and knee osteoarthritis [44]. Other studies have examined risk factors for low back pain, but not specifically in association with degenerative disc disease [45,46]. Radiographic evidence of degenerative disc disease has a higher prevalence among men than women of the same age [47]. However, other investigators found that vertebral and facet joint osteoarthritis resulted in an increased spinal motion and instability that was more prevalent in women than in men [48]. Factors such as body

INTERVERTEBRAL DISC REPLACEMENT

453

mass index (BMI), occupational exposures and heavy physical activity continue to be controversial with respect to their potential causal role in spinal osteoarthritis [49,50].

Treatments for lumbar disc degeneration While other levels of the spine may present problems, the main reason why patients seek surgical treatment is for low back pain in the lumbar and lumbosacral areas. Painful lumbar disc degeneration is one of the most common indications for surgery. The pain of degenerative joint disease is linked to mobility. While pain is alleviated by surgery to suppress motion by spinal fusion (arthrodesis), it is at the cost of impaired function. Arthrodesis is the currently accepted ‘gold standard’ surgical treatment for lumbar degenerative disc disease when nonoperative therapies fail. There are a variety of approaches to arthrodesis, including: anterior, posterior and posterolateral surgical approaches and the use of spinal fixation devices, with and without pedicle screw instrumentation and interbody fusion cages, and with or without bone graft, cement or bone substitutes to fixate the treatments. The various implants, pedicle screws, and fixation devices are used to remove loose pieces of bone and cartilage, resurface the bone, and reposition or stabilize the disc. The ultimate goal is to relieve the pain and disability by stabilizing spinal structures and alignment, while maintaining disc height. While arthrodesis often successfully relieves pain with an acceptably low rate of complications, it can also cause mechanical stress that may lead to degeneration of adjacent vertebral levels and joints, with consequent motion problems. There are also limitations with respect to restored function. The results of arthrodesis are significantly worse for patients who have had prior spinal surgeries [51]. Overall, despite almost 80% improvement in the short-term follow-up for large case series [52–54], a longer-term follow-up of patients up to 10 years demonstrated high rates of pain, medication use, and further surgery [55,56]. A systematic review of 47 different nonrandomized studies revealed that roughly two-thirds of patients had successful (usually defined as absence of reoperation, removal, or revision) spinal fusion (range 16–95% of patients). The two most common complications were failed fusion (i.e. pseudoarthrosis, 14%) and chronic pain at the bone graft donor site (9%) [57].

Intervertebral disc replacement More recently a variety of arthroplasty techniques have been developed as alternatives to spinal fusion. This discussion will focus on total intervertebral disc replacement, which has become increasingly popular. Its purpose is to relieve nerve root irritation or alleviate the compression caused by the materials of a herniated disc, as well as to restore disc height in order to protect neural elements, thereby providing pain relief and retained spinal motion (to prevent posterior facet arthropathy and adjacent segment disease) [58], and restoring function. Intervertebral disc replacement offers some advantages over arthrodesis. Because of the small incision site and reduced operative time, blood loss, hospital stay, recovery

454

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

time, and rehabilitative time are reduced, and there is likely to be less postoperative pain and other operative complications. It aims to decrease problems with adjacent vertebral segment degeneration, avoids complications with more invasive fusion surgeries, and allows earlier return to activities. This may be ideal for patients who are elderly, with comorbid conditions, or who are younger and more active.

Assessment of the safety and effectiveness of intervertebral disc replacement It is crucial to conduct long-term follow-up studies to see whether, over time, artificial discs hold up and meet the primary goals of good motion and fewer problems with adjacent segment degeneration [59]. Even 2 year results need to be interpreted with caution: Historically, hip arthroplasty results seemed very encouraging early on. However, 10 year follow-up of patients demonstrated disappointing outcomes, particularly for adults under 55 years of age, who had a surgical revision rate of 28% [60]. Most importantly, no randomized clinical controlled trials have evaluated the longterm results of intervertebral disc replacement [61]. Its success needs to be evaluated and compared with the clinical outcomes for spinal arthrodesis, especially perioperative complications, implant survivability, and measures of patient improvement in pain, disability, and activity level. In addition it is important to understand the acceptability and safety of subsequent salvage procedures by examining the rates of complications for patients who require further surgery and/or device removal. As with the hip, a primary end point of interest will be implant survivability, i.e. the absence of failures requiring reoperation, revision, or surgical removal of the disc. There may be disc extrusion or explantation, and wear particles or host inflammatory response or reaction may occur. Currently there are insufficient data to adequately assess the longterm performance of total disc arthroplasty. Review of the literature revealed a case series study of 50 patients who received the Charite´ III disc device [62]. It found that 24 (48%) of the patients required reoperation, with the majority having complications at the site of the prosthesis or adjacent vertebral sites. Another study that followed a case series of 27 patients for a mean time period of 53 months found problems with polyethylene wear, subsidence of the prosthesis, and adjacent vertebral deterioration [63]. Metal sensitivity and immune reaction to materials of the artificial discs have been understudied [64,65]. The carcinogenic potential of orthopedic implants has been an area of concern, partly as a result of animal studies [66]. While there appears to be no concern for long-term risk of excess cancer for patients with hip arthroplasty, the occurrence of cancer among persons receiving spinal implant devices has not been studied [67,68]. Another important endpoint is the clinical outcome for the patient. Full assessment of pain, neurological impairment, disability, activity level, and patient satisfaction should be performed. Standard measurement tools for patient evaluation of persons with spinal surgeries include the Oswestry disability index (ODI), short-form health-related quality of life questionnaire (SF-36), and pain assessment using the Visual Analogue Score

INTERVERTEBRAL DISC REPLACEMENT

455

(VAS). Additional information before and after surgery is collected regarding employment and activity level, clinical neurological examination findings, and measurements of vertebral range of motion with flexion/extension/rotation. Data about hospitalizations and use of medications (especially pain and antiinflammatory medications) also are important in demonstrating the effect of the surgery on the clinical status of the patient. In addition to the need for better understanding of the overall long-term success of disc implant surgery, we need a better understanding of what population subgroups will likely benefit or fail. Evaluation of patient factors should include age, underlying conditions, prior back surgery, duration of back pain, occupation, activity level, smoking status, body mass index, and severity of disease. These all may have an effect on the survivability of the device and clinical patient outcome. Degenerative disease factors, such as trauma, osteoporosis, osteoarthritis, and cancer, will alter the success rates and rates of complications. It seems likely that the numbers of complications would increase with longer follow-up time, as was seen in total hip arthroplasty [60]. In addition, the type of surgical approach, e.g. percutaneous or open procedure, and number of levels of operated discs have also been shown to be important predictive factors in various spinal treatments. Several studies suggest that multilevel disc procedures have worse clinical outcomes, especially with respect to disability and activity level. One study revealed that college athletes who underwent two-level disc surgery had higher rates of drop-out for persistent back symptoms than those with one-level procedures [69]. Another study of elderly patients found worse results for those who had more than one-level surgery [70]. Similar results are emerging for total disc replacement, as found in some short-term studies [71]. Long-term prospective studies will be needed to adequately assess whether artificial disc replacement will lead to many of the same complications and adjacent vertebral problems that have been noted with spinal arthrodesis. Radiographic assessment provides data regarding device loosening, subsidence, displacement or migration, presence of radiolucencies (representing separation), disc extrusion, disc height, nearby soft tissue reaction and evaluation of adjacent level degeneration. Diagnostic criteria to define degenerative disc disease is not standard, and while X-rays and CT scans may be used in evaluations, MRIs appear to offer more accurate assessments of disc disorders and have become the gold standard for spine imaging. However, there is no consensus about possible prognostic radiological indicators [72,73]. Discography can be a very painful procedure, with variable results. It has problems with accuracy and precision and many centers report high false positive rates of disease [74]. Many experts feel it is seldom helpful prior to surgery. It is also important to assess specific surgical risks. Overall rates of perioperative complications of total disc replacement may vary from 2% to 10%, depending on the study [62,75,76]. The anterior placement of an artificial disc is highly technical surgery that may pose several safety concerns. Vascular complications are rare (< 0.17%), but potentially fatal [77,78]. These may include bleeding, phlebitis, emboli, and acute leg ischemia. They may occur because the surgical approach is anterior to the spinal column in close proximity to the great vessels. To clear the degenerative disc area for insertion, retracting surgical instruments are placed under the bifurcation of the great vessels (aorta and iliac arteries and veins). This calls for specialized training.

456

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

Perioperative complications vary according to the surgeon’s experience. One study that looked at complication rates according to the surgeon’s experience found that the most experienced surgeon’s complication rate (2.2%) was statistically different from that of less experienced surgeons (10.7%) [79]. Possible gastrointestinal injuries include peritoneal or intestinal tears leading to bowel perforation, peritoneal scarring, and adhesions. Genitourinary complications such as retrograde ejaculation, incontinence, or ureter damage might occur. Peripheral nerve damage, spinal cord injuries, dural tear, epidural hematoma, or herniated nucleus pulposus may occur [80]. Postoperative complications include deep vein thrombosis, pulmonary embolus and infection. Late complications may include pain and leg edema.

Data sources and selected methodological issues to consider in epidemiologic studies of orthopedic medical devices Epidemiologic research has provided much of the clinical trial data used to evaluate the efficacy and safety of orthopedic devices and to build the body of scientific knowledgewe have about these devices, but the full potential of the clinical trial and other epidemiologic research designs to expand our knowledge of the long-term effectiveness of orthopedic devices and improve clinical practice has yet to be realized. This section describes data sources available for use in epidemiologic studies of orthopedic devices and some of the methodological issues researchers should consider in conducting future epidemiologic studies to address the information needs and fill the data gaps in our current understanding of the safety, efficacy, and effectiveness of orthopedic devices. This section is intended to complement the discussion presented of methodological issues common to many epidemiologic studies of medical devices.

Data sources for epidemiologic studies of orthopedic medical devices The goals of epidemiologic research on orthopedic medical devices are to improve patient care by: (a) estimating the incidence of the health conditions that lead to the use of important orthopedic devices, such as total hip replacement; (b) identifying the risk factors for the health conditions that are indications for orthopedic procedures; (c) estimating the long-term effectiveness of orthopedic devices that improve health outcomes beyond the benefits conventional medical treatment offer; (d) identifying, estimating, and monitoring over time the prevalence and incidence of adverse events that occur as a result of exposure to specific orthopedic devices; (e) identifying the risk factors for complications and other adverse events associated with the use of orthopedic devices; and (f) measuring the effectiveness of interventions that aim to reduce the risk and occurrence of adverse effects associated with the use of orthopedic devices. Epidemiologic data needed to address these goals are often not available, or if available, information about the patient, his/her medical device exposure and other relevant factors is incomplete or not sufficiently detailed.

DATA SOURCES AND SELECTED METHODOLOGICAL ISSUES

457

Torrence [81] identified several government-sponsored and private data sources that can be used in epidemiologic studies of medical devices and described the strengths and weaknesses of each source. The Agency for Healthcare Research and Quality’s Nationwide Inpatient Sample (NIS), with data from over 7 million discharge summaries and 900 hospitals, is of value, particularly due to the ability to generate nationally representative estimates. The National Health Interview Survey, managed by the Centers for Disease Control and Prevention’s (CDC) National Center for Health Statistics, includes data on orthopedic implants. The CDC’s National Nosocomial Infections Surveillance System has data on joint prostheses. Centers for Medicare and Medicaid Services’ Medicare data provide useful claims data regarding orthopedic device implants. Outside the USA, valuable orthopedic implant registries have been established [17,18]. Analytic epidemiologic studies designed specifically to answer questions about device performance and safety and surveillance systems, and databases designed specifically to collect data needed to monitor relevant health outcomes and conduct analytic studies, offer the best hope for achieving the objectives of orthopedic device epidemiology. National and regional hospital hip replacement registries provide the best available orthopedic device exposure data. Registry data can be linked to and augmented by other sources, including insurance claims data, data extracted from medical records, and laboratory records.

Methodological issues to consider in orthopedic device epidemiology Since the development of specialized databases and studies and the linkage of records from disparate sources often requires a substantial investment of time and other resources, the prudent researcher’s careful consideration of the methodological issues described below, before undertaking a study, will maximize the return on his/her investment. Artificial hips and artificial intervertebral disks represent different stages in the evolution of an implanted orthopedic device: orthopedists have extensive experience with hip prostheses, but artificial intervertebral disks are a novel and emerging technology with which the experience has been much shorter. These two devices are used here to illustrate the range of methodological issues that may arise in epidemiologic studies of orthopedic devices.

Study design Research objectives and the analytic design best suited to meet the study questions should guide the choice of study design. In the USA and in many countries throughout the world, the double-blind randomized clinical trial is the gold standard of medical evidence. Although ideal, for ethical reasons the design features of the double-blind randomized clinical trial are rarely, if ever, feasible in clinical trials of orthopedic devices. Random allocation of eligible patients selected from the same population increase the similarities between the treatment and comparison group on known and

458

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

unknown characteristics that may confound the results. The factors that make a novel hip prosthesis an attractive alternative to patients who are eligible for inclusion in the treatment group of a premarket clinical trial also make it difficult to find an appropriate control group for these patients. Candidates for the new device either fared poorly on hip prostheses that are already available commercially or they were not good candidates for those devices. Given the clinical presentation of these patients, a control group comprised of patients who can use existing hip prostheses successfully would be inappropriate. Historical controls and the efficacy rates and complication rates observed for a comparable device are frequently used as the comparison group for patients treated with the novel hip prosthesis. It would be worthwhile to consider randomly allocating patients eligible to benefit from a novel device either to the new device group or to a group that would receive the best available non-surgical medical treatment. Such a comparison better reflects the treatment choices available to the treating physician than a trial comparing the treatment group to historical controls. Many of the published articles on effectiveness and safety are case series describing the experience of a hospital’s or a physician’s patients who underwent hip replacement. These studies may be retrospective or prospective and may range in duration from 1 year as many as 20 or more years. Many of the studies which are identified as long-term follow-up studies of hip arthroplasty lack a control group. The device survival rate and complication rates are reported for the period of observation without reference to a comparison group. More classic observational epidemiologic study designs are being used with increasing frequency in orthopedic device epidemiology. Cross-sectional studies have provided estimates of prevalence of orthopedic device use and of osteoarthritis. The cohort study is the design preferred for studying the long-term effectiveness and safety profile of orthopedic devices. This design makes it easier to evaluate bias and to validate exposure history. A few long-term follow-up studies of hip prostheses have been reported. Additional well-designed longitudinal cohort studies with sufficient power to detect rare, but often important, complications of hip replacement are warranted. Postmarket surveillance must be done continuously to assess trends in adverse event reporting over time. The case-control design has been applied successfully to study the therapeutic effect of anticoagulation on the risk of postsurgical deep venous thrombosis in patients who recently underwent hip replacement. Poor documentation of device exposure in medical records makes it difficult to conduct case-control studies of the relationship between artificial hips and specific complications. Ecological studies have little to contribute to safety evaluation.

Summary Orthopedic medical devices can relieve pain and restore function for millions of patients. Advances in materials have improved the durability and enhanced the clinical utility of

REFERENCES

459

many of these devices. The aging of the American and world populations will increase the future demand for these invaluable devices.

References 1. March LM, Bagga H. Epidemiology of osteoarthritis in Australia. Medical Journal of Australia 2004; 180(5 suppl): S6–S10. 2. Karlson EW, Mandl LA, Aweh GN et al. Total hip replacement due to osteoarthritis; the importance of age, obesity, and other modifiable factors. Am J Med 2003; 114: 155–159. 3. Flugsrud GB, Nordsletten L, Espehaug B, Havelin LI, Meyer HE. Risk factors for total hip replacement due to primary osteoarthritis: a cohort study in 50 034 persons. Arthritis Rheum 2002; 46: 675–682. 4. Pavelka K. Osteonecrosis. Baillie`res Best Prac Res Clin Rheumatol 2000; 14: 399–414. 5. Assouline D, Chang Y, Greenspan A, Shoenfeld Y, Gershwin ME. Pathogenesis and natural history of osteonecrosis. Semin Arthritis Rheum 2002; 32: 94–129. 6. Lavernia CJ, Sierra RJ, Grieco FR. Osteonecrosis of the femoral head. J Am Acad Orthop Surg 1999; 7(4): 250–261. 7. Bachmeier CJ, March LM, Cross MJ et al. A comparison of outcomes in osteoarthritis patients undergoing total hip and knee replacement surgery. Osteoarthritis Cartilage 2001; 9: 137–146. 8. Ethgen O, Bruyere O, Richy F, Dardennes C, Reginster JY. Health-related quality of life in total hip and total knee arthroplasty. J Bone Joint Surg 2004; 86A: 963–973. 9. Parvizi J, Johnson BG, Rowland C, Ereth NH, Lewallen DG. Thirty day mortality after elective total hip arthroplasty. J Bone Joint Surg Am 2001; 83: 1524–1528. 10. Miller KA, Callaghan JJ, Goetz DD, Johnston RC. Early postoperative mortality following total hip arthroplasty in a community setting: a single surgeon experience. Iowa Orthop J 2003; 23: 36–42. 11. Nunley RM, Lachiewicz PF. Mortality after total hip and knee arthroplasty in a medium volume university practice. J Arthroplasty 2003; 18: 278–285. 12. Astor BC, Kaczmarek RG, Hefflin BJ, Daley WR. Mortality after aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000; 70: 1939–1945. 13. Taylor HD, Dennis DA, Crane HS. Relationship between mortality rates and hospital patient volume for Medicare patients undergoing major orthopedic surgery of the hip, knee, spine, and femur. J Arthroplasty 1997; 12: 235–242. 14. Kreder HJ, Deyo RA, Koepsell T, Swiontkowski MF, Kreuter W. Relationship between the volume of total hip replacements performed by providers and the rates of postoperative complications in the state of Washington. J Bone Joint Surg Am 1997; 79: 485–494. 15. Losina E, Barrett J, Mahomed NN, Baron JA, Katz JN. Early failures of total hip replacement; effect of surgeon volume. Arthritis Rheum 2004; 50: 1338–1343. 16. Franklin J, Robertsson O, Gestsson J, Lohmander LS, Ingvarsson T. Revision and complication rates in 654 Exeter total hip replacements, with a maximum follow-up of 20 years. Musculoskeletal Disorders 2003; 4(1): 6. 17. Lucht U. The Danish Hip Arthroplasty Register. Acta Orthop Scand 2000; 71: 433–439. 18. Soderman P, Malchau H, Herberts P et al. Outcone after total hip arthroplasty, Part II. Diseasespecific follow-up and the Swedish National Total Hip Arthroplasty Register. Acta Ortop Scand 2001; 72: 113–119.

460

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

19. Torchia ME, Klassen RA, Biancio AJ. Total hip arthroplasty with cement in patients less than twenty years old: long-term results. J Bone Joint Surg Am 1996; 78: 995–100. 20. Callaghan JJ, Albright JC, Goetz DD, Olejniczak JP, Johnston RC. Charnley total hip arthroplasty with cement. Minimum twenty-five year follow-up. J Bone Joint Surg Am 2000; 82(4): 487–497. 21. Nicholls MA, Selby JB, Hartford JM. Athletic activity after total joint replacement. Orthopedics 2002; 11: 1283–1287. 22. Brown TE, Larson B, Shen F, Moskal JT. Thigh pain after cementless total hip arthroplasty: evaluation and management. J Am Acad Orthop Surg 2002; 10: 385–392. 23. Anderson FA Jr, Hirsh J, White K, Fitzgerald RH Jr. Temporal trends in prevention of venous thromboembolism following primary total hip or knee arthroplasty 1996–2001: findings from the Hip and Knee Registry. Chest 2003; 124(6 suppl): 349S–356S. 24. Wroblewski BM, Siney PD, Fleming PA. Fatal pulmonary embolism and mortality after revision of failed total hip arthroplasties. J Arthroplasty 2000; 15: 437–439. 25. Hirakawa, Jacobs JJ, Urban R, Saito T. Mechanisms of failure of total hip replacements: lessons learned from retrieval studies. Clin Orthop 2004; 420: 10–17. 26. Clohisy JC, Calvert G, Tull F, McDonald D, Maloney WJ. Reasons for revision hip surgery: a retrospective review. Clin Orthop 2004; 429: 188–192. 27. Lieberman IH, Moran E, Hasttings DE, Bogoch ER. Heterotopic ossification after primary cemented and non cemented total hip arthroplasty in patients with osteoarthritis and rheumatoid arthritis. Can J Surg 1994; 37: 135–139. 28. Visuri T, Pukkala E, Pulkkinen P, Paavolainen P. Decreased cancer risk in patients who have been operated on with total hip and knee arthroplasty for primary osteoarthrosis. Acta Orthop Scand 2003; 74: 351–360. 29. Kurtz S, Mowat F, Ong K et al. Prevalence of primary and revision total hip and knee arthroplasty in the United States from 1990 through 2002. J Bone Joint Surg 2005; 87A: 1487–1497. 30. Sharkness CM, Hamburger S, Moore RM Jr, Kaczmarek RG. Prevalence of artificial hip implants and use of health services by recipients. Publ Health Rep 1993; 108: 70–75. 31. Pagnano MW, McLamb LA, Trousdale RT. Primary and revision total hip arthroplasty for patients 90 years of age and older. Mayo Clin Proc 2003; 78: 285–288. 32. Ostendorf M, Johnell O, Malchau H, Dhert WJ et al. The epidemiology of total hip replacement in The Netherlands and Sweden: present status and future needs. Acta Orthop Scand 2002; 73: 282–286. 33. Wells VM, Hearn TC, McCaul KA et al. Changing incidence of primary total hip arthroplasty and total knee arthroplasty for primary osteoarthritis. J Arthroplasty 2002; 17: 267–273. 34. http://www.cdc.gov/nccdphp/dnpa/obesity/trend/prev_char.htm 35. Milner PC, Payne JN, Stanfield RC et al. Inequalities in accessing hip joint replacement for people in need. Eur J Publ Health 2004; 14: 58–62. 36. Kim S, Koebel S, Duffy R. Cost burden of hip and knee replacements in Ohio: estimates from the National Hospital Discharge Survey, 2000. Managed Care Interface August 2004; 22–25. 37. Birkemeyer NJ, Weinstein JN. Medical vs. surgical treatment for low back pain: evidence and clinical practice. Eff Clin Pract 1999; 2(5): 218–227. 38. Deyo RA, Cherkin D, Conrad D et al. Cost, controversy, crisis: low back pain and the health of the public. Ann Rev Publ Health 1991; 12: 141–156. 39. Szpalski M, Gunzburg R, Mayer M. Spine arthroplasty: a historical review. Eur Spine J 2002; 11(suppl 2): S65–84. 40. Waddell G. Low back pain: a twentieth century healthcare enigma. Spine 1996; 21: 2820–2825. 41. NIH Conference – Osteoarthritis: new insights. Part 1: the disease and its risk factors. Ann Intern Med 2000; 133(8): 635–646.

REFERENCES

461

42. Felson DT, Naimark A, Anderson J, Kazis L et al. The prevalence of knee osteoarthritis in the elderly. The Framingham Osteoarthritis Study. Arthritis Rheum 1987; 30: 914–918. 43. Bijkerk C, Houwing-Duistermaat JJ, Valkenburg HA et al. Heritabilities of radiologic osteoarthritis in peripheral joints and of disc degeneration of the spine. Arthritis Rheum 1999; 42(8): 1729–1735. 44. Hassett G, Hart DJ, Manek NJ et al. Risk factors for progression of lumbar spine disc degeneration: the Chingford study. Arthritis Rheum 2003; 48(11): 1312–1317. 45. Deyo RA, Bass JE. Lifestyle and low back pain. The influence of smoking and obesity. Spine 1989; 14(5): 501–506. 46. Frymoyer JW, Pope MH, Clements JH. Risk factors in low back pain. An epidemiological survey. J Bone Joint Surg 1983; 65(2): 213–218. 47. Harada A, Okuizumi H, Miyagi N et al. Correlation between bone mineral density and intervertebral disc degeneration. Spine 1998; 23: 857–861. 48. Fujiwara A, Lim T-H, An HS et al. The effect of disc degeneration and facet joint osteoarthritis on the segmental flexibility of the lumbar spine. Spine 2000; 25(23): 3036–3044. 49. O’Neill TW, McCloskey EV, Kanis JA et al. The distribution, determinants, and clinical correlates of vertebral osteophytosis: a population-based survey. J Rheumatol 1999; 26(4): 842–848. 50. Yoshimura N, Dennison E, Wilman C et al. Epidemiology of chronic disc degeneration and osteoarthritis of the lumbar spine in Britain and Japan: a comparative study. J Rheumatol 2000; 27(2): 429–433. 51. Williams KD, Park AL. Lower back pain and disorders of intervertebral discs. In Canale ST (ed.), Campbell’s Operative Orthopaedics, vol 2, 10th edn. St. Louis, MO: Mosby, 2003; 2016–2028. 52. Kozak JA, O’Brien JP. Simultaneous combined anterior and posterior fusion. An independent analysis of a treatment for the disabled low-back pain patient. Spine 1990; 15(4): 322–328. 53. Linson MA, Williams H. Anterior and combined anteroposterior fusion for lumbar disc pain. A preliminary study. Spine 1991; 16: 143–145. 54. Moore KR, Pinto MR, Butler LM. Degenerative disc disease treated with combined anterior and posterior arthrodesis and posterior instrumentation. Spine 2002; 27: 1680–1686. 55. Penta M, Fraser RD. Anterior lumbar interbody fusion. A minimum 10-year follow-up. Spine 1997; 22: 2429–2434. 56. Tropiano P, Huang RC, Girardi FP et al. Lumbar total disc replacement. Seven to eleven-year follow-up. J Bone Joint Surg Am 2005; 87A(3): 490–496. 57. Turner JA, Ersek M, Herron L et al. Patient outcomes after lumbar spinal fusions. J Am Med Assoc 1992; 268: 907–11. 58. Gamradt SC, Wang JC. Lumbar disc arthroplasty. Spine J 2005; 5(1): 95–103. 59. de Kleuver M, Oner FC, Jacobs WCH. Total disc replacement for chronic low back pain: background and a systematic review of the literature. Eur Spine J 2003; 12: 108–116. 60. Puolakka TJ, Pajamaki KJ, Halonen PF et al. The Finnish Arthroplasty Register: report of the hip register. Act Orthop Scand 2001; 72: 433–441. 61. Gibson JNA, Grant IC, Waddell G. The Cochrane review of surgery for lumbar disc prolapse and degenerative lumbar spondylosis. Spine 1999; 24(17): 1820–1832. 62. Zeegers WS, Bohnen LMLJ, Laaper M, Verhaegen MJA. Artificial disc replacement with modular type SB Charite´ III: 2-year results in 50 prospectively studied patients. Eur Spine J 1999; 8: 210–17. 63. van Olij A, Oner FC, Verbout AJ. Complications of artificial disc replacement: a report of 27 patients with SB Charite´ disc. J Spinal Disord Tech 2003; 16(4): 369–383. 64. Hallab N, Merritt K, Jacobs JJ. Metal sensitivity in patients with orthopaedic implants. J Bone Joint Surg Am 2001; 83A(3): 428–436.

462

CH29

ORTHOPEDIC DEVICES: EPIDEMIOLOGIC CONSIDERATIONS

65. Merritt K, Rodrigo JJ. Immune response to synthetic materials. Sensitization of patients receiving orthopedic implants. Clin Orthop 1996; 326: 71–79. 66. Hallab N, Link HD, McAfee PC. Biomaterial optimization in total disc arthroplasty. Spine 2003; 28(20S): S139–152. 67. Olsen JH, McLaughlin JK, Nyren O et al. Hip and knee implantations among patients with osteoarthritis and risk of cancer: a record-linkage study from Denmark. Int J Cancer 1999; 81(5): 719–722. 68. Signorello LB, Ye W, Fryzek JP et al. Nationwide study of cancer risk among hip replacement patients in Sweden. J Natl Cancer Inst 2001; 93(18): 1405–1410. 69. Wang, JC, Shapiro MS, Hatch JD. The outcome of lumbar discectomy in elite athletes. Spine 1999; 24: 570–573. 70. Silvers HR, Lewis PJ, Asch HL et al. Lumbar microdiscectomy in the elderly patient. Br J Neurosurg 1997; 11: 16–24. 71. Cinotti G, David T, Postacchini F. Results of disc prosthesis after minimum follow-up period of 2 years. Spine 1996; 21: 995–1000. 72. Cihangiroglu M, Yildurum H, Bozgeyik Z et al. Observer variability based on the strength of MR scanners in the assessment of lumbar degenerative disc disease. Eur J Radiol 2004; 51(3): 202–208. 73. Modic TM. Degenerative disc disease and back pain. Magn Reson Imag Clin N Am 1999; 7(3): 481–491. 74. Resnick DK, Malone DG, Ryken TC. Guidelines for the use of discography for the diagnosis of painful degenerative lumbar disease. Neurosurg Focus 2002; 13(2):1–9. 75. Griffith SL, Shelokov AP, Buttner-Janz K et al.. A multicenter retrospective study of the clinical results of the LINK SB Charite´ intervertebral prosthesis. The initial European experience. Spine 1994; 19: 1842–1849. 76. Lemaire JP, Skalli W, Lavaste F et al. Intervertebral disc prosthesis. Results and prospects for the year 2000. Clin Orthop 1997; 337: 64–76. 77. Baker JK, Reardon PR, Reardon MJ et al. Vascular injury in anterior lumbar surgery. Spine 1993; 18(15): 2227–2230. 78. Bingol H, Cingoz F, Yilmaz AT et al. Vascular complications related to lumbar disc surgery. J Neurosurg 2004; 100: 249–253. 79. Wiese M, Kra¨mer J, Bernsmann K et al. The related outcome and complication rate in primary lumbar microscopic disc surgery depending on the surgeon’s experience: comparative studies. Spine J 2004; 4(5): 550–556. 80. Geisler FH, Blumenthal SL, Guyer RD et al. Neurological complications of lumbar artificial disc replacement and comparison of clinical results with those related to lumbar arthrodesis in the literature: results of a multicenter, prospective, randomized investigational device exemption study of Charite´ intervertebral disc. J Neurosurg Spine 2004; 1(2): 143–154. 81. Torrence ME. Data sources: use in the epidemiologic study of medical devices. Epidemiology 2002; 13: S10–S14.

30 Clinical epidemiology of intrapartum fetal monitoring devices Danica Marinac-Dabic US Food and Drug Administration, Rockville, MD, USA

Barry S. Schifrin Kaiser Permanente Los Angeles Medical Center, Los Angeles, California, USA

Cara J. Krulewitch University of Maryland School of Nursing, Baltimore, MD, USA

Roscoe M. Moore Jr. US Public Health Service (Retired), Rockville, MD, USA

Introduction Prior to the advent of the electronic fetal monitor (EFM) in the early 1970s, fetal surveillance during labor was limited to intermittent auscultation of the fetal heart rate (FHR) and observation of presence of meconium. The information was difficult to ascertain accurately and the data that was collected was irreproducible. The auscultatory

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

464

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

criteria for fetal distress, at that point, had not been modified for more than half a century. The detection of fetal distress was unreliable, especially in the early stages, and there were no experts in auscultation. The intrapartum stillbirth rate was high but, more importantly, apparently occurred without warning. Thus, until the fetus was actually delivered there could be no reassurance whether the fetus was healthy, modestly compromised, hopelessly moribund, or dead. During the past three decades the introduction of continuous electronic fetal monitoring (EFM), followed by gradual additions of adjunctive techniques, dramatically changed the maternal/fetal intrapartum monitoring and care. EFM of the fetal heart rate during labor offered the promise of enhancing fetal outcome by reducing both intrapartum stillbirth and neonatal death rates as well as the risk of subsequent adverse neurological outcome in the offspring. The notion was that EFM would serve as a screening test for severe asphyxia (i.e. asphyxia severe enough to cause neurological; damage or fetal death). By this precept, EFM would allow the early recognition of asphyxia so that timely obstetric intervention could avoid asphyxia-induced brain damage or death in the newborn [1]. It was also expected to reduce the need for cesarean section for fetal distress, since most distress determined by auscultation did not correlate with adverse outcome [2]. At least one of the expectations of EFM was realized: there was a virtual disappearance of intrapartum stillbirths, going from about 3 per 1000 births to almost nil [3,4]. This benefit was difficult to attribute to any other cause. However, it is unclear whether the decrease in neonatal death rate was attributable solely to EFM or to a combination of EFM and modern neonatal intensive care units. This inability to quantify the benefits of EFM is a recurring theme in modern obstetrics. We examine the epidemiologic evidence of the value of four intrapartum monitoring modalities, including EFM, fetal blood gases analyses, fetal pulse oximetry and fetal ECG waveform analysis, and discuss the limitations and challenges in interpretation of the epidemiologic findings.

Electronic fetal monitoring As initially proposed, the purpose of the continuous EFM during labor was to obtain information on response of the FHR to the stresses of repeated uterine contractions to determine the need for intervention or further evaluation.

Brief history EFM was introduced into clinical medicine in the early 1970s and became an almost universal tool for the evaluation of the well-being of the fetus during labor [3,4]. By 1980 about 45% of laboring patients were monitored by this technology. It is estimated that 62% and 74% of laboring patients in 1988 and 1992, respectively, received fetal monitoring. By 2004, about 85% of approximately 4 million births in the USA were

ELECTRONIC FETAL MONITORING

465

monitored in this way [3]. These estimates do not include the widespread use of these devices during the antepartum evaluation of high-risk pregnancies.

Device description The EFM obtains two channels of information and records them continuously on a moving strip chart recorder (see e.g. Figure 30.1). For the display of the fetal heart rate (actually the fetal heart rate pattern), the monitor, by means of either internal or external transducers, determines the interval between consecutive heart beats. For each interval that it detects, the monitor calculates and then plots a ‘rate’, as if all intervals in one minute were exactly the same as the detected interval. Thus, if the device determines that the interval between two consecutive beats is 1 second, it calculates that there would be 60 beats in 1 minute – as if all the beats in 1 minute were 1 second apart. When it detects a new interval, it discards the last interval and performs the calculation anew. Thus, if the next interval were 0.8 seconds the device would calculate a rate of 75 beats per minute (bpm), again, as if all the intervals in 1 minute were the same. By determining an ‘instantaneous rate’ for each interval and then plotting that ‘rate’ on a graph, the graph in fact represents the rhythm of the fetal heart rate (or the fetal heart rate pattern). While an ‘average rate’ may be discerned usually easily with the naked eye, the

Figure 30.1 Fetal monitoring system, 1980. Fetal monitoring system with digital and recorded display of fetal heart rate, intrauterine pressure, and uterine activity. Reproduced with permission from the Science Museum/Science and Society Library

466

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

interpretation of EFM strips is essentially a description of the rhythm and its changes over time. The second channel of the monitor provides information about the duration and intensity of uterine contractions, along with (less reliably) fetal and maternal movements. The detection of contractions is an important part of the monitoring scheme. Uterine contractions consistently decrease uterine blood flow in direct proportion to the intensity and duration of the contractions. Indeed, beyond about 50 mmHg of intrauterine pressure there is no option for further exchange between mother and fetus of oxygen, carbon dioxide, nutrients, and heat across the placenta. Contractions also stimulate the fetus, create pressure on the fetal head and may jostle the fetus so that it impinges on its umbilical cord. The active, neurologically normal fetus signals the onset of a contraction by making truncal movements [5]. These diminish as the contraction peaks. At the end of the contraction the fetus often exhibits ‘breathing movements’. Each of these behavioral activities, in turn, is reflected in a unique FHR pattern [6]. On the other hand, any critical diminution of oxygen availability associated with the contraction will be immediately reflected in a deceleration of the FHR pattern before changes in baseline rate or variability [7]. EFM is very different from monitoring of an adult in a coronary care unit, or the fetus monitored by intermittent auscultation (IA). In the coronary care unit, the patient is usually at rest. The role of all the attached devices is to detect any perceived abnormality of rhythm or ECG pattern as quickly as possible and alert the qualified health care provider so that proper intervention is as timely and appropriate as possible. As a practical matter, the FHR-based system of fetal surveillance system is quite different. The fetus is not at rest with the monitor waiting for some alteration. Rather, the fetus is obliged to deal with the provocations and impositions of the frequently recurring contractions with their manifold effects on blood flow, etc. In addition, the mature fetus has many resources for manipulating its heart rate and cardiac output. Abnormal heart rate rhythms may indicate fetal compromise but they may also be indicative of other conditions unrelated to health or well-being. Breathing, resting, moving, sucking, mouthing, and REM sleep all produce discernible, reproducible changes in the FHR pattern [8]. There are no auscultatory patterns in the fetus corresponding to the fetal behavioral changes or the early indicators of hypoxemia that are readily seen in FHR patterns. IA is used exclusively to determine whether requirements for intervention (i.e. a persistently low fetal heart rate – the only criterion for intervention with IA) have been met.

Epidemiologic assessment There are no studies that demonstrate that IA, by itself, is of any value! Indeed, the largest study of its kind, involving over 50 000 deliveries (at a time before EFM was available) found that IA could not be used for the detection of early fetal distress [9]. Initial studies of monitoring compared immediate fetal outcomes before and after the introduction of EFM and suggested considerable benefit for every measure of maternal and fetal

ELECTRONIC FETAL MONITORING

467

outcome [1,10,11]. In 1984, Niswander et al. reported that fetuses whose deaths were ascribed to asphyxia or trauma, and babies born at term who had seizures within 48 hours of delivery, were significantly more likely than healthy controls to have had signs of fetal distress on the monitor that had not been properly attended to [12]. Babies with seizures, as well as those with terminal apnea, were also substantially more likely than controls to have been born after a failure to react appropriately to signs of severe fetal distress during labor [12]. A meta-analysis by Vintzeleos et al., using the findings of nine randomized controlled trials (RCT), concluded that compared to intermittent auscultation, EFM increased the cesarean section rate overall (OR 1.53, 1.17–2.01) as well as the cesarean rate associated with suspected fetal distress (OR 2.55, 1.81–2.53). EFM also increased the use of operative vaginal delivery by forceps (OR 2.4, 1.97–3.18) and vacuum extraction (OR 2.4, 1.97–3.18). While the overall perinatal mortality was not reduced (OR 0.87, 0.57– 1.33), there was a decreased risk of mortality due to hypoxia in the monitored group (OR 0.41, 0.17–0.98) [13]. Importantly, one of the studies contributing to this meta-analysis and to the finding of decreased death related to hypoxia was conducted in Greece, where there was no widespread monitoring before the RCT and a preexisting high baseline perinatal mortality in the community. These demographics suggest that the higher the mortality was before the introduction of monitoring, the greater the benefit in mortality reduction related to hypoxia that could be demonstrated. Every other RCT was conducted in an institution in which EFM was already in wide use. This suggests that once EFM had been introduced into standard care, it was more difficult to demonstrate a specific benefit in terms of perinatal death without an extraordinary number of patients, since there were then only a very limited number of controls available for analysis. In the meta-analysis carried out by Thacker [14,15], the authors included 12 randomized controlled trials involving over 18 000 patients. Seizures occurred less frequently in EFM patients than in auscultated patients (0.8% vs. 0.1%). This meta-analysis again found an increased risk of operative delivery and no obvious benefit in terms of mortality or demonstrable morbidity other than seizures. The experimental design, methodology, and to some extent the results of the individual RCTs and other studies on the effectiveness of EFM diverge so widely that they permit almost any conclusion about the benefit, or lack thereof, of EFM or auscultation [16,17]. One may thus conclude from individual studies that FHR patterns have not been proved to relate to neonatal outcome, that it is ‘equivalent to’ IA (although IA, by itself, has not been shown to be either predictive or improving of outcome), that it has no impact on fetal death or disability, or that it increases the risk of adverse outcome. In the study of Shy et al., for example, the risk of neurological handicap in the study group was significantly greater than that in the control group [18]. The largest long-term RCT comparing EFM and auscultation derives from Dublin, Ireland [19]. In this study, the control group (IA) had about a 21% higher incidence of seizures compared to the electronically monitored patients [19]. Those newborns with neonatal seizures, however, failed to show significant neurological sequelae at 4 year follow-up [20]. This is important, because the frequency of neonatal seizures is widely regarded as a measure of the quality of obstetrical care, and their prevention is considered

468

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

medically and economically beneficial. More importantly, the majority of newborns in the trial who had cerebral palsy were not in the group of those fetuses who had FHR tracings that were considered ominous [20,21]. When data from the Dublin study was analyzed retrospectively, there was no relationship between auscultation findings (the basis for monitoring in the control group) and any adverse outcome – but there was an obvious, direct correlation between abnormal FHR patterns and adverse outcomes [22]. Similarly, in the RCT of Shy et al., which looked exclusively at preterm fetuses, despite more detrimental outcomes in electronically monitored preterm fetuses, there was no relationship between auscultation and outcome, but a direct correlation between the abnormal patterns and adverse outcome [23]. The potential reasons for fetal heart rate monitoring that have not been shown to be effective are many, some of them discernible from identifiable flaws in the design of the studies, including selection bias and other methodological flaws [1]. However, it is also possible that technical difficulties in the use and interpretation of EFMs are at least partly responsible.

Methodological deficiencies The failure to find correlation between the use of EFM and improvement in outcome relate in part to the use of outcome measures that may not be related to hypoxia or neurological injury during labor. Indeed, most of the studies are limited to immediate measures of outcome such as death, requirement for intensive care support, seizures or acidosis. In these studies, there was no assurance that the fetus was neurologically normal at the outset of monitoring, neither were other causes of adverse outcome in the newborn during labor or at delivery considered, including such factors as birth trauma, sepsis, drug depression, or inadequate resuscitation. These obvious influences on neonatal outcome would, however, not be expected to be associated with specific FHR abnormalities. Long-term follow-up was generally not part of the studies – except as noted above in the Dublin study. In fact, later neuro-radiological assessments have shown the limitations of trying to ascertain correlations between clinical evaluations at the time of birth and subsequent neurological handicap [24,25]. No comprehensive evaluation of the correlation between FHR patterns and subsequent neuro-radiological abnormalities or other long-term impact has yet been reported. Other criticisms of the FHM studies include a lack of standardized interpretation of fetal heart rate patterns. The Dublin study made reference to two different, fundamentally incompatible classifications. There is a broad understanding that marked differences exist in the interpretation of FHR patterns, even among experts and even when the classification is agreed upon [26]. Finally, there is lack of agreement regarding algorithms for intervention for specific FHR patterns [3] – a condition that continues today even with more modern schemes of interpretation [27]. Another methodological problem is that almost invariably the clinical trials that compare EFM with IA have excluded subjects at risk for adverse outcomes, thus making

ELECTRONIC FETAL MONITORING

469

comparative safety assessments dubious. In the Dublin trial, for example, patients were included only after membranes were ruptured and the presence of meconium excluded [19]. Another common failure leading to potential selection bias is the lack of consideration of the impact of antepartum surveillance with EFM on the candidacy of a patient to enter the trial. It is reasonable to conclude, as Chalmers has, that existing studies on the effectiveness of EFM have been beset by technical and statistical difficulties [28]. It is therefore reasonable to argue that the benefits of EFM remain unsatisfactorily tested. Notwithstanding the recommendations by the American College of Obstetricians and Gynecologists (ACOG), International Federation of Gynecology and Obstetrics (FIGO), the Society of Obstetricians and Gynecologists of Canada, and the Royal College of Obstetricians and Gynaecologists, based on the results of these RCTs, all assert that auscultation is equivalent to EFM, that EFM is not required on admission to labor and delivery, and that intermittent auscultation is relevant as a primary, or indeed the only, necessary method of surveillance [3,29]. Yet a more recent publication from the American College of Obstetricians and Gynecologists seems to require EFM in high-risk patients [3].

Lack of standardized FHR tracings nomenclature As mentioned above, the nomenclature of patterns used to characterize FHR tracings have not been standardized and there is often disagreement, even among ‘experts’ [30–37], neither has the nomenclature been systematized to provide widely agreed upon guidelines for intervention [1]. The National Institute of Child Health and Development (NICHD) Guidelines (for research) dealt only with a limited number of features of FHR patterns and failed to provide guidelines for intervention in the large number of ‘nonreassuring’ FHR patterns [35]. It is also clear that while different classifications function primarily to direct attention to phenomena within the tracing (decelerations, variability, tachycardia, etc.), they do little to explain the source of the heart rate pattern abnormalities or the probability that they arise from clinically important pathology. Moreover, these classifications are deemed relevant only to the detection of hypoxia and acidosis and not in relationship to the detection of such assessable fetal parameters as neurological integrity and behavior [38–41].

Technical difficulties The source of the FHR and uterine contraction signals (internal or external), the weight of the mother, and expulsive efforts will all influence both the accuracy of the interpretation and the potential for misrepresentation of the EFM results. Abnormalities of the tracing

470

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

are found predominantly during the second stage of labor, the expulsive phase, when technical difficulties are most likely to interfere with interpretation. Indeed, in some cases the monitor may double-count or half-count the fetal or maternal rates, or mistake the maternal heart rate pattern for the fetal pattern, with sometimes catastrophic results [42–44]. In the Dublin trial, tracings could not be properly analyzed in about 40% of patients during the second stage [19].

Scalp and umbilical cord blood gases analyses Brief history Fetal scalp blood sampling was developed shortly before EFM and revolutionized intrapartum care by obtaining information directly from the fetus. The premise of monitoring fetal capillary blood was, again, based on the notion that acidosis (hypoxia) was directly related to fetal injury. Presumably, its timely detection would permit prompt and effective intervention.

Epidemiologic assessment The fetal blood sampling technique was somewhat difficult to implement; the sampling of fetal blood was intermittent and prone to error. Nevertheless, initially, it was included in several of the RCTs [19,45] that were designed to assist in the evaluation of certain nonreassuring FHR patterns. It is not used as a primary mode of surveillance. In the RCTs it was credited with diminishing the incidence of cesarean section compared to EFM alone [19,45]. Because of the difficulties in obtaining the samples, the need for multiple samples and the increasing reliance on FHR patterns, there has been a progressive diminution in the use of this technique [46]. Assessment of umbilical cord pH and blood gases immediately after birth is widely used to assess the acid–base balance of the fetus in utero just before delivery. To properly interpret fetal scalp and umbilical cord blood gases, there must be an understanding of numerous technical pitfalls that can compromise interpretation of both the timing and severity of fetal hypoxia. These may include: (a) internal inconsistency in the analysis of a blood gas sample; (b) blood gas samples (labeled ‘artery’ and ‘vein’) may actually derive from the same vessel, or the reporting of the results may be reversed; or (c) the samples may be contaminated with an air bubble(s). On occasion, only a single sample or only a single pH may be available. Unless both an umbilical venous and an umbilical arterial sample are obtained and a base excess calculated in each, one cannot be certain that the only sample obtained is indeed from an umbilical artery. Westgate et al., for example, found that only about 75% of supposedly paired umbilical cord samples had validated data from both an artery and a vein; 54 (2.8%) had only one blood sample available, 90 (4.6%) had an error in the pH or pCO2 of one vessel, and in 350 (18%) pairs the differences between vessels indicated that they were not

FETAL PULSE OXIMETRY

471

sampled from artery and vein as intended [47]. Only 60% of the cases with an arterial pH less than 7.05 had evidence of a metabolic acidosis (base deficit in the extracellular fluid 10 mmol/l or more). Severe intracellular acidosis of a specific fetal structure may not be detected in the umbilical cord arterial sample if there is no venous return from an ischemic area, for example, a cerebral infarct. Occlusion of the umbilical arteries by either stretching or external compression becomes easier as fetal arterial pressure falls. If partial or total umbilical venous occlusion occurs periodically, as is usually the case, the umbilical artery cord gas may recover partially or completely between episodes. If the episodes are severe enough or frequent enough, they may or may not be reflected in the umbilical cord gases. Similarly, in obtaining samples from the fetal scalp during labor it is important to obtain the sample between contractions, as this will provide the values associated with recovery, as opposed to the time during the embarrassment. Severe acidosis may be absent from newborns who have suffered catastrophic injury during labor [48]. Conversely, severe neonatal acidosis, even associated with abnormal neurological findings, is not a good predictor of subsequent neurological injury [49]. Westgate et al., along with the Pomerance, have concluded that confusion about the value of cord gas measurements may be due to the pitfalls in obtaining reliable umbilical cord data, the use of erroneous data, and finally, inadequate definitions of acidosis [47,50]. These pitfalls have been poorly addressed in various outcome studies.

Fetal pulse oximetry Brief history In May 2000, the US Food and Drug Administration (FDA) granted conditional approval of the OxiFirstTM Fetal Oxygen Saturation Monitoring System, to be used as an adjunct to FHR monitoring to continuously monitor fetal intrapartum oxygen saturation in a singleton vertex fetus with a gestational age of at least 36 weeks, in the presence of a nonreassuring heart rate pattern, after fetal membranes have ruptured.

Device description This device was designed to continuously measure fetal oxygen saturation (FSpO2) in the presence of a nonreassuring fetal heart rate (FHR) pattern. It was not to be used as a primary form of surveillance. A specialized sensor is inserted through the dilated cervix after the membranes have been ruptured and positioned against the fetal face, permitting measurement of FSpO2 during labor. The stated and implied benefits of the technology include: (a) reduced cesarean section rate for so-called ‘nonreassuring fetal heart rate patterns’ (NRFHRP); (b) simplicity, in that even a single reading of more than 30% between contractions is assurance that the fetus is adequately oxygenated [51,52].

472

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

Epidemiologic assessment A clinical trial used to support device approval included 1010 women with labors complicated by nonreassuring FHR patterns that were randomly assigned to either EFM alone or EFM plus continuous fetal pulse oximetry [51]. In this randomized clinical trial there was indeed a reduction in cesarean delivery rates for ‘nonreassuring fetal heart rate patterns’, from 10.2% to 4.5% (p ¼ 0.007), but there was no difference in the immediate neonatal outcomes or the overall cesarean delivery rates (the latter as a result of an increase in abdominal delivery for dystocia in the pulse oximetry group) [51]. Based on these findings, the ACOG did not endorse the device for use in clinical practice [3,53]. The conditions of FDA approval included the completion of three postapproval studies. The ‘General Use Study’ was stopped early because of the inability to recruit sufficient patients. The ‘Dystocia Study’, a prospective non-randomized observational cohort study in nulliparous women at term, examined the relationship between nonreassuring fetal heart rate patterns and operative delivery for dystocia, compared two distinct classes of nonreassuring fetal heart rate patterns (class I, intermittent, mildly nonreassuring; class II, persistent, progressive, and moderate to severely nonreassuring) among nulliparous patients with the use of fetal pulse oximetry to confirm fetal wellbeing. The study found that ‘significantly’ nonreassuring fetal heart rate patterns predict cesarean delivery for dystocia among nulliparous patients with normally oxygenated fetuses in a setting of a standardized labor management protocol. This finding confirmed the observations from the earlier randomized controlled trial of fetal pulse oximetry [54]. As a part of the postapproval study requirements, the National Institute of Child Health and Human Development (NICHD) Maternal–Fetal Medicine Units (MFMU) Network performed a randomized trial designed to evaluate the effectiveness and safety of fetal pulse oximetry. The primary objective was to measure whether fetal oximetry, as an adjunct to EFM, would be associated with a reduction in overall cesarean rate. The secondary objectives were to study whether knowledge of FSpO2 changed the risk of a cesarean delivery for nonreassuring FHR and for dystocia. Additional objectives also included potential maternal and neonatal adverse effects. The study was stopped prematurely because the findings, up to that point, revealed no significant difference in the overall cesarean rate between the control group (EFM alone) and the study group (EFM plus pulse oximetry) (OR 26.3% vs. 27.5%, respectively; p ¼ 0.31). The cesarean rates for the separate indications of nonreassuring FHR (OR 7.1% vs. 7.9%; p ¼ 0.30) and dystocia (OR 18.6% vs. 19.2%; p ¼ 0.59) were similar for the device vs. the control group. The neonatal condition was similar in both groups. The authors concluded that knowledge of fetal oxygen saturation is associated with neither a reduction in the cesarean section rate nor an improvement in neonatal well-being [55]. Another randomized controlled trial comparing fetal pulse oximetry plus electronic fetal heart rate monitoring to monitoring alone found no difference in cesarean delivery for nonreassuring fetal heart rate tracings or for labor disorders [56]. Compared to IA or EFM accompanied by ancillary techniques (blood sampling, pulse oximetry, fetal ECG analysis; see below), there may be an increased risk of interventions for fetal distress (including cesarean section) when EFM is used alone. This is an

FETAL ECG WAVEFORM MONITORING

473

indicator of the poor specificity of the current approach to identifying the ‘nonreassuring’ FHR tracing. Adding an adjunctive device that could increase the specificity of the moderately ‘nonreassuring’ FHR pattern during labor was felt to be an important contribution to care of the mother and fetus during labor and delivery. However, the conflicting results of the study (no actual decrease in the total number of cesarean sections) and the uncertainty of the relationship of hypoxia or acidosis to fetal injury as just discussed, makes the value of this device unclear [57–59]. Parenthetically, the manufacturer of the fetal pulse oximeter has discontinued production.

Fetal ECG waveform monitoring Brief history The analysis of the ST segment of the fetal electrocardiogram (ECG) has emerged as an adjunctive tool that obstetricians may use to further evaluate the ability of the fetus to respond to the stress of labor, in the face of moderately abnormal ‘nonreassuring’ FHR pattern.

Device description The ST waveform of the fetal ECG provides continuous information on how the fetal heart muscle responds to the stress of labor. Thus, an elevation of the ST segment and T wave, measured as T:QRS ratio, identifies those fetuses that react to stress by releasing stress hormones that mobilize glycogen stored in the heart. An ST depression may indicate situations where the fetal heart is not capable of responding. During the last decade, clinicians in Europe gained extensive clinical experience with the use of this advanced surveillance technology. In 2005, the FDA approved the marketing of the STAN1 S31 Fetal Heart Monitor. This is a fetal heart monitoring system used during labor and delivery to measure, display, and analyze the ST waveform of the fetal ECG (FECG). The STAN System provides this ST waveform analysis as an adjunct to standard Electronic Fetal Monitoring (EFM/CTG) to provide doctors and midwives with more detailed information about the fetal hypoxic situation during labor. Device use requires that an electrode be applied directly to the fetal presenting part. Prerequisites for use of this device include accessibility of the fetal presenting part, sufficient dilatation of the cervix and ruptured membranes.

Epidemiologic assessment Early studies examined: (a) the feasibility of evaluating ST waveform changes during labor; (b) a comparison of the ability of FHR plus elevated T:QRS ratio compared to FHR

474

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

alone to predict fetal acidemia during labor and delivery; and (c) studies directed at enhancing the quality of the fetal ECG complex [60–64]. These early studies demonstrated the feasibility of using STwaveform analysis as an adjunct to standard fetal heart rate monitoring to improve the assessment of the fetal condition during labor. The feasibility studies were followed by a large, randomized controlled trial in a non-academic setting in Plymouth, UK, to collect scientific evidence of the clinical value of fetal heart rate monitoring and ST waveform analysis. This trial tested the hypothesis that the combination of ST waveform and FHR analysis, compared to FHR analysis only, would reduce operative interventions for fetal distress without placing the fetus at risk. Results demonstrated a significant reduction (p < 0.001) in operative deliveries for fetal distress without increasing the risk of hypoxia. This study also demonstrated a trend toward reduction in the number of newborns with metabolic acidosis and low 5 minute Apgar scores. The major limitation of the study was the lack of sufficient power to detect statistically significant differences in these last two endpoints [65]. The Plymouth trial was followed by the Swedish Randomized Control Trial (SRCT), a multicenter trial conducted to evaluate the ability of obstetricians and midwives to use ST segment information and very specific clinical guidelines to monitor labor, and to make clinical management decisions that would result in a reduced number of fetuses being born with metabolic acidosis. The secondary outcome variable was the rate of operative deliveries (cesarean sections, forceps or vacuum deliveries) for fetal distress (ODFD). In addition, the neonatal morbidity was evaluated in terms of Apgar scores at 1 and 5 minutes and admissions to the Special Care Baby Unit (SCBU). Women (4966) in active labor with a single term (> 36 completed gestational weeks) fetus in cephalic presentation were entered into the SRCTafter a clinical decision had been made to apply a fetal scalp electrode for internal monitoring. Of these, 2447 women were randomized to the CTG (the designation of EFM in Europe)-only group and 2519 to CTG þ ST. This study found that the rate of metabolic acidosis at birth was significantly reduced with CTG þ ST monitoring, compared with CTG alone. The CTG þ ST (treatment) group, compared to the CTG-only (control) group, showed a 61% reduction in the number of cases with umbilical artery metabolic acidosis (pH < 7.05 and base deficit > 12 mmol/l) (p ¼ 0.01); 78% reduction in cases with metabolic acidosis admitted to special care baby unit (SCBU) (p ¼ 0.04); 27% fewer operative deliveries for fetal distress (p ¼ 0.009); and 88% fewer neonates with signs of moderate/severe neonatal encephalopathy (OR ¼ 0.12, 95% CI ¼ 0.01–0.94; p ¼ 0.02). The rates of operative interventions for failure to progress did not differ between the two groups. During the second half of the clinical trial, and after additional education on the use of the STAN readings, investigators became more proficient in their abilities to interpret the ST analysis and to manage labor according to specific guidelines unique to the STAN program. The results from the second half of the trial led to a greater reduction in cord metabolic acidosis (75%) and operative deliveries for fetal distress (44%). Data from this study demonstrated that intrapartum fetal heart rate monitoring combined with ST waveform analysis increases the ability of obstetricians to more appropriately intervene in cases of a nonreassuring fetal status during labor, as compared to use of FHR alone [66].

FETAL ECG WAVEFORM MONITORING

475

These two randomized clinical trials, Plymouth and the SCRT, were subject to metaanalysis. The analysis included 6826 cases and demonstrated that with the ST analysis, the proportion of babies born with metabolic acidosis can be reduced from 1.46% to 0.65% (OR ¼ 0.44; p ¼ 0.003). The analysis also showed that the rates of ODFD were reduced from 9.2% to 6.8% (OR 0.72; p < 0.001). At the same time, the rate of moderate/ severe encephalopathy was reduced from 3.3 to 0.5 per 1000 (OR 0.16; p ¼ 0.01) [67]. A recent Cochran review of the benefits of the STAN analysis system supports its use as an adjunct to electronic fetal monitoring [68]. Observational studies contributed tremendously to the body of knowledge regarding this new technique. The European Centers of Excellence project (EU project) initially included 10 academic institutions across Europe (The Gothenburg Study was the first in this series) that became the Europeans Centers of Excellence. The project started in 2000, finished in 2002 and included approximately 8000 fetal monitoring recordings. The STAN Fetal Heart Monitor was used as part of normal clinical management practice at each center and included an education and training program. The main purpose of the EU Project was to assess the success of the training program. The first prospective, observational study was conducted at two maternity wards in Gothenburg, Sweden. Between October 2000 and September 2002, a total of 4830 highrisk deliveries were monitored, using the STAN S21 Fetal Heart Monitor. The purpose of the study was to assess the value of the STAN monitor for intrapartum fetal monitoring using CTG and STwaveform analysis. A comparison between the first and second years showed use of the STAN Fetal Heart Monitor increased from 22% to 38%. The increase in usage was correlated with an overall reduction in metabolic acidosis rate from 0.77% to 0.44% in the total population, the same low incidence of hypoxic neonatal encephalopathy as noted in the SRCT, and no increase in operative interventions for fetal distress. The study investigators concluded that, with continued use and experience, the STAN Fetal Heart Monitor provides consistent improvement in fetal outcome. Of note, a retrospective study conducted at the University of Western Ontario, Ontario, Canada, reported very different findings. The study was conducted to determine the ability of intrapartum electronic fetal heart rate monitoring (EFM) plus fetal electrocardiogram analysis (STAN) to predict metabolic acidemia (defined as an umbilical cord artery pH < 7.15 and base deficit 12 mmol/l) at birth [69]. Only fetal heart rate information (using the FIGO classification) was used for clinical management of the 143 patients in the study. Attending physicians were blinded to the ST analysis information. Based on retrospective analysis of the ability of the ST waveform to predict metabolic acidosis at birth (BD > 12 mmol/l), the two study investigators found a sensitivity of 43%, specificity of 74%, negative predictive value of 96%, and a positive predictive value of 8%. The authors concluded that the STAN clinical guidelines have a poor positive predictive value and a sensitivity of less than 50% for predicting metabolic acidemia at birth. Similar results were reported by Haberstich et al. (30%) and Kwee (46%) for the same degree of metabolic acidosis [70,71]. Even when STAN is credited with reducing the risk of metabolic acidosis (pH < 7.05 and BD > 12 mmol/l) [64], no significant differences in Apgar scores, admissions to the NICU or neonatal encephalopathy were found between those monitored with

476

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

CTG þ STAN and those monitored with CTG alone [72]. Finally, it has yet to be shown whether determining myocardial hypoxia (from STAN analysis) will diminish the risk of subsequent neurological handicap, as measured by long-term follow-up [72]. The outcomes from this study appear inconsistent with those of the Swedish RCT. Potential reasons for these differences include the use of a standard spiral scalp electrode that had not previously been tested for ST analysis (resulting in ECG signal loss) and interpretation of STAN indications out of clinical context [69]. In all these studies, the outcome results depend on the abnormalities in the tracing that are considered ‘nonreassuring’, as well as the scheme of intervention when problematic tracings are combined with problematic labors. Finally, there is the technical issue of poor-quality signals (ECG signal loss) that occur about 10% of the tracing time [69,70,73]. The failure rate of EFM tracings is essentially nil. It must be understood that the premise of both fetal pulse oximetry and the fetal ECG waveform is that monitoring these modalities of care would reduce the rate of intervention for ‘nonreassuring’ FHR patterns thought to be ‘potentially’ indicative of fetal distress. In both studies, the normal FHR tracing was considered sufficiently reliable as to not require ancillary testing. Similarly, certain markedly abnormal FHR patterns were considered so severe, that intervention was required irrespective of the STAN or pulse oximetry reading. The results of the studies evaluating the STAN system seem to conclude that, with proper training, the users were able to use the STAN Fetal Heart Monitor according to its intended use to improve assessment of the fetal condition during labor as an adjunct to standard fetal heart rate monitoring [70,74–76]. The results are seemingly more likely due to an enhancement of the clinical skills of STAN users, due to training and experience, and less from an improvement in the technology. To date there has been no analysis of the data to show which aspects of the entire STAN program are responsible for the apparent improvement in both the operative intervention rate and the rate of metabolic acidosis. There are no studies of long-term outcome. If the results of the STAN study are valid and reproducible, then the STAN technology and its associated program for monitoring labor should be able to assist in the reclassification of the FHR patterns into those rarely associated with STAN abnormalities and those frequently associated with STAN abnormalities.

Challenges of intrapartum fetal monitoring modalities When the FHR pattern is entirely reassuring, there is reasonable diagnostic certainty that there is no fetal indication for intervention (that the fetus is tolerating the stresses of contractions), and that no ancillary testing is required. It does not exclude intervention on the basis of fetal size, or presentation, or position, or pelvic size, or a maternal condition. Thus, the determination of the timing of intervention will depend as much on the estimated feasibility and timing of safe vaginal delivery as it does on the heart rate pattern. The value of the evaluations of fetal well-being by whatever modality (or combination) and the feasibility of safe vaginal delivery are modified by the reliability

CHALLENGES OF INTRAPARTUM FETAL MONITORING MODALITIES

477

and predictability of the information, the skill of the interpreter, and the clinical context. The approach being sought from EFM, and likely by the STAN program, emphasizes the need to examine and codify trends in surveillance parameters in an effort to identify the boundaries that separate the normally oxygenated and normally behaving fetus from the hypoxemic one. It seems necessary to avoid discussing tracings or STAN or pH values as a ‘snapshot’ without discussing their evolution [77]. Such a limited perspective cannot permit an accurate assessment of the fetal condition or thoughtful deliberation about the urgency and route of delivery. Assessing ‘nonreassuring patterns’ would seem to require less attention to specific features of the types of decelerations or variability, but greater attention to changes in these values, including the stability of the baseline over time. In other words, we need to use the fetus as its own control. The original incentive for implementing fetal monitoring was the notion that the decelerations in FHR produced from uterine contractions would allow early detection of fetal hypoxia and early intervention, in other words rescue, of the fetus from impending decompensation. This often-repeated precept of EFM regarding intervention may indeed be the major impediment to realizing its benefits. To optimize the timing of intervention, there must be knowledge about the types of FHR patterns and mechanisms (plural) that produce injury, the speed of deterioration and the options for recoverability. Clearly, some patterns develop slowly with ample, non-critical, opportunities for intervention, with the expectation of normal outcome. Other patterns progress more rapidly and the timing of intervention becomes critical if the fetus is to be rescued. Still other patterns develop so rapidly and unpredictably that timely intervention is improbable and, irrespective of the speed of intervention, the fetus may not escape injury or death [38,78]. In other circumstances, the opportunity for timely intervention has long passed and intervention, certainly required by the standard of care, may not change the inevitable handicap or death. There has been little meaningful discussion of these issues, or correlation of specific FHR patterns with various specific clinical abnormities. As a practical matter, therefore, the decelerations (or pH or STAN abnormalities) that can be tolerated at full dilatation of the cervix with a head ready to deliver may not reasonably be tolerated in early labor before descent of the head or rupture of the membranes – factors likely to further increase the severity of decelerations. Indeed, the lower the feasibility of safe vaginal delivery, the lower the threshold for intervention should be for abnormal FHR patterns, acidosis, or STAN abnormality. For example, in the patient with an arrest of dilatation and a modestly deteriorating FHR pattern, what can be the benefit of waiting for an abnormal fetal oxygen saturation or critically abnormal STAN reading or low pH? Indeed, fetal pulse oximetry, for example, may be inappropriate for the evaluation of certain abnormal patterns (severe, variable decelerations) [79]. In addition, abnormal FHR patterns, in fact, anticipate difficulty in labor [54]. Perhaps decelerations occur for a reason. The increased frequency of variable decelerations, operative intervention, and subsequent adverse neurological outcome with fetuses in occiput posterior positions have been known for some time [80–84]. It stands to reason that the longer and more difficult the labor, the greater is the potential for decelerations from fetal head or cord compression, and for subsequent harm, and the less likely is safe

478

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

vaginal delivery, especially from the OP position [39,82–85]. Continued, unproductive labor may be associated with increased cranial molding and deeper lodgement of the fetal head into the pelvis. It may provoke protracted fetal bradycardia or make ultimate operative delivery by either vaginal or abdominal routes that much more difficult.

Conclusion As much insight as FHR tracings, along with ancillary techniques, are able to provide, they cannot be used to explain many adverse outcomes. Indeed, they provide limited affirmative evidence of adverse outcomes associated with inadequate resuscitation, congenital anomaly, traumatic or difficult delivery, sepsis, and medication. The majority of infants depressed at birth have normal pH and the majority of infants with low pH have normal outcome. There is, however, no reported example of sudden, unexpected fetal death in a fetus with a reassuring FHR pattern. Thus, all of these surveillance modalities (EFM, fetal pulse oximetry, pH, STAN Monitor ) should at least anticipate fetal death in labor, as this outcome, unlike neurological injury, is universally associated with either progressive hypoxia and acidosis or an acute, catastrophic event, e.g. massive abruption. However, whether any of these techniques, by a single value or feature or even a combination thereof, can dictate the critical time of delivery of the fetus with a problematic FHR pattern and a problematic labor, is yet unresolved. A more thorough understanding of this subject would benefit from a comprehensive trial with authoritative, meaningful nomenclature and a broadened perspective on the nature and urgency of intervention. We believe that much more is known about the determinants of fetal heart rate patterns than is currently used clinically in their analysis or classification. ‘Perinatal asphyxia’ does not mean that the fetus was injured. ‘Perinatal injury’ does not mean that the fetus was asphyxiated or that it was preventable. The evaluation of FHR tracings, as well as ancillary techniques, therefore, requires that testing must be performed properly and in a timely way, and the results must be interpreted and communicated correctly using an agreed-upon, widely understood nomenclature and scheme of intervention. Reasonable interpretation must provoke an appropriate, reasoned response in the larger clinical context relating to the condition of the mother and the feasibility of safe vaginal delivery. All these components must be included in any comparative evaluation of surveillance parameters.

References 1. Parer JT, King T. Fetal heart rate monitoring: is it salvageable? Am J Obstet Gynecol 2000; 182(4): 982–987. 2. Paul RH, Gauthier RJ, Quilligan EJ. Clinical fetal monitoring. The usage and relationship to trends in cesarean delivery and perinatal mortality. Acta Obstet Gynecol Scand 1980; 59(4): 289–295.

REFERENCES

479

3. ACOG Practice Bulletin No. 70: Intrapartum fetal heart rate monitoring. Obstet Gynecol 2005; 106(6): 1453–1460. 4. Parer JT. Electronic fetal heart rate monitoring: a story of survival. Obstet Gynecol Surv 2003; 58(9): 561–563. 5. Mulder EJ, Visser GH. Braxton Hicks’ contractions and motor behavior in the near-term human fetus. Am J Obstet Gynecol 1987; 156(3): 543–549. 6. Mulder EJ, Boersma M, Meeuse M, van der Wal M et al. Patterns of breathing movements in the near-term human fetus: relationship to behavioural states. Early Hum Dev 1994; 36(2): 127–135. 7. Murata Y, Martin CB Jr, Ikenoue T et al. Fetal heart rate accelerations and late decelerations during the course of intrauterine death in chronically catheterized rhesus monkeys. Am J Obstet Gynecol 1982; 144(2): 218–223. 8. Nijhuis JG. Fetal behavior. Neurobiol Aging 2003; 24(suppl 1): S41–6; discussion S7–9, S51–2. 9. Benson RC, Shubeck F, Deutschberger J, Weiss W, Berendes H. Fetal heart rate as a predictor of fetal distress. A report from the collaborative project. Obstet Gynecol 1968; 32(2): 259–266. 10. Paul RH. Clinical fetal monitoring. Experience on a large clinical service. Am J Obstet Gynecol 1972; 113(5): 573–577. 11. Mueller-Heubach E, MacDonald HM, Joret D, Portman MA et al. Effects of electronic fetal heart rate monitoring on perinatal outcome and obstetric practices. Am J Obstet Gynecol 1980; 137(7): 758–763. 12. Niswander K, Henson G, Elbourne D et al. Adverse outcome of pregnancy and the quality of obstetric care. Lancet 1984; 2(8407): 827–831. 13. Vintzileos AM, Nochimson DJ, Guzman ER, Knuppel RA et al. Intrapartum electronic fetal heart rate monitoring vs. intermittent auscultation: a meta-analysis. Obstet Gynecol 1995; 85(1): 149–155. 14. Thacker SB, Stroup DF, Peterson HB. Efficacy and safety of intrapartum electronic fetal monitoring: an update. Obstet Gynecol 1995; 86(4 pt 1): 613–620. 15. Thacker SB, Stroup DF. Revisiting the use of the electronic fetal monitor. Lancet 2003; 361(9356): 445–446. 16. Mahomed K, Nyoni R, Mulambo T, Kasule J, Jacobus E. Randomised controlled trial of intrapartum fetal heart rate monitoring. Br Med J 1994; 308(6927): 497–500. 17. Mahomed K, Nyoni R, Mlambo T, Jacobus E, Kasule J. Intrapartum foetal heart rate monitoring – continuous electronic vs. intermittent Doppler – a randomised controlled trial. Cent Afr J Med 1992; 38(12): 458–462. 18. Shy KK, Olshan AF, Hickok DE, Luthy DA. Electronic fetal monitoring during premature labor and the occurrence of perinatal mortality in very low birthweight infants. Birth 1988; 15(1): 14–8. 19. MacDonald D, Grant A, Sheridan-Pereira M, Boylan P, Chalmers I. The Dublin randomized controlled trial of intrapartum fetal heart rate monitoring. Am J Obstet Gynecol 1985; 152(5): 524–539. 20. Grant A, O’Brien N, Joy MT, Hennessy E, MacDonald D. Cerebral palsy among children born during the Dublin randomised trial of intrapartum monitoring. Lancet 1989; 2(8674): 1233–1236. 21. Grant A. Cerebral palsy in infants born during trial of intrapartum monitoring. Lancet 1990; 335(8690): 660. 22. Ellison PH, Foster M, Sheridan-Pereira M, MacDonald D. Electronic fetal heart monitoring, auscultation, and neonatal outcome. Am J Obstet Gynecol 1991; 164(5 pt 1): 1281–1289. 23. Larson EB, van Belle G, Shy KK, Luthy DA et al. Fetal monitoring and predictions by clinicians: observations during a randomized clinical trial in very low birth weight infants. Obstet Gynecol 1989; 74(4): 584–589.

480

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

24. Cowan F, Rutherford M, Groenendaal F et al. Origin and timing of brain lesions in term infants with neonatal encephalopathy. Lancet 2003; 361(9359): 736–742. 25. Mercuri E, Cowan F, Rutherford M, Acolet D et al. Ischaemic and haemorrhagic brain lesions in newborns with seizures and normal Apgar scores. Arch Dis Child Fetal Neonatal Ed 1995; 73(2): F67–74. 26. Donker DK, van Geijn HP, Hasman A. Interobserver variation in the assessment of fetal heart rate recordings. Eur J Obstet Gynecol Reprod Biol 1993; 52(1): 21–28. 27. Electronic fetal heart rate monitoring: research guidelines for interpretation. National Institute of Child Health and Human Development Research Planning Workshop. Am J Obstet Gynecol 1997; 177(6): 1385–1390. 28. Chalmers I. Continuous fetal heart rate monitoring. J R Soc Med 2001; 94(5): 258. 29. Liston R, Crane J, Hamilton E et al. Fetal health surveillance in labour. J Obstet Gynaecol Can 2002; 24(3): 250–276; quiz 77–80. 30. Melchior J, Bernard N. Second-stage fetal heart rate patterns. In Jad S (ed.), Fetal Monitoring, Physiology and Techniques of Antenatal and Intrapartum Assessment. Tunbridge Wells, UK: Castle House Publications; 1989. 31. Hammacher K, Brun del Re R, Gaudenz R, De Grandi P, Richter R. [Cardiotocographic diagnosis of fetal hazard using a CTG-score]. Gynakol Rundsch 1974; 14(suppl 1): 61–63 [in German]. 32. Hon EH. The classification of fetal heart rate. I. A working classification. Obstet Gynecol 1963; 22: 137–146. 33. Dellinger EH, Boehm FH, Crane MM. Electronic fetal heart rate monitoring: early neonatal outcomes associated with normal rate, fetal stress, and fetal distress. Am J Obstet Gynecol 2000; 182(1 pt 1): 214–220. 34. Rooth G, Huch A, Huch R. Guidelines for the use of fetal monitoring. Int J Gynaecol Obstet 1987; 25: 159067. 35. Electronic fetal heart rate monitoring: research guidelines for interpretation. The National Institute of Child Health and Human Development Research Planning Workshop. J Obstet Gynecol Neonatal Nurs 1997; 26(6): 635–640. 36. Ayres-de-Campos D, Bernardes J, Costa-Pereira A, Pereira-Leite L. Inconsistencies in classification by experts of cardiotocograms and subsequent clinical decision. Br J Obstet Gynaecol 1999; 106(12): 1307–1310. 37. Jonker FH, Van Geijn HP, Chan WW, Rausch WD et al. Characteristics of fetal heart rate changes during the expulsive stage of bovine parturition in relation to fetal outcome. Am J Vet Res 1996; 57(9): 1373–1381. 38. Schifrin BS. The CTG and the timing and mechanism of fetal neurological injuries. Best Pract Res Clin Obstet Gynaecol 2004; 18(3): 437–456. 39. Asakura H, Schifrin BS, Myers SA. Intrapartum, atraumatic, non-asphyxial intracranial hemorrhage in a full-term infant. Obstet Gynecol 1994; 84(4 pt 2): 680–683. 40. Freeman JM, Nelson KB. Intrapartum asphyxia and cerebral palsy. Pediatrics 1988; 82(2): 240–249. 41. Freeman RK. Problems with intrapartum fetal heart rate monitoring interpretation and patient management. Obstet Gynecol 2002; 100(4): 813–826. 42. Murray ML. Maternal or fetal heart rate? Avoiding intrapartum misidentification. J Obstet Gynecol Neonatal Nurs 2004; 33(1): 93–104. 43. Sherman DJ, Frenkel E, Kurzweil Y, Padua A et al. Characteristics of maternal heart rate patterns during labor and delivery. Obstet Gynecol 2002; 99(4): 542–547. 44. Schifrin B, Harwell R, Hamilton-Rubinstein T, Visser GH. Maternal heart rate pattern - a confounding factor in intrapartum fetal surveillance. Prenat Neonat Med 2001; 6: 75–82.

REFERENCES

481

45. Haverkamp AD, Orleans M, Langendoerfer S, McFee J et al. A controlled trial of the differential effects of intrapartum fetal monitoring. Am J Obstet Gynecol 1979; 134(4): 399–412. 46. Goodwin TM, Milner-Masterson L, Paul RH. Elimination of fetal scalp blood sampling on a large clinical service. Obstet Gynecol 1994; 83(6): 971–974. 47. Westgate J, Garibaldi JM, Greene KR. Umbilical cord blood gas analysis at delivery: a time for quality data. Br J Obstet Gynaecol 1994; 101(12): 1054–1063. 48. Korst LM, Phelan JP, Wang YM, Martin GI, Ahn MO. Acute fetal asphyxia and permanent brain injury: a retrospective analysis of current indicators. J Matern Fetal Med 1999; 8(3): 101–106. 49. Socol ML, Garcia PM, Riter S. Depressed Apgar scores, acid-base status, and neurologic outcome. Am J Obstet Gynecol 1994; 170(4): 991–998; discussion 8–9. 50. Pomerance J. Interpreting Umbilical Cord Blood Gases. Pasedena, CA: BNMG, 2004. www.cordgases.com 51. Garite TJ, Dildy GA, McNamara H et al. A multicenter controlled trial of fetal pulse oximetry in the intrapartum management of nonreassuring fetal heart rate patterns. Am J Obstet Gynecol 2000; 183(5): 1049–1058. 52. Dildy GA III. Fetal pulse oximetry: a critical appraisal. Best Pract Res Clin Obstet Gynaecol 2004; 18(3): 477–484. 53. ACOG Committee Opinion No. 258, September 2001. Fetal pulse oximetry. Obstet Gynecol 2001; 98(3): 523–524. 54. Porreco RP, Boehm FH, Dildy GA et al. Dystocia in nulliparous patients monitored with fetal pulse oximetry. Am J Obstet Gynecol 2004; 190(1): 113–117. 55. Bloom S. The MFMU Network randomized trial of pulse oximetry. Am J Obstet Gynecol 2006; 193(6): S2. 56. Klauser CK, Christensen EE, Chauhan SP et al. Use of fetal pulse oximetry among high-risk women in labor: a randomized clinical trial. Am J Obstet Gynecol 2005; 192(6): 1810–1817; discussion 7–9. 57. Rijnders RJ, Mol BW, Reuwer PJ, Drogtrop AP et al. Is the correlation between fetal oxygen saturation and blood pH sufficient for the use of fetal pulse oximetry? J Matern Fetal Neonatal Med 2002; 11(2): 80–83. 58. Kirkendall C, Phelan J. Ischemic fetal brain injury and normal fetal pulse oximetry. Obstet Gynecol 2002; 99: 628. 59. Schifrin BS. Is fetal pulse oximetry ready for clinical practice?: Writing for the CON position. Am J Matern Child Nurs 2003; 28(2): 65. 60. Arulkumaran S, Lilja H, Lindecrantz K, Ratnam SS et al. Fetal ECG waveform analysis should improve fetal surveillance in labour. J Perinat Med 1990; 18(1): 13–22. 61. Maclachlan NA, Spencer JA, Harding K, Arulkumaran S. Fetal acidaemia, the cardiotocograph and the T:QRS ratio of the fetal ECG in labour. Br J Obstet Gynaecol 1992; 99(1): 26–31. 62. Murphy KW, Russell V, Johnson P, Valente J. Clinical assessment of fetal electrocardiogram monitoring in labour. Br J Obstet Gynaecol 1992; 99(1): 32–37. 63. Luzietti R, Erkkola R, Hasbargen U, Mattsson LA et al. European Community multi-center trial, ‘Fetal ECG Analysis During Labor’: ST plus CTG analysis. J Perinat Med 1999; 27(6): 431– 440. 64. Amer-Wahlin I, Bordahl P, Eikeland T et al. ST analysis of the fetal electrocardiogram during labor: Nordic observational multicenter study. J Matern Fetal Neonatal Med 2002; 12(4): 260–266. 65. Westgate J, Harris M, Curnow JS, Greene KR. Randomised trial of cardiotocography alone or with ST waveform analysis for intrapartum monitoring. Lancet 1992; 340(8813): 194–198.

482

CH30

CLINICAL EPIDEMIOLOGY OF INTRAPARTUM FETAL MONITORING DEVICES

66. Amer-Wahlin I, Hellsten C, Noren H et al. Cardiotocography only vs. cardiotocography plus ST analysis of fetal electrocardiogram for intrapartum fetal monitoring: a Swedish randomised controlled trial. Lancet 2001; 358(9281): 534–538. 67. Noren H, Amer-Wahlin I, Hagberg H et al. Fetal electrocardiography in labor and neonatal outcome: data from the Swedish randomized controlled trial on intrapartum fetal monitoring. Am J Obstet Gynecol 2003; 188(1): 183–192. 68. Neilson JP. Fetal electrocardiogram (ECG) for fetal monitoring during labour. Cochrane Database Syst Rev 2003; 2: CD000116. 69. Dervaitis KL, Poole M, Schmidt G, Penava D et al. ST segment analysis of the fetal electrocardiogram plus electronic fetal heart rate monitoring in labor and its relationship to umbilical cord arterial blood gases. Am J Obstet Gynecol 2004; 191(3): 879–884. 70. Kwee A, van der Hoorn-van den Beld CW, Veerman J, Dekkers AH, Visser GH. STAN S21 fetal heart monitor for fetal surveillance during labor: an observational study in 637 patients. J Matern Fetal Neonatal Med 2004; 15(6): 400–407. 71. Haberstich R, Vayssiere C, David E et al. [Routine use of ST-segment of fetal electrocardiogram for monitoring labor. A year’s experience (preliminary results)]. Gynecol Obstet Fertil 2003; 31(10): 820–826. 72. Boog GJ. Fetal monitoring with the ST analyser: need for a long-term follow-up of the infants. Am J Obstet Gynecol 2005; 193(5): 1884; author reply 5. 73. Gagnon R, Dervaitis KL. Fetal monitoring with the ST analyser: need for a long-term follow-up of the infants. Am J Obstet Gynecol 2005; 193(5): 1885. 74. Luttkus AK, Noren H, Stupin JH et al. Fetal scalp pH and ST analysis of the fetal ECG as an adjunct to CTG. A multi-center, observational study. J Perinat Med 2004; 32(6): 486–494. 75. Rosen KG. Fetal electrocardiogram waveform analysis in labour. Curr Opin Obstet Gynecol 2005; 17(2): 147–150. 76. Noren H, Blad S, Carlsson A et al. STAN in clinical practice – the outcome of 2 years of regular use in the city of Gothenburg. Am J Obstet Gynecol 2006 (in press). 77. Quintero R, Meyers S, Schifrin B. Problems with intrapartum fetal heart rate monitoring interpretation and patient management. Obstet Gynecol 2003; 101(3): 617; author reply 8. 78. Schifrin BS, Ater S. Fetal hypoxic and ischemic injuries. Curr Opin Obstet Gynecol 2006; 18(2): 112–122. 79. Salamalekis E, Bakas P, Saloum I, Vitoratos N, Creatsas G. Severe variable decelerations and fetal pulse oximetry during the second stage of labor. Fetal Diagn Ther 2005; 20(1): 31–34. 80. Ingemarsson E, Ingemarsson I, Solum T, Westgren M. Influence of occiput posterior position on the fetal heart rate pattern. Obstet Gynecol 1980; 55(3): 301–304. 81. Sokol RJ. Occiput posterior position and FHR patterns. Obstet Gynecol 1981; 57(2): 266–267. 82. Pearl ML, Roberts JM, Laros RK, Hurd WW. Vaginal delivery from the persistent occiput posterior position. Influence on maternal and neonatal morbidity. J Reprod Med 1993; 38(12): 955–961. 83. Ponkey SE, Cohen AP, Heffner LJ, Lieberman E. Persistent fetal occiput posterior position: obstetric outcomes. Obstet Gynecol 2003; 101(5 pt 1): 915–920. 84. Fitzpatrick M, McQuillan K, O’Herlihy C. Influence of persistent occiput posterior position on delivery outcome. Obstet Gynecol 2001; 98(6): 1027–1031. 85. Akmal S, Kametas N, Tsoi E, Howard R, Nicolaides KH. Ultrasonographic occiput position in early labour in the prediction of caesarean section. Br J Obstet Gynaecol 2004; 111(6): 532–536.

31 The postmarket surveillance of medical devices: meeting the challenge Susan N. Gardner and Daniel Schultz US Food and Drug Administration, Rockville, MD, USA

One need only scan the broad spectrum of topics covered by this book – ranging from the materials used to fabricate medical devices to ethical considerations in studying their use – to recognize the breadth of the challenge faced by the US Food and Drug Administration (FDA) in assuring the safety of these products. Part of that challenge, of course, is met through a program of evaluation of high-risk, new devices prior to marketing. In this sense, the FDA operates the gate through which new devices must pass before they can be used in clinical practice – and it must be demonstrated that they are reasonably safe and effective before the gate opens and they are allowed onto the market. But the premarket evaluation program alone cannot assure continued safety and effectiveness of marketed devices. The need to get life-saving technology to the healthcare provider and the patient necessitates relatively small clinical studies, of short duration, prior to marketing. Therefore, it is only after marketing, when the product is used in much larger populations over a longer period of time, that many safety problems become evident. It may be years after the product has moved from clinical trials into community use before a significant number of patients are accrued to identify rare, but important, adverse events, and the effects of long-term implants may not become apparent until years following implant [1,2]. Also, the practitioner and patient populations using and receiving the device change significantly after it is marketed, and this can have a profound impact on the nature and frequency of adverse events. In premarket clinical trials, the clinicians participating in a Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

484

CH31

THE POSTMARKET SURVEILLANCE OF MEDICAL DEVICES

clinical trial are well trained in using the device and alert to its possible limitations, and patients are carefully selected for a limited set of medical indications. Once the device is in widespread use by less-skilled or less-trained practitioners and on a broader population of patients, adverse effects can show up that were not apparent prior to marketing. There are other challenges in carrying out a postmarket surveillance program for devices that stem from certain characteristics distinctive to the manufacture and use of devices – characteristics that may not be shared by the drugs and vaccines also regulated by FDA, as described in Chapter 3 of this book. For example, most drug products consist of a distinct molecular entity that remains essentially unchanged over the product’s lifetime. Manufacturers of devices, on the other hand, often make incremental changes to their products, sometimes in response to the concerns of clinicians. And so it can be difficult to determine which version of a device is under question as we try to evaluate its safety. Another factor that poses a special challenge in assessing the safety of medical devices is the role of the user. The ICU nurse faced with multiple patient monitoring devices, the anesthesiologist working with a complex gas delivery device, and the elderly diabetic trying to read a glucose monitor, all play a critical role in determining device safety. When investigating an adverse outcome in device use, we are often faced with the question, ‘Was this incident caused by the device itself or was it caused by the person using it?’. Although the answer may be ‘both’, it is important that we dissect and reconstruct each safety issue so as to tease apart the primary sources of the problem. With this information, we can feed product problems back to manufacturers for future design improvements and develop effective educational vehicles for users. The postmarket surveillance program currently in place in the FDA responds in large measure to all of these challenges. Through our Medical Device Reporting (MDR) system (see Chapter 2), we are receiving almost 200 000 reports of adverse events each year, and our Medical Product Surveillance Network (MedSun) is providing more detailed information from a subset of healthcare facilities enrolled nationwide (see Chapter 5). Based in part on the information provided by these systems, we alert manufacturers and users to problems as they arise, work with device users and the manufacturer to assess and understand the problem and, when necessary, work with manufacturers to recall devices whose safety problems cannot be resolved by other means. In acting on the information provided by our postmarket surveillance systems, we must be aware of the potential negative consequences of what we do. For example, if we were to remove from the market the only product available to treat a critical condition because we discovered that it could produce a relatively rare adverse event, this could do more harm to patients than allowing it to remain on the market and working to correct the problem through other means, such as labeling changes, training programs, or safety messages. And if our actions were to result in patients foregoing needed treatment, or opting for riskier alternatives, this too would compromise public health. Major difficulties remain in our quest to develop an optimally effective postmarket surveillance program, as discussed in Chapter 4. One is the ‘denominator’ problem. We often do not have access to information on the actual number of a particular device in use, and so we cannot calculate the rate at which the product may be failing or malfunctioning.

THE POSTMARKET SURVEILLANCE OF MEDICAL DEVICES

485

For example, if a particular adverse event is occurring five times more frequently with product A than with product B, one might think, on the face of it, that a warning to users of product A or a corrective action might be in order. But if product A is used five times more frequently than product B, its failure rate may actually be unremarkable. Another obstacle on the road to an optimally effective postmarketing program is the lack of unique device identifiers (see Chapter 4). Too many of the adverse event reports we receive from clinicians and patients lack vital information on the manufacturer, the model, the serial number, and sometimes even the name of the device. Without this information, it can be very difficult to analyze – or even identify – the problem. Unfortunately, the information is often found on the device’s packaging, which users are prone to discard. A universal system of device nomenclature and permanent labeling would be a major milestone in improving adverse event reporting, and is a goal we are pursuing (see Chapter 7). In the shorter term, FDA is taking several important steps to enhance the postmarket surveillance of medical devices. For example, the MedSun program, which provides active and highly focused monitoring of devices in selected hospitals across the nation, has been expanded and enhanced (see Chapter 5). Data-mining utilities, an analytic technique already in use for the monitoring of drugs, is now being explored for devices (see Chapter 8). The National Electronic Injury Surveillance System (NEISS), which collects data on product-related injuries treated in emergency rooms, is being used as a means of gathering intelligence on medical device adverse events (see Chapter 6). A center specifically devoted to devices has recently been added to the Center for Education and Research on Therapeutics (CERTS), a consortium of academic centers conducting research in medical care [3], and FDA is strengthening the requirements and monitoring for mandated post-approval studies of devices conducted by manufacturers [4]. The need for an effective postmarket surveillance program for medical devices is not new (see Chapter 4). We know that postmarket surveillance can identify safety problems not foreseen during the premarketing phase, that it can help characterize at-risk subgroups of patients, that it can lead to life-saving corrective actions by regulatory agencies and clinicians, and that it can help manufacturers improve the safety and effectiveness of their products. A postmarket surveillance system that enables continual learning and feedback throughout the product life cycle will foster innovation and support the best medical practice. Achieving such a system will require significant progress towards the improvements mentioned above, along with improved collaboration and communication among all stakeholders. The growing contribution of the discipline of epidemiology will be critical in generating the information needed to advance medical device safety and effectiveness, both before and after marketing. Epidemiology and surveillance are key components of a healthy postmarket system. As the Center for Devices and Radiological Health (CDRH) stretches to reach its full potential for monitoring marketed medical devices, the growing contribution of the discipline of epidemiology and the practice of surveillance will be critical in generating the information needed to advance medical device safety and effectiveness, both before and after marketing.

486

CH31

THE POSTMARKET SURVEILLANCE OF MEDICAL DEVICES

References 1. Brown SL, Bright RA, Tavris DR. Medical device epidemiology and surveillance: patient safety is the bottom line. Expert Rev Med Devices 2004; 1: 1–2. 2. O’Shea JC, Kramer JM, Califf RM, Peterson ED. Part I. Identifying holes in the safety net. Am Heart J 2004; 147: 977–984. 3. AHRQ expands therapeutics education and research network to focus on critical issues facing the healthcare system. Centers for Education and Research on Therapeutics, April 25 2006: http:// www.certs.hhs.gov/whats_new/archive/20060425_01.html [accessed May 2006]. 4. Schultz D. Ensuring the safety of marketed devices. Center for Devices and Radiological Health, Food and Drug Administration, January 4 2006: http://www.fda.gov/cdrh/postmarket/mdpi.html [accessed May 2006].

Index Note: page numbers in italics refer to figures and tables abdominal aortic aneurysms 355–63 endovascular graft postmarket assessment 358–60 endovascular repair 356, 357–8 indications for treatment 355–6 mortality rate 358, 359 natural history 355–6 outcomes of treatment 357–8 risk:benefit profile 360 rupture risk 357 stent-grafts 356, 358–63 surgery 356, 357 ABT-578 sirolimus analog 340–1, 343–4 Acanthamoeba keratitis 433 access, equitable to medical devices 376 hip implants 451 accredited persons 198–9 acidemia, metabolic 475 acidosis, fetal/infant 471 active implantable medical devices (AIMDs) 178–9, 182 active medical device 179 acupuncture adverse event reports 325–6 safety 325 acupuncture needles 322–6 embedded auricular 325–6 manufacturers 324 1 Adept Registry for Clinical Evaluation (ARIEL) 110–11 adolescence, corrective surgery 209–10 AdvaMed trade association 194, 200 Advance Bionics Corporation 396–7

adverse event(s) reduction 24, 26 reports 13 industry use 165–6 adverse event reporting 11–14 data elements 294 data management 163 FDA 160–3 healthcare providers 162–3 improvement 196 industry use of information 165–6 information sources 161–2 underreporting 50–1, 64, 161–2 adverse medical device events (AMDE) 24, 26 data collection 57 data elements of reports 294 estimate production 81–2 frequency 36 indicators 53–4 multiple types 36 postmarket studies 155–6 public health burden 83–4 rates 27 records 51 registries 52–3 reporting 47–8, 49, 50–2 enhanced 52 surveillance 43–58, 47–8, 49, 50–2, 291–2 conditions 54–5 program 55–7 underreporting 50–1, 64

Medical Device Epidemiology and Surveillance Edited by S. Lori Brown, Roselie A. Bright and Dale R. Tavris # 2007 John Wiley & Sons, Ltd ISBNs: 0 470 85275 5 (cased) 0 470 85276 3 (Pbk)

488 Agency for Healthcare Research and Quality (AHRQ) databases 114 technology assessment program 118 alternative and complementary medical devices 319–32 acupuncture needles 322–6 adverse events 331–2 ear candles 326–8 magnets 328–31 Alternative Summary Reporting (ASR) database 411–12 American College of Cardiology (ACC), National Cardiovascular Data Registry (NCDR) 106 American College of Epidemiology (ACE) guidelines 136 American Registry of Pathology 110 American Society of Testing and Materials (ASTM) 7 AneuRx stent-graft 358, 359 Angio-Seal Hemostatic Puncture Closure Device 381, 382, 387, 390 annual reports 46 anticoagulant therapy heart valves 369 hemostasis devices 380, 381 hip implants 447–8 suture devices 381 anti-insulin antibodies 230 antiproliferative agents, drug-eluting coronary stents 336–42, 343–4 aortic valves 367–77 access issues 376 device reprogramming 375–6 elderly patients 371–2 operative mortality 370–2, 375 patient tracking 375–6 postmarket performance 374 risk factors 371 withdrawn 374–6 Armed Forces Institute of Pathology registry 110 Association for Advancement of Medical Instrumentation (AAMI) 7 auscultation, intermittent 466, 467–8 Bacillus cereus, ventilator-related outbreaks 248–9

INDEX

balloon catheter, procoagulant combination 381 Belmont Report 130 bench testing 149, 150 beneficence principle 129 bias 276 bileaflet valve 368 billing patterns 173–5 bioethics foundations 128 Biolimus A9 341 blood gas analysis, fetal 470–1 body mass index (BMI), hip implants 444 bone cement 443–4 bovine components of medical devices 265, 266–7, 267 bovine spongiform encephalitis (BSE) 259, 260 transmission risk minimization 267–8 breast implants adolescents 210 augmentation 410–11, 421 capsular contracture 414 complications 419–20 congenital anomalies 416 connective tissue disease 420–1 FDA surveillance 411–21 infections 414 mammography 414–15, 417 maternal–child problems 415–16 medical purposes 410–11 psychotropic drug use 421 regulation 409–11 re-operation after mammoplasty 419–20 rupture 416–19 safety concerns 409 saline-filled 409, 414, 417 soybean oil 408 suicide 421 tissue expanders 414 types 407–9 see also silicone gel breast implants breast pumps 248, 249 bronchoscopes cleaning 242–3 disease outbreak 239, 240–1, 242–3 disinfection 242–3 material failure 275 recall 197

INDEX Burkholderia cepacia 252 pediatric patients 251 caged-ball valve 367, 368 cancer, hip implant association 448 carbon dioxide partial pressure detector 251, 252 cardiac catheterization 380 cardiac monitors, electromagnetic interference 296, 298 cardiac pacemakers, electromagnetic interference 303, 304, 306, 308, 310, 311, 312 cardiovascular devices aortic valves 367–77 electromagnetic interference 293, 295–6 cardioverter defibrillators see implantable cardioverter defibrillators care, standards of 46 case-control studies 37, 435–6 orthopedic devices 458 cataract 427–8 CE marking 181, 182, 184 cellular phones, electromagnetic interference 311 Center for Devices and Radiological Health (CDRH) 5 adverse event report data management 163 epidemiology program 100 postmarket surveillance 159 Center for Education and Research on Therapeutics (CERTS) 485 Centers for Disease Control and Prevention (CDC) meningitis with cochlear implants 399 National Center for Health Statistics 115–18 surveillance systems 104 Centers for Medicare and Medicaid Services (CMS) databases 113 mandated registries 110 ceramics, orthopedic materials 443 cesarean section rate 467 cessation of use 29 children clinical studies 204 device-related disease 251 meningitis association with cochlear implants 397 orthopedic devices 207–9

489

pediatric medical devices 203–15 radiation exposure 212 thermometers 205–6 CLARION cochlear implant models 396–7 Clarke Error Grid 226–7 Class I devices 8, 182 premarket notification 9, 10 Class II devices 8 postmarket surveillance 153 premarket notification 9, 10 regulatory changes 198–9 Section 522 studies 154 tracking 16 Class IIa devices 182 Class IIb devices 182 Class III devices 8, 182 clinical trial information 10 cochlear implants 396 postmarket surveillance 153 premarket approval application 9–10 premarket notification 9 regulatory changes 198–9 Section 522 studies 154 tracking 16 classification of medical devices 8, 91 Europe 182–3 cleaning bronchoscopes 242–3 endoscopes 243 clinical data 10 clinical epidemiologic studies 138–40 clinical research studies 127 clinical studies 150–2 children 204 design 151–2 clinical trials information for premarket notification 10 premarket 483–4 cobalt–chrome alloys 441, 442 cochlear implants 13, 395–405 DC current leakage 403, 404 device failures 401, 402–4 Good Manufacturing Practice 401–2 hazard evaluation 402–4 hermeticity failures 401, 402–4 intended use 395–6 investigations 397–400 manufacturer Complaint Report Database 403–4

490

INDEX

cochlear implants (continued ) material failure 275 meningitis association 396–404 patient harm assessment 404 positioner component 400, 401 recommendations 400–1 tissue reactions 277 codes product 91 unique 54, 90, 485 see also product coding systems cohort studies 36–7 collagen plug sealant devices 381–2, 388, 389–90 COMBI 40þ cochlear implant 401 Common Rule 141 Competent Authorities (CAs) 183, 184, 185 Complaint Report Database 403–4 complementary medical devices see alternative and complementary medical devices complex devices 29 human factors 31 complex multi-device situations 29–30 computed tomography (CT), pediatric 212 computerization, confidentiality 141–2 Concept Permanence 90 condition of approval (CoA) studies 154, 155, 164 confidentiality computerization 141–2 databases 119 violation in published reports 166 Conformity Assessment Bodies (CABs) 181, 188–9 confounding by indication 276 congenital anomalies, breast implants 416 Congressional Black Caucus (US Congress) 195 connective tissue disease, breast implants 420–1 consensus standards, national/international 7 consumer guides 196 Consumer Product Safety Commission (CPSC) 103 consumer protection 6 consumers 187–200 accomplishments 199–200 concerns 193–7 manufacturers 193–4 regulatory safeguards 194–5 safety of implants 195–7

legislative change requests 198–9 organizations 188 regulatory change requests 198–9 regulatory mechanism recommendations 197–9 contact lenses 274, 427 case-control studies 435–6 continuous wear 437 corneal ulcers 430–7 daily wear 432 extended wear 432, 434, 435 hyperpermeable 434 pre-/post-approval studies 434 relative risk 434–5 tissue reaction 278 Continuous Glucose Monitoring System (CGMS) 225–8, 231 continuous subcutaneous insulin perfusion (CSII) 220, 223–4 disadvantages 224–5 contrast medium-related malaria 250–1, 252 controlled medical vocabulary 90 corneal scarring 431 corneal transplant, prion disease transmission 261–2 corneal ulcers contact lenses 430–7 infectious/non-infectious 436–7 coronary stents, bare metal 337–8, 346 coronary stents, drug-eluting 335–50 antiproliferative agents 336–42, 343–4 biocompatible coatings 336 comparative trials 342, 345 complex coronary lesions 345–6 postapproval requirements 346–7 postmarket adverse events 348–50 regulation 346–50 safety 346 corrections 17–18 cosmetic surgery adolescence 210 see also breast implants, augmentation Cosmetic Surgery National Data Bank 108 Cosmetic Surgery Telephone Survey 107 Council for International Organizations of Medical Sciences (CIOMS) guidelines 135, 141 Creutzfeldt–Jakob disease (CJD) 259 iatrogenic transmission 260–1

INDEX new variant 259, 260 surgical instrument transmission 260–1, 262, 263, 264, 265, 267 transplant tissue transmission 262 critical care areas 32, 34 cupping 324 Dalkon Shield 189, 190 Danish Hip Arthroplasty Register 447 Danish Registry for Plastic Surgery of the Breast 111–12, 192 data collection 105 event recovery 165 elements of adverse event reports 294 insurance claims 173 management and adverse event reporting 163 national surveys 115–18 sources 99–120 submission from manufacturers 197 see also registries data mining 119, 167 databases 100–1 academic 115 adverse event reporting 162 automated large administrative 112–15 Complaint Report 403–4 confidentiality 119 European 184, 185 growing capabilities 119 privacy issues 119 private healthcare 113 surveillance 101–4 technology assessment 118 use in epidemiology studies 140–2 see also named databases dawn phenomenon 224 Declaration of Geneva 128 Declaration of Helsinki 128 decontamination of surgical instruments 263–5, 266–7, 267 defibrillators external 296 see also implantable cardioverter defibrillators Device Experience Network (DEN) database see Medical Device Reporting (MDR) database Device Safety-Exchange (DS-X) 70, 72

491

device use 32 DeviceNet design 64, 65–6, 67–8 implementation 68 di-2-ethylhexylphthalate (DEHP) 211 diabetes mellitus control 227 devices used in management 219–31 diabetic ketoacidosis (DKA) 225, 230 diabetic neuropathy, magnet therapy 329–30 diagnosis codes 26 Dialysis Surveillance Network 104 diseases breast pump-related 248, 249 bronchoscope-related 239, 240–1, 242–3 endoscope-related 240–1, 243–4 hemodialysis-related 244, 245, 246–7 medical device-related 237–53 needle-related 250 neonatal/pediatric intensive care unit outbreaks 247–50 prevention 24, 25 syringe-related 250 tongue depressor-related 248, 250 treatment 24, 25 ventilator-related outbreaks 248–50 disinfection bronchoscopes 242–3 endoscopes 243 surgical instruments 263–5, 266–7, 267 disk valve 368 documentation, device use 32, 53, 55, 57, 76 Drug Safety Oversight Board 167–8 Duett sealing device 381, 382 dura mater transplantation 264 Dynamic Registry 109 ear candles 326–8 complications 327–8 ear infections 206–7 EC declaration of conformity 181 economics of medical devices 148 effectiveness checks 18 elderly patients aortic valve replacement 371–2 hip implants 450–1 intraocular lenses 429 electrical acupuncture point detector 324 electro-acupuncture stimulator 323, 326

492

INDEX

electrocardiography (ECG), fetal waveform monitoring 473–6 electromagnetic compatibility 292 electromagnetic interference with medical devices 291–316 cardiovascular devices 293, 295–6 causes 311–13 coding problems 314, 315 deaths 303, 306–7 device types 303 FDA reports 297–316 implantable cardioverter defibrillators 303, 304–5 MRI devices 298, 305, 308–9, 310 other medical devices 312 reports 298, 299, 300–5 sources 303, 305, 309 statistical procedures 297–8 tip-of-the-iceberg paradigm 314, 315 electronic fetal monitoring (EFM) 464–70 antepartum surveillance 469 device 465–6 fetal ECG waveform monitoring 473–6 intervention rate 477 risk for fetal distress 472–3 methodology 468–9 nonreassuring fetal heart rate patterns 472–3 studies 466–8 surveillance parameters 477 technical difficulties 469–70 electronic muscle stimulators 320, 321, 322 Emergency Care Research Institute (ECRI), technology assessment databases 118, 167 emergency departments, NEISS 79–80, 81, 83, 103–4 endoscopes, GI 240–1, 243–4 endotoxins, hemodialysis-related 244, 245, 246 enforcement activities 15–18 environmental factors 30 epidemiology 1, 2, 3, 21–38 ethical guidelines 134 importance 24, 26 epidemiology studies clinical 138–40 data sources 99–120

database use 140–2 descriptive 35–6 design 35–7 cultural features 32 device features 26–32, 33–4 intrinsic features 28–32 regulatory features 27–8 ethical requirements 136–42 heart valve replacement 373–6 independent review 139–40 informed consent 140 materials for devices 273–6, 277–8, 278 medical records use 140–2 postmarket 156 respect for subjects 140 survey use 140–2 epinephrine, acute poisoning 311–12 essential requirements (EU) 180 ethic(s) 127–42 ethical codes 128 ethical guidelines epidemiologic 134 professional 133–6 requirements for epidemiology studies 136–42 ethnic differences 195 EUDAMED database 184, 185 European Economic Area (EEA), regulation of medical devices 93 European Union (EU) 177–85 classification of medical devices 182–3 consumer issues 188–9 harmonization process 180, 188–9 medical device groups 178 Member States Competent Authorities 183, 184, 185 regulation of medical devices 93 traceability of medical devices 184 vigilance system 184–5 event recovery data 165 Everolimus 341, 344 experts, internal 13 exports of medical devices 148, 149 exposure to devices 32 failure of device, reporting 48 Fair Access to Clinical Trials Act (2005) 168 FDA Modernization Act (1997) 162

INDEX femoral artery catheterization site complications 388 hemorrhage 383, 384–5 postprocedure hemostasis 379 femoral head, aseptic necrosis 444 fetal distress in labor 467 intervention risk 472–3 operative deliveries 474 fetal ECG waveform monitoring 473–6 fetal monitoring, intrapartum 463–78 epidemiological assessment 466–8 modalities 476–8 fetal pulse oximetry 471–3 fetus blood gas analysis 470–1 see also electronic fetal monitoring (EFM); heart rate, fetal Fibroid Registry for Outcomes Data (FIBROID) registry 108 fibromyalgia 420, 421 field inspections 17–18 Finnish Arthroplasty Registry 112 fixation devices 208, 209 fomites 238 Food, Drug and Cosmetics Act 164 see also Section 522 Food and Drug Administration (FDA) 5 academic opportunities 167–8 adverse event reporting 160–3 breast implant regulation 409–11 surveillance 411–21 BSE transmission risk minimization 267–8 causality assessment 13–14 cochlear implant investigations 397–400 contact lens regulation 431 corrections 17–18 data mining 119, 167 database of device adverse events 12 diabetes device regulation 221 drug-eluting coronary stent regulation 346–7 electromagnetic interference reports 297–316 guidance 7–8 heart valve replacement guidance 372 hemostasis device evaluation 387–9 postmarket performance 382–7 safety study 389–90

493

human subjects protection regulations 131–2, 133 intraocular lens regulation 428–30 MED-EL cochlear implant recommendations 404 NEISS long-term applications 84 passive reporting system 95 pediatric medical devices 213–14 postmarket tools 11–15, 199–200 publications 166, 167 regulation of medical devices 93 regulatory controls 6, 194–5 removals 17–18 responsibility 6 scientific article writing 166 stent-graft postmarket assessment 358–60 surveillance program 45 surveillance systems 101–3 tools 46–7 tracking program 16–17 see also Medical Device Reporting (MDR) database Food and Drug Modernization Act (1997) 68 fractures, children 208, 209 General and Plastic Surgery Devices Panel 410 general controls 6–7 general medical devices 178, 182 glargine 224 Global Harmonization Task Force (GHTF) 103 Global Medical Device Nomenclature (GMDN) 92 gloves, medical 273 latex 279–82, 283–5, 285–6 sensitivity 278 starch peritonitis 251, 252 tissue reactions 277 glucose blood level control 220, 224, 229 continuous monitoring devices 225–8 fluctuations 229 GlucoWatch biographer 226, 227, 231 Good Clinical Practice (GCP) guidelines 133 Good Manufacturing Practice (GMP), cochlear implants 401–2 government agencies databases 113–14 government regulatory agencies 196, 198 Gram-negative bacteria, hemodialysis-related outbreak 245, 246

494

INDEX

grommets 206–7 growth plates, mechanical 209 gynecomastia 210–11 Health and Human Services Department (HHS, US) 128, 129–30 human subject protection regulatory requirements 130–1, 132, 133 Health Insurance Portability and Accountability Act (1996) 141 health regulatory agencies 198 Health Risk Assessment (HRA), cochlear implant failures 403 Healthcare Cost and Utilization Project (HCUP) 114 healthcare institutions, device management 94 healthcare providers adverse event reporting 162–3 statistical samples 57 heart rate, fetal 463–4 abnormal rhythms 466 analysis 474 decelerations 477 electronic monitoring 464–70 nomenclature for tracings 469 nonreassuring pattern 471–2, 476 patterns 468, 469, 477 rhythm 465–6 technical difficulties 469–70 heart valves access issues 376 adverse event reports 373 allograft 369–70 animal tissue 369 anticoagulant therapy 369 autograft 369 bio-prosthetic 369–70 complications of replacement 373–5 device reprogramming 375–6 epidemiological studies for replacement 373–6 failure 275 FDA guidance for replacement 372 homograft 369–70 human tissue 369–70 mechanical 367–9 objective performance criteria for replacement 372 patient tracking 375–6

tissue 369–70 tissue reaction 278 see also aortic valves hemodialysis disease outbreaks 244, 245, 246–7 membrane materials 274, 277, 278 hemoglobin A1c (HbA1C) levels 226, 227, 229 testing 220 hemostasis, postprocedure of femoral artery 379 hemostasis devices 379–91 adverse events 37, 382–7, 388–9 collagen plug sealant devices 381–2, 388, 389–90 complication rates 388 FDA study 387–9 femoral artery hemorrhage 383, 384–5 gender differences 383–5 hematoma 383, 384–5 infection 383 meta-analyses 386–7 non-invasive patches 382 postmarket performance 382–7 protective effects 389 registry studies 37 risk evaluation study 387–9 safety assessment study 389–90 safety issues 386–7 suture 380–1, 388 types 380–2 women relative risk 383–5 hepatitis B hemodialysis-related outbreak 245, 246–7 needle-related 250, 252 hernia repair 210 hip arthroplasty, total 444–5 complications 447–8 device survival 446–7 operative mortality 445–6 rates 451 hip implants adverse event reports 448–50 cancer association 448 complications 447–8 costs 451 elderly patients 450–1 heterotopic ossification 448 methodological issues 457–8 rates 451

INDEX retrieval analysis 447–8 revision 447–8 risk factors 444–5 study design 457–8 thromboembolism 447–8 trends 450–1 wear rates 447 hip prostheses 274, 442 tissue reaction 278 HIV infection hemodialysis-related outbreak 244, 245 needle-related 250 hospitals, surveillance systems 175 human factors 30–1 surveillance studies 48 human subjects information 141 informed consent 132, 133 protection 127–8 regulations 129–30, 131–2, 133 regulatory requirements 130–1 respect for 140 hybrid studies 37 hyperglycemia 225, 230 hypodermic needles, disease outbreaks 250 hypoglycemia 225 control 228 insulin pumps 224 hypothesis-testing studies 36–7 identifier, unique 54, 90, 485 implantable cardioverter defibrillators 24 electromagnetic interference 298, 303, 304–5, 311, 312 registry 106 tracking 16–17 in vitro diagnostic medical devices (IVDs) 178, 179, 182–3 Independent Review Boards (IRB) 137–8 epidemiology studies 139–40 Industrial Epidemiology Forum (IEF) guidelines 134–5 industry, use of information for adverse event reports 165–6 infants acidosis 471 blood gas analysis 470–1 seizures 467

495

infections breast implants 414 hemostasis devices 383 late-onset 208 information industry use from adverse event reports 165–6 research participants 141 sources for adverse event reporting 161–2 informed consent 132, 133 epidemiology studies 140 infusion pumps, external 223–5 inguinal hernia 210 Institute of Medicine (IOM), pediatric medical devices 213–14 insulin 219–20 closed loop systems 230–1 concentrated 230 delivery devices 221–5 external infusion pumps 223–5 implantable pumps 228–30 jet injectors 222–3 long-acting 224 pen device 221–2 sensors 231 therapy 220 insurance claims data 173 International Epidemiologic Association (IEA) guidelines 135 International Society for Environmental Epidemiology (ISEE) guidelines 135 international vigilance surveillance system 103 intervertebral disc, degenerative disease 452–3 intervertebral disc replacement 452–6 effectiveness 454–6 implant survivability 454 methodological issues 457–8 patient factors 455 patient outcome 454–5 perioperative complications 455–6 radiographic assessment 455 safety assessment 454–6 study design 457–8 surgical risk 455 intraocular lenses 427 access issues 376 epidemiology 427–30 outcomes 428–9, 430 younger patients 429–30

496 intrauterine contraceptive device (IUD) 189 Investigational Device Exemption (IDE) regulation 137 investigational plans 151 jet injectors 222–3 disease outbreaks 250, 252 joint replacement costs 451 justice principle 129–30 keratitis, microbial 431–2, 433–5, 436–7 Kids’ Inpatient Database (KID) 114 Klebsiella oxytaca related outbreaks 248, 249 Klebsiella pneumoniae, hemodialysis-related outbreak 244, 245 kuru 259 labeling 6, 10 requirements 46 labor fetal distress 467 long 477–8 lasers 187 latex allergy 280, 281, 282, 283–5, 285–6 latex gloves 279–82, 283–5, 285–6 sensitivity 278 starch peritonitis 251, 252 legislation change and consumer requests 198–9 pediatric medical devices 213–14 liposuction disease outbreak 252 ultrasound-assisted 210–11 listing 6 literature review 171–3 long-term implantable insulin pump (LIIP) 228–30 lumbar disc degeneration 453 magnet therapy 328–31 magnetic resonance imaging (MRI) electromagnetic interference 298, 305, 308–9, 310 embedded auricular needles 325–6 malaria, contrast medium-related 250–1, 252 mammography, breast implants 414–15, 417 mammoplasty, re-operations 419–20 manual compression 379 hemostasis device comparison 386

INDEX

Manufacturer and User Facility Device Experience Database (MAUDE) 101–2, 162, 291–2 breast implant surveillance 411, 412, 415 cochlear implant investigations 398 electromagnetic interference reports 296, 297, 311 intraocular lenses 430 problem codes 313–14 reports 292–3 manufacturers acupuncture needles 324 ADME reports 293, 301–2 Complaint Report Database 403–4 consumer concerns 193–4, 197 data submission 197 drug-eluting coronary stent regulation 347 ear candles 326–7 inspections 13 use of information for adverse event reports 165–6 manufacturing practices 7 market for medical devices 147–8 export 148, 149 growth potential 148 market surveillance, European 183 marketing applications 8–10 materials for devices 32, 273–86 epidemiology 276, 278 failures 275 surveillance 276, 278 tissue reactions 276, 277–8 see also latex allergy MED-EL cochlear implants 401, 402–4 Medical Device Reporting (MDR) database 293, 296, 412 breast implant surveillance 411 intraocular lenses 430 postmarketing studies 155 surveillance 291 Medical Device Reporting (MDR) system 81 adverse event reports 484 hip implant reports 448–50 regulation 12 Medical Device User Fee and Modernization Act (2002) 198–9, 203, 213 medical devices definition 1–2, 5–6, 22–3 EU 178–9

INDEX incremental changes 484 scope 71 terminology 27 Medical Product Surveillance/Safety Network (MedSun) 52, 63–76, 103 adverse event reporting 162–3, 484 current status 71–4 design 69–70, 71 enhanced reporting 52 epidemiology 75–6 facilities 72–3, 76 feedback 70 networks 74 postmarketing surveillance 485 reports 71–2, 73 safe use promotion 74–5 surveillance 54 training 70 underreporting 51 medical records, use in epidemiology studies 140–2 MedWatch 56–7, 162, 291 meningitis, incidence 401 meningitis with cochlear implants 396–404 epidemiology 399–400 FDA response 397–401 mercury amalgam tooth fillings, tissue reaction 278 metabolic acidemia prediction 475 metals, orthopedic devices 441–3 Methylbacterium mesosphilicum, bronchoscope-related outbreak 240, 242 microbial keratitis 431–2, 433–5, 436–7 milk contamination 248, 249 MiniMed Continuous Glucose Monitoring System 225–8, 231 mobile phones, electromagnetic interference 311 monitoring of devices 44, 46 moxibustion 322, 324 Multi-Item Gamma Position Shrinker (MGPS) 119 multiple daily insulin (MDI) 220 Mutual Recognition Agreements (MRAs) 181 Mycobacterium chelonae, bronchoscope-related outbreak 240–1, 242 Mycobacterium tuberculosis, bronchoscope-related outbreak 239, 240–1, 242

497

naming of devices 88–9 National Center for Health Statistics 115–18 National Clearinghouse of Plastic Surgery Statistics 107 National Drug Code (NDC) 173, 174 Directory 94 National Electronic Injury Surveillance System (NEISS) 54, 67, 79–84, 103–4 data collection/analysis 81–2 limitations 81 long-term applications 84 medical device surveillance utilization 83–4 medical device-associated adverse event estimates 81–2 non-sampling errors 81 postmarketing surveillance 485 sampling errors 81 uses 80–1 National Eyecare Outcome Network (NEON) database 429–30 National Health Interview Survey 115 National Healthcare Survey (NHCS) 117 National Home and Hospice Care Survey (NHHCS) 117–18 National ICD Registry 106 National Institutes of Health (NIH) registries 108–10 National Maternal and Infant Health Survey (NMIHS, 1998) 116 National Medical Association 195 National Mortality Followback Survey (1993) 115–16 National Nosocomial Infections Surveillance Systems (NNIS) 104 National Research Center (NRC) for Women and Families 188, 195 Nationwide Inpatient Sample (NIS) 114 neonatal intensive care unit, disease outbreaks 247–50 neurological instruments, CJD transmission 260–1 New Approach concept (EU) 178, 179–81 New Approaches to Coronary Interventions registry 108–9 nomenclature of medical devices 87–96 applications 93–5 attributes 90 commercial applications 94 current terminologies 91–2

498

INDEX

nomenclature of medical devices (continued ) definition 90 epidemiology of devices 94–5 hierarchies 89 management of devices 94 naming of devices 88–9 quality indicator 90 relationship supporting 89–90 relationships of devices 90 specificity level of devices 88–9 surveillance of devices 94–5 technical elements 87–90 UMDNSTM 91–2 non-significant risk device study 137–8 Notified Bodies 180–1 Nuremberg Code 128 obesity, osteoarthritis risk 444 objective performance criteria, heart valve replacement 372 Office of Device Evaluation 163 cochlear implants 402 Office of Surveillance and Biometrics (OSB) 164 operating rooms 32, 33 operator error see user error ophthalmic devices 427–38 ORA inspection, cochlear implants 401–2 orthopedic devices 207–9, 441–59 data sources 456–7 effectiveness 458 intervertebral disc replacement 452–6 materials 441–4 methodological issues 457–8 safety 458 study design 457–8 ossification, heterotopic 448 osteoarthritis hip 444 spine 452–3 otitis media with effusion 206–7 oximetry, fetal pulse 471–3 pacemakers see cardiac pacemakers paclitaxel, drug-eluting coronary stents 338–40, 341–2, 343, 344 postmarket adverse events 349 pain, magnet therapy 329, 330–1 pancreas, artificial mechanical 230–1

patches, non-invasive 382 Patient and Consumer Coalition 196 patient tracking, heart valves 375–6 pediatric intensive care units, disease outbreaks 247–50 pediatric medical devices 203–15 regulatory framework 212–14 peer review 166 Perclose devices 380–1, 382, 387 percutaneous coronary intervention manual compression 379 stenting 335–6 performance failure of devices 160 pharmacoepidemiology 171–5 pharmacy reimbursement 173 Plasmodium falciparum, contrast medium-related 250–1, 252 polyethylene 443 ultra-high molecular weight 443 polymers, orthopedic materials 443–4 polymethylmethacrylate (PMMA) 443–4 polyurethane breakdown products 278 post-approval studies 46, 197 postmarket condition of approval studies 163–5 postmarket enforcement 15–18 postmarket oversight 2, 3, 10–18 clinical data 10 postmarket performance, hemostasis devices 382–7 postmarket studies 148, 152–6 adverse event reports 155–6 condition of approval 154, 155, 164 consumer advocates 196–7 epidemiologic 156 follow-up 152–3 mandated 14–15 Medical Device Reporting 155 Section 522 154–5, 164–5 postmarket surveillance 2, 11–15, 46, 483–5 academic perspective 159–68 Europe 183 program 484–5 studies 47, 153 preclinical studies 148, 149–50 premarket agreements 197 premarket application 9–10 clinical data 10, 11 conditions of approval 14

INDEX Pre-Market Approval (PMA) process application 9–10 cochlear implants 396 diabetes device regulation 221 premarket notification 7, 9 premarket review 6–8 premarket studies 46, 148–53 premarket submissions, pediatric devices 212–13 pressure-equalizing tubes 206–7 prevention options 24, 25 prions 259–68 iatrogenic transmission 260–1 medical devices with tissue component 264–5, 266–7, 267 proteins 260 transplant tissue transmission 261–2 privacy, databases 119 Privacy Rule 141 private healthcare databases 113 procedure codes 32 reimbursement 174 product codes 91 product coding systems 28, 50, 54, 55, 76 identification 28 unique codes 54, 90, 485 product development protocol (PDP) 9 product life cycle 148–53 product problem reporting 11–14 product recall 165 proplast TMJ implants 193 prostheses, implanted 23 Prosthetic Titanium Rib implant 209 Pseudomonas aeruginosa 253 bronchoscope-related 243 Pseudomonas cepacia 253 public health 6 burden of adverse events 83–4 disease outbreaks 238 epidemiologic studies 100 interventions 75–6 notifications 13 publications 166, 167 pulsed magnetic fields 328–31 pyrogenic reactions, hemodialysis-related 244, 245, 246 Quality Systems regulation 6, 7, 14, 46 cochlear implants 401–2

499

QuickSeal Arterial Closure System 382 racial differences 195 racial issues, access to medical devices 376 radiation exposure, children 212 rapamycin, drug-eluting coronary stents 336–8 recall of device 197–8 authority 17–18 mandatory order 18 records 7 registration 6 registries 52–3, 105–12 breast implants 192 data collection 105 design 105 government-sponsored 108–12 hip implant data 447 manufacturer-sponsored 110–11 national form other countries 111–12 sponsored by professional organizations 106–8 regulation of medical devices change and consumer requests 198–9 consumer concerns 194–5, 197–200 pediatric 212–14 US 5–18, 93 removals 17–18 reporting 64 barriers 47–8, 49 data elements 102 mandatory 102 voluntary 56–7 see also adverse event reporting; adverse medical device events (AMDE), reporting reports 7 annual 46 incompleteness 48, 50 MedSun 71–2, 73 see also adverse event(s), reports respect for persons 129 re-use of devices 29 surveillance studies 48 Rhizopus microsporus 248, 250, 253 risk to health, definition 17 risk:benefit ratio, favorable 139 rods, spinal 209

500

INDEX

Royal College of Physicians of London, ethical code 128 rubber, natural 280 Safe Medical Devices Act (1990) 64, 102, 161 safe use promotion 74–5 safety alert 165 children 211–12 long-term 195–7 surveillance 95 saline-filled breast implants 409, 414 rupture 417 scalp blood gas analysis 470–1 scientific articles, peer review 166 scoliosis congenital 208, 209 repair 209–10 Section 522 studies 164–5 postmarket 154–5 seizures, infants 467 sensors, insulin 231 Serratia marcescens 253 significant risk device study 137–8 silicone extracapsular 417, 420 TMJ implants 192 silicone gel breast implants 191–2, 407–22 adverse event reports 412–14 coding of adverse events 413–14 fibromyalgia 420, 421 infections 414 moratorium 410 rupture 413, 416–19 silent 417 tissue reaction 278 sirolimus adverse events 346 drug-eluting coronary stents 336–8, 343–4, 346 postmarket adverse events 348–9 Society of Thoracic Surgeons (STS) National Adult Cardiac Surgery Registry 106–7 soybean oil, breast implants 408 special controls 7–8 specificity level of devices 88–9 Sphingomonas paucimobilis outbreaks 248, 249

spinal cord stimulator, electromagnetic interference 303, 304, 308, 310, 311, 312–13 spinal fusion 209 spinal growing rods 209 spinal osteoarthritis 452–3 spinal surgery arthrodesis 453 pediatric 208–9 risk factors 452 ST waveform analysis 474 stainless steel 441, 442 STAN fetal heart monitor 474, 475–6, 477 standards development organizations (SDOs) 7 starch peritonitis 251, 252 Starr–Edwards caged-ball valve 367, 368 stent-grafts abdominal aortic aneurysms 356, 357–63 attachment 362 delivery 361 design 361–3 mortality rate 358, 359 patency 363 performance 361–3 postmarket assessment 358–60 structure 362 stents see coronary stents sterilization, surgical instruments 263–5, 266–7, 267 Streptococcus pneumoniae, meningitis association with cochlear implants 397, 399 surgical instruments CJD transmission 260–1, 263, 264, 265, 267 decontamination 263–5, 266–7, 267 peripheral tissues 262 prion disease transmission 260–1, 262 surveillance 1, 2, 3 active 53–4 AMDE 43–58 reports 47–8, 49, 50–2 conditions 54–5 data collection 57 goals 44, 45 rationale 44, 45, 46–7 registries 52–3 statistically-based program 57 systems 64, 65–6, 67 hospitals 175

INDEX technique improvement 165 tools 57 surveys national 115–18 use in epidemiology studies 140–2 suture hemostasis devices 380–1, 388 syringes, disease outbreaks 250 tampons 189–91 technology management 94 Teflon TMJ implants 192 telemetry system, electromagnetic interference 303, 306, 307 temperature-taking devices 205–6 temporomandibular joint (TMJ) implants 192–3 terminologies, current 91–2 thermometers 205–6 thromboembolism, hip implants 447–8 tissue components of medical devices 264–5, 266–7, 267 tissue reactions, materials for devices 276, 277–8 titanium alloys 441, 443 tongue depressors 248, 250, 253 tooth fillings, tissue reaction 278 Total Product Life Cycle (TPLC) program 163–4 toxic shock syndrome 190–1 traceability of medical devices 184 tracking 16–17 transplant tissues, prion disease transmission 261–2 treatment options 24, 25 Trilucent breast implant 408 tympanostome tubes 206–7

501

umbilical cord blood gas analysis 470–1 umbilical hernia 210 underreporting 50–1, 64, 161–2 Unique Medical Device Identification Code 54, 90, 485 Universal Medical Device Nomenclature SystemTM (UMDNSTM) 91–2 universal precautions 280 user error avoidance 146–7 reporting 48 safety of devices 484 user-related error 48, 49 uterine contraction detection 466 valid scientific evidence concept 10 validity of epidemiology studies 139 value of epidemiology studies 139 Vasoseal 387, 390 venous catheters, tissue reactions 277 ventilators 31 disease outbreaks 248–50 electromagnetic interference 305 vigilance system 184–5 webVDMETM 119 World Health Organization (WHO), Good Clinical Practice guidelines 133 World Medical Association, Declaration of Helsinki 128 Zotoralimus 340–1, 343–4