Editorial
A new year in medicine The change of year provides an opportunity for Januslike reflection on the matters that dominated health care in 2010 and challenges ahead for 2011. But how much really changes? To provide perspective, one might consider how 100 years ago, The Lancet, under the editorship of Squire Sprigge, welcomed the new decade in an editorial titled, “The promise of 1911”. The editorial would be familiar in tone and content to Lancet readers today. It cited progress against rabies, diphtheria, and the plague, and praised advances in surgery that would have seemed miraculous to a previous generation. Sprigge anticipated that history would regard the previous decade’s achievements of “wireless telegraphy,…the mechanically-propelled vehicle, and… aviation [as] among the most prominent feats of human ingenuity”. He lamented that advances in medicine moved more slowly, and that the “demon of tuberculosis” had not been exorcised in 1910, although he hoped that better understanding would one day result in mastery of the disease. There had been two general elections in the UK in 1910, so concern was expressed about the effect of political uncertainty on social issues, such as the health of disadvantaged people, the maintenance of charitable hospitals in a depressed economy, occupational health, and workers’ compensation for industrial accidents. In addressing the profession, legislation was urged against the “grasping charlatan and dangerous quack” (echoing a letter about homoeopathy in the correspondence section). He argued that the public would be best protected by better-educated doctors, referring to the issue’s lead Article, which attacked the contemporary curriculum in medical schools and absence of leadership for progress in education. While the eloquent prose and emphasis on syphilis of the Jan 7, 1911 issue seems dated, there is more of relevance to practice in 2011 than one might comfortably admit. Case reports from regional medical associations in the UK would be familiar today, as would reports “from our own correspondent”, which describe cocaine addiction in Montreal, Canada, and identified tuberculosis, measles, diarrhoea, and respiratory infection as leading causes of death in South Africa. The behaviour of expert medical witnesses and the reporting of medicine by the lay press also came under discussion, as did jurisprudence and anaesthesia, and even medical tourism. There was www.thelancet.com Vol 377 January 1, 2011
also complaint about philanthropists whose charity is excessive, poorly coordinated, and indiscriminate. In addition to medical education, which was the focus of The Lancet on Dec 4, 2010, three other topics are particularly timely. A discussion about the origin of cancer is a reminder of how much remains to be understood about this disease. How welcome, therefore, is the study by Peter Rothwell and colleagues in today’s issue of the potential protective benefit of aspirin against some cancers. A review of an 800-page textbook of paediatric surgery suggests a level of competency in 1911 that sits uncomfortably with the recent announcement by the Royal College of Surgeons that half of NHS district hospitals in England lack the facilities or staff to undertake emergency surgery on children. The correspondent from New York reported a crackdown on trade in rotten eggs. Food safety continues to threaten health, yet new food legislation in the USA seems uncertain because of cost. When Thomas Wakley founded The Lancet in 1823, he set out to inform, reform, and entertain. Sprigge’s Lancet was certainly an entertaining read that covered more than health. Writing a book review seems to have been a blood sport in 1911. In addition to medical books, the reviews included a book on Eastern religions and philosophies, diaries, medical journals, and literary and art magazines. Another section was devoted to The Lancet laboratory, in which new products were described. Peripheral articles included archaeology in Egypt, and the science of tea—in which the chemistry of Chateau Lafite is mentioned by way of comparison. History renders some content poignant. The review of A handbook for medical officers in the field foreshadowed the world war that would soon destroy the world that readers knew in 1911. A provincial UK hospital announced plans to acquire an x-ray machine, citing among other reasons that it could be used to treat ringworm; years later those treated would have higher risks of cancer. From Vienna came news about superior health among the city’s 180 000 Jewish people, whom a generation later would face lethal persecution. Between 1911 and 2011 there is much for medicine to be proud of—and also to be humble about. New years bring new promise and new opportunities, but some old demons remain. ■ The Lancet
See Articles page 31
For previous issues of The Lancet see http://www.sciencedirect. com/ For Health professionals for a new century: transforming education to strengthen health systems in an interdependent world see The Lancet Commissions 2010; 376: 1923–58
1
Editorial
Presidential Commission for the Study of Bioethical Issues
On synthetic biology
For The President’s Bioethics Commission see http:// bioethics.gov/ For the Commission’s report on synthetic biology see http:// bioethics.gov/documents/ synthetic-biology/PCSBISynthetic-BiologyReport-12.16.10.pdf For Creation of the first synthetic cell see http://www. sciencemag.org/content/329/ 5987/52.full?sid=db4dfc35f47d-4dcf-b1cb-53b130242cf7
On Dec 16, the 13-member US Presidential Commission for the Study of Bioethical Issues published a 188-page document entitled New Directions: The Ethics of Synthetic Biology and Emerging Technologies. President Barack Obama asked the Commission last May to assess the status of synthetic biology on the same day that the J Craig Venter Institute announced the creation of the first synthetic living cell. The Commission rightly concludes that a new life form was not created by Venter’s team, which inserted a synthetic (man-made) genome of a naturally existing bacterium into a related bacterial cell. The process, the group stated, only represents an alteration of an already existing life form. The report’s 18 recommendations were unanimously endorsed by the Commission. On the one hand, the group encourages the expansion of research within the emerging specialty of synthetic biology. On the other, the group emphasises that the field of synthetic biology is in its infancy and does not yet pose any immediate bio-terror (deliberate) or bio-error (inadvertent) threats in creating new organisms that might warrant formation
of new federal oversight bodies. Instead, crossgovernmental coordination, oversight, and monitoring of the specialty’s scientific progress are recommended. To aid transparency, the report advises that a biology equivalent of the political factcheck.org site is created, as a resource and forum for discussion. Artificially altered organisms tailored to deliver customised drugs or targeted vaccines are already being made. Early in 2012, more efficient full-scale production of the antimalarial drug (artemisinin) from engineered Escherichia coli is to begin. Production of influenza vaccine is another key area of research. The global market for synthetic biology is projected to exceed US$4·5 billion by 2015. Most of the relevant research is currently being pioneered in the USA. The rest of the world should reorganise its existing synthetic biology brain power and technological potential. What is also lacking within the promising emerging field of synthetic biology is international guidance and prioritisation. Synthetic biology must be a global endeavour. ■ The Lancet
World Health Organization
A breath of fresh indoor air
For WHO’s guidelines see http:// www.euro.who.int/__data/ assets/pdf_file/0009/128169/ e94535.pdf
2
Since they were first published in 1987, WHO’s guidelines for air quality have been fundamental for providing information to regulatory authorities in air pollution. On Dec 15, WHO, together with a multidisciplinary panel of experts, released new guidelines for indoorair quality, this time focusing on chemical indoor-air pollutants including carbon monoxide, formaldehyde, and nitrogen dioxide. Poor quality indoor air is a major cause of morbidity and mortality worldwide, accounting for 2·7% of the global disease burden and contributing to about 1·6 million deaths every year (mostly due to acute infections of the lower respiratory tract in children younger than 5 years in low-income countries). This effect on health is substantial and the burden of disease is much greater than that caused by outdoor-air pollutants; however, indoor-air pollution remains lower on public-health agendas than does outdoor-air pollution. The new guidelines emphasise the threat of harmful indoor chemicals and combustion products that are released from solid fuels, for example,
which are still used by more than 3 billion people worldwide for cooking and heating. Because implementation of such measures is beyond the capacity of the individual building user, the guidelines are directed not only at public-health officials, but also at specialists and authorities who have the power to implement the relevant regulatory measures to ensure access to clean indoor air. The guidelines do not instruct on how to take action; rather, they provide scientific bases and uniform recommendations that countries can adopt and develop into legal standards. However, WHO will support its member states in compiling evidence and developing and applying the relevant policies. Whether this assistance will be sufficient is uncertain—WHO should make a call to action to set the ball rolling. The importance of interventions to reduce exposure to indoor-air pollution is reflected in Millennium Development Goals 1, 3, 4, and 7. As 2011 looms, governments worldwide need to ensure that they provide access to clean indoor air for all. ■ The Lancet www.thelancet.com Vol 377 January 1, 2011
Comment
Will an aspirin a day help keep fatal cancer away?
www.thelancet.com Vol 377 January 1, 2011
Which fatal cancers, in addition to colorectal cancer, could aspirin help to prevent? On the basis of results from the current analysis and from previous studies, effects on oesophageal, stomach, and lung cancer mortality seem likely. The reduction in oesophageal and stomach cancer mortality is supported by consistent reductions in observational studies.7,8 Although results from observational studies of lung cancer have been varied,7,9,10 lung cancer mortality was significantly reduced in both Rothwell and colleagues’ analysis and in the Women’s Health Study.6 Results for prostate and pancreatic cancer mortality are suggestive, but should be interpreted more cautiously. The reduction in prostate cancer mortality was not statistically significant and, although slightly lower incidences of prostate cancer have been noted in some observational studies,11 few have examined prostate cancer mortality. Pancreatic cancer mortality was significantly lower (p=0·04) after the first 5 years in the intervention period analysis, but observational studies do not support an effect.12 Further research focusing on longterm daily use is needed to clarify whether aspirin can reduce mortality from prostate and pancreatic cancer. Can we assume that after 5 years on a regimen of daily aspirin, an individual will experience a 34% reduction in risk of fatal cancer, as suggested by the intervention period analysis? Assumptions about the exact magnitude of effects on cancer mortality should be made with caution because the confidence interval indicates the reduction in risk could plausibly be as low
See Articles page 31
Photolibrary
Observational studies and randomised trials indicate that long-term aspirin use can reduce incidence and mortality from colorectal cancer;1,2 however, evidence from randomised trials about other cancers is limited. Peter Rothwell and colleagues, in The Lancet,3 provide important new evidence that long-term daily aspirin lowers mortality from several cancers other than colorectal cancer, and could have a meaningful effect on overall cancer mortality. In a pooled analysis, including the intervention periods of eight randomised trials that lasted up to 9 years, cancer mortality was 21% lower in the aspirin group than in the control group, driven mainly by a 34% reduction in cancer mortality after the first 5 years of follow-up. In a longer-term analysis, including 20 years of follow-up from the intervention and post-intervention periods of three of the eight trials, cancer mortality was 22% lower in participants randomised to receive aspirin for 5–9 years than in those not randomised to aspirin. Rothwell and colleagues’ analyses are informative about the dose and duration of aspirin use that might be necessary to reduce cancer mortality. 75–100 mg per day seems to have been as effective as 300–1200 mg at reducing cancer mortality. However, even low doses of aspirin cannot be used without substantial risk of serious side-effects. Doses of 75–100 mg per day increase the risk of serious gastrointestinal bleeding, possibly as much as do doses of 300–325 mg.4,5 For duration of use, in the long-term analysis no reduction in cancer mortality was noted in participants who were randomised to receive aspirin for less than 5 years, indicating that daily use for at least 5 years will probably be needed to reduce cancer mortality significantly. Results from Rothwell and colleagues’ analysis, which included only trials of daily use, contrast with the null results for overall cancer mortality in the Women’s Health Study,6 a large 10-year randomised trial of 100 mg aspirin taken every other day. These different results suggest that aspirin might need to be used daily to reduce cancer mortality significantly; however, differences in study populations and chance could also have contributed to the contrast in results. Delayed effects of aspirin use could possibly be detected in the future during long-term post-intervention follow-up of the Women’s Health Study.
3
Comment
as 13%, and results for overall cancer mortality might not be completely generalisable to populations in which the proportion of deaths from specific types of cancer is different. In the long-term analysis, which provides the most precise results about specific cancers, about 39% of participants were current smokers and lung cancer was one of the major contributors to the overall reduction in cancer mortality. Most of the additional reduction in overall cancer mortality was due to fewer deaths from colorectal, oesophageal, and prostate cancer. In view of this pattern of results, the generalisability of results for overall cancer mortality to specific groups of patients should be considered. For example, whether a similarsized reduction in cancer mortality could be expected for a woman who has never smoked and has recently had a negative screening colonoscopy is unclear. Clinical guidelines for aspirin use from the US Preventive Services Task Force recommend not using aspirin specifically for colorectal cancer prevention,13 and do not consider cancer when balancing the risk of serious gastrointestinal bleeding against the benefit from prevention of cardiovascular disease.14 Future guideline committees should consider whether effects on cancer mortality might contribute to the overall balance of risks and benefits of daily aspirin use.
I declare that I have no conflicts of interest.
Eric J Jacobs
14
Epidemiology Research Program, American Cancer Society, Atlanta, GA 30303, USA
[email protected]
1
2
3
4 5 6
7 8
9 10
11
12
13
Flossman E, Rothwell PM, for the British Doctors Aspirin Trial and the UK-TIA Aspirin Trial. Effect of aspirin on long-term risk of colorectal cancer: consistent evidence from randomised and observational studies. Lancet 2007; 369: 1603–13. Rothwell PM, Wilson M, Elwin CE, et al. Long-term effect of aspirin on colorectal cancer incidence and mortality: 20-year follow-up of five randomised trials. Lancet 2010; 376: 1741–50. Rothwell PM, Fowkes FGR, Belch JFF, Ogawa H, Warlow CP, Meade TW. Effect of daily aspirin on long-term risk of death due to cancer: analysis of individual patient data from randomised trials. Lancet 2010; published online Dec 7. DOI:10.1016/S0140-6736(10)62110-1. Derry S, Loke YK. Risk of gastrointestinal haemorrhage with long term use of aspirin: meta-analysis. BMJ 2000; 321: 1183–87. Laine L. Review article: gastrointestinal bleeding with low-dose aspirin—what’s the risk? Aliment Pharmacol Ther 2006; 24: 897–908. Cook NR, Lee IM, Gaziano JM, et al. Low-dose aspirin in the primary prevention of cancer: the Women’s Health Study: a randomized controlled trial. JAMA 2005; 294: 47–55. Bosetti C, Gallus S, La Vecchia C. Aspirin and cancer risk: an updated quantitative review to 2005. Cancer Causes Control 2006; 17: 871–88. Abnet CC, Freedman ND, Kamangar F, Leitzmann MF, Hollenbeck AR, Schatzkin A. Non-steroidal anti-inflammatory drugs and risk of gastric and oesophageal adenocarcinomas: results from a cohort study and a meta-analysis. Br J Cancer 2009; 100: 551–57. Hernández-Díaz S, García Rodríguez LA. Nonsteroidal anti-inflammatory drugs and risk of lung cancer. Int J Cancer 2007; 120: 1565–72. Feskanich D, Bain C, Chan AT, Pandeya N, Speizer FE, Colditz GA. Aspirin and lung cancer risk in a cohort study of women: dosage, duration and latency. Br J Cancer 2007; 97: 1295–99. Mahmud SM, Franco EL, Aprikian AG. Use of nonsteroidal anti-inflammatory drugs and prostate cancer risk: a meta-analysis. Int J Cancer 2010; 127: 1680–91. Larsson SC, Giovannucci E, Bergkvist L, Wolk A. Aspirin and nonsteroidal anti-inflammatory drug use and risk of pancreatic cancer: a meta-analysis. Cancer Epidemiol Biomarkers Prev 2006; 15: 2561–64. US Preventive Services Task Force. Routine aspirin or nonsteroidal anti-inflammatory drugs for the primary prevention of colorectal cancer: US Preventive Services Task Force recommendation statement. Ann Intern Med 2007; 146: 361–64. US Preventive Services Task Force. Aspirin for the prevention of cardiovascular disease: US Preventive Services Task Force recommendation statement. Ann Intern Med 2009; 150: 396–404.
Rituximab maintenance in follicular lymphoma: PRIMA Published Online December 21, 2010 DOI:10.1016/S01406736(10)62272-6 See Articles page 42
4
Follicular non-Hodgkin lymphoma is the most common indolent lymphoma in Europe and the USA. This disease was thought to be incurable with standard therapeutic modalities, with major causes of death being disease progression, transformation to more aggressive lymphoma histology, and complications of therapy. Rituximab—an anti-B-cell monoclonal antibody, which targets the CD20 antigen—has had a profound impact on the treatment and outcome of follicular lymphoma in the past decade. Several cohorts of patients from network-based and population-based clinical trials have had improved survival since rituximab use became routine practice.1,2 Indeed, for patients with follicular
lymphoma, prospective trials have shown better benefits in overall survival with rituximab plus chemotherapy than with chemotherapy alone.3,4 The optimum dose and schedule of rituximab is still unknown.5 Extended schedules or maintenance approaches have been assessed in patients with follicular lymphoma after initial treatment with single-agent rituximab and after chemotherapy alone. These studies showed a substantial impact of rituximab on progression-free survival, with suggestions of benefit in overall survival.6,7 Additionally, no major safety concern arose; despite B-cell depletion being prolonged when extended schedules were used, numbers of infections were increased only slightly. www.thelancet.com Vol 377 January 1, 2011
Rituximab plus chemotherapy is now the most commonly used regimen in the initial treatment of follicular lymphoma,8 with median progression-free survival exceeding 3 years. Until now, the benefit of rituximab maintenance after rituximab plus chemotherapy was unknown. In The Lancet, Gilles Salles and colleagues9 report the initial outcome of the PRIMA trial, which enrolled 1217 patients, and randomised (1:1) 1019 of these to 2 years of rituximab maintenance therapy (single dose every 8 weeks) or to observation, both following remission after rituximab plus chemotherapy. This trial is the largest randomised study ever in follicular lymphoma and is a huge accomplishment by the GELA lymphoma group. The patients were a highrisk group, with almost 80% having intermediate-risk or high-risk scores on a validated follicular lymphoma international prognostic index (FLIPI). All had indications for therapy including bulk of disease, symptoms, or laboratory abnormalities. Although three different chemotherapy regimens were used as induction therapy, most patients (75%) were treated with rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP), rendering the results especially relevant to current practice. Of the 1217 patients who received induction therapy, 16% did not proceed to randomisation. Only a third of these patients were lost because of disease progression; most of the remaining patients were withdrawn because of toxic effects during chemotherapy. The characteristics of the randomised group were very similar to the group of patients who were initially enrolled in the study. Patients receiving rituximab maintenance therapy had significantly better rates of 3-year progression-free survival than did those receiving observation (75% vs 58%). The benefit to maintenance was observed in all FLIPI groups. Time to next antilymphoma treatment was also longer in the maintenance group than in the observation group, although the next lymphoma therapy could have been single-agent rituximab in the observation group. Toxic effects were raised in the maintenance group, with most increased effects being infections. The toxic effects were mostly slight and self-limiting. A few patients had immunoglobulin concentrations followed up; there was no significant decrement in concentrations of IgG or IgM in this subset. Other experiences of prolonged maintenance suggest that concentrations of IgM often fall, with a few patients developing IgG deficiencies.6 www.thelancet.com Vol 377 January 1, 2011
Science Photo Library
Comment
Non-Hodgkin Lymphoma (green)
B-cell recovery has not been presented in the PRIMA study; in past studies, at least 6–12 months are needed for full B-cell recovery after rituximab.10 What are the implications of these impressive findings for clinical practice? Should all patients with follicular lymphoma receive rituximab maintenance after induction therapy? The scarcity of a benefit in overall survival after maintenance therapy should be emphasised. Moreover, of considerable interest, two quality-of-life questionnaires were administered, and there were no differences noted between the maintenance group and the observation group. Although Salles and colleagues interpreted this finding as no decrement in quality of life with maintenance therapy, an alternative and equally plausible explanation is that there is no benefit in quality of life to the patient in maintaining remission. The endpoint of progression-free survival in cancer is used as a surrogate of clinical benefit presuming improved quality of life maintaining remission. These findings suggest that this improvement might not be the case in this indolent (and often asymptomatic) disease. Longer follow-up of the PRIMA trial is needed to answer additional important questions. Does the observed safety profile persist in longer follow-up? Does rituximab resistance develop in the maintenance patients therefore precluding future therapies? Is a strategy of observation and rituximab therapy at time of disease progression equivalent or superior to scheduled maintenance in the long term (a key question also being addressed by 5
Comment
the RESORT trial11)? Do the PRIMA findings extend to the large group of follicular lymphoma patients with asymptomatic disease not requiring therapy? Finally, in an era of increased health-care costs, what benefit is necessary to justify the cost of this maintenance strategy, which at my institution would cost Medicare more than US$60 000 per patient? An analysis of cost-effectiveness would be very helpful in addressing this issue. Until these questions are answered, to state that maintenance is needed for all patients with follicular lymphoma who are initially treated with rituximab plus chemotherapy seems premature. However, maintenance is an option and the PRIMA investigators are to be congratulated for this important contribution, and are strongly encouraged to continue follow-up of these patients to answer the questions that remain.
2
3
4
5
6
7
8
9
Jonathan W Friedberg Division of Hematology and Oncology, James P Wilmot Cancer Center, University of Rochester, NY 14642, USA
[email protected] I am a Scholar of Clinical Research of the Leukemia & Lymphoma Society. 1
Fisher RI, LeBlanc M, Press OW, Maloney DG, Unger JM, Miller TP. New treatment options have changed the survival of patients with follicular lymphoma. J Clin Oncol 2005; 23: 8447–52.
10
11
Swenson WT, Wooldridge JE, Lynch CF, Forman-Hoffman VL, Chrischilles E, Link BK. Improved survival of follicular lymphoma patients in the United States. J Clin Oncol 2005; 23: 5019–26. Marcus R, Imrie K, Belch A, et al. CVP chemotherapy plus rituximab compared with CVP as first-line treatment for advanced follicular lymphoma. Blood 2005; 105: 1417–23. Hiddemann W, Kneba M, Dreyling M, et al. Frontline therapy with rituximab added to the combination of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) significantly improves the outcome for patients with advanced-stage follicular lymphoma compared with therapy with CHOP alone: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group. Blood 2005; 106: 3725–32. Friedberg JW. Unique toxicities and resistance mechanisms associated with monoclonal antibody therapy. Hematol Am Soc Hematol Educ Program 2005; 1: 329–34. Martinelli G, Schmitz SF, Utiger U, et al. Long-term follow-up of patients with follicular lymphoma receiving single-agent rituximab at two different schedules in trial SAKK 35/98. J Clin Oncol 2010; 28: 4480–84. Hochster H, Weller E, Gascoyne RD, et al. Maintenance rituximab after cyclophosphamide, vincristine, and prednisone prolongs progression-free survival in advanced indolent lymphoma: results of the randomized phase III ECOG1496 Study. J Clin Oncol 2009; 27: 1607–14. Friedberg JW, Taylor MD, Cerhan JR, et al. Follicular lymphoma in the United States: first report of the national LymphoCare study. J Clin Oncol 2009; 27: 1202–08. Salles G, Seymour JF, Offner F, et al. Rituximab maintenance for 2 years in patients with high tumour burden follicular lymphoma responding to rituximab plus chemotherapy (PRIMA): a phase 3, randomised controlled trial. Lancet 2010; published online Dec 21. DOI:10.1016/S0140-6736(10)62175-7. Anolik JH, Friedberg JW, Zheng B, et al. B cell reconstitution after rituximab treatment of lymphoma recapitulates B cell ontogeny. Clin Immunol 2007; 122: 139–45. Kahl BS. Eastern Cooperative Oncology Group 4402: Rituximab Extended Schedule or Retreatment Trial (RESORT). Clin Lymphoma Myeloma 2006; 6: 423–26.
Wormy mothers, healthy babies: case closed or conundrum? Published Online December 21, 2010 DOI:10.1016/S01406736(10)62271-4 See Articles page 52
6
Vaccines seem to be less effective in poor communities of low-income countries than in the high-income countries where they were developed. Responses to oral vaccines for polio, cholera, or rotaviruses,1 and to parenterally administered vaccines for BCG, measles, or typhoid, are lower in African and Asian populations2 than in wealthy countries. Whether modulation of immune responses by exogenous factors contributes to these outcomes remains unresolved; therefore, parasitic infections in developing countries are especially interesting. Three decades ago, Williams and Greenwood3 found that children infected with malaria parasites respond poorly to meningococcal vaccination, and, in cross-sectional or non-randomised longitudinal studies, parasitic helminths have been shown to be associated with poor vaccine-induced responses.4 Because vaccination is the most cost-effective way of reducing child mortality, any measure that improves vaccine effectiveness in developing countries should be seriously considered.
In The Lancet, Emily Webb and colleagues5 test the hypothesis that helminth infections during pregnancy attenuate responses of infants to vaccination. In more than 2300 mothers and newborn babies, the investigators found that single-dose anthelmintic treatment of Ugandan mothers during pregnancy affected neither the cytokine nor the antibody responses of their infants to vaccine antigens. The similarity in the immunological responses of infants to vaccines, irrespective of maternal treatment, was paralleled by similar birth outcomes and incidence of disease episodes in the first year of life. In a subanalysis of the effect of albendazole on mothers who were hookworm-positive, some effects were seen in responses of interferon-γ to mycobacterial antigens, and in T-helper-2 cytokines in response to tetanus toxoid;5 however, these responses were not of clinical relevance. Thus, this trial resolves speculation about whether helminth infections in the second and third trimester of pregnancy could affect www.thelancet.com Vol 377 January 1, 2011
immune responses of newborn babies to vaccines, and are thereby associated with detrimental health indicators. This finding has serious implications for policy makers who advocate the use of anthelmintic treatment during pregnancy. Webb and colleagues’ trial raises several issues. First, the conclusions can only be applied to maternal hookworm infection, because neither Trichuris trichiura nor filaria are affected by albendazole or praziquantel treatment; ascariasis was rare, and whether the study was well powered enough to accurately ascertain the effects of schistosomiasis is unclear. Importantly, the investigators had previously shown that hookworm infection affected immunological responses of mothers and infants,6 which raised the question of whether those effects were due to the infection itself or to confounders, and justified the reported trial. However, Webb and colleagues do not indicate whether they observed an effect of hookworms on immune responses by comparing the hookworm-infected and uninfected subgroups within the placebo group. Second, whether or not exposure to infection was interrupted is unclear; the study was randomised at the individual level and other members of the pregnant mothers’ household would act as reservoirs of infection. Could household or village randomisation be a more powerful approach? A related issue is the low intensity of helminth infection observed in this adult Ugandan population, which might benefit from the use of new, more sensitive PCR-based methods of detection to more accurately establish the treatment effect.7,8 In Europe, exposure during pregnancy to farming environments that are rich in microorganisms, and possibly helminth antigens, results in a change in immune responsiveness of cord blood and reduced susceptibility to allergic diseases in childhood.9 Regulatory T cells are upregulated in mononuclear cells of cord blood in infants with in-utero exposure to farm environments.10 Two issues arise from these studies: (1) allergic outcomes in infancy—as previously studied in Uganda by the Elliott group11—could be a useful control in the current study by Webb and colleagues, because an effect of maternal anthelmintic treatment in that context would emphasise that certain immune responses can be modified by helminths, but not, importantly, immune responses to early-life vaccines; and (2) exposures at conception (Sangare L, Walson JA, www.thelancet.com Vol 377 January 1, 2011
Abbie Trayler-Smith/Panos
Comment
personal communication) or in the first trimester of pregnancy, might hold the key to profound effects. Finally, the duration of postnatal follow-up might be an issue, because at least two published studies have reported that the effects of in-utero sensitisation due to pregnancy-related infection with Plasmodium falciparum become apparent in only the second year of life.12,13 Helminths are thought to affect immune responses to coinfection (such as P falciparum) and responses to vaccination in childhood; however, Webb and colleagues’ study did not determine the status of helminth infection in the infants aged 12 months. Therefore, the results of placebo-controlled albendazole treatment to determine the effect of intestinal helminths on vaccine-induced responses in childhood are eagerly awaited. The study’s elegant design and its effectiveness in a challenging setting, where frequent data collection must have been daunting, are noteworthy and should ensure that it is a template for future trials that aim to expand our knowledge of the complex, but extremely important, infections that might leave clinically relevant immunological imprints in early life. *Maria Yazdanbakhsh, Adrian J F Luty Department of Parasitology, Leiden University Medical Center, Leiden 2300, Netherlands
[email protected] We declare that we have no conflicts of interest. 1
Serazin AC, Shackelton LA, Wilson C, Bhan MK. Improving the performance of enteric vaccines in the developing world. Nat Immunol 2010; 11: 769–73.
7
Comment
2 3 4
5
6
7
Labeaud AD, Malhotra I, King MJ, King CL, King CH. Do antenatal parasite infections devalue childhood vaccination? PLoS Negl Trop Dis 2009; 3: e442. Williamson WA, Greenwood BM. Impairment of the immune response to vaccination after acute malaria. Lancet 1978; 1: 1328–29. van Riet E, Hartgers FC, Yazdanbakhsh M. Chronic helminth infections induce immunomodulation: consequences and mechanisms. Immunobiology 2007; 212: 475–90. Webb EL, Mawa PA, Ndibazza J, et al. Effect of single-dose anthelmintic treatment during pregnancy on an infant’s response to immunisation and on susceptibility to infectious diseases in infancy: a randomised, double-blind, placebo-controlled trial. Lancet 2010; published online Dec 21. DOI:10.1016/S0140-6736(10)61457-2. Elliott AM, Namujju PB, Mawa PA, et al, for the Mother and Baby study team. A randomised controlled trial of the effects of albendazole in pregnancy on maternal responses to mycobacterial antigens and infant responses to Bacille Calmette-Guérin (BCG) immunisation [ISRCTN32849447]. BMC Infect Dis 2005; 5: 115. Supali T, Verweij JJ, Wiria AE, et al. Polyparasitism and its impact on the immune system. Int J Parasitol 2010; 40: 1171–76.
8
9
10 11
12
13
Verweij JJ, Brienen EA, Ziem J, Yelifari L, Polderman AM, Van Lieshout L. Simultaneous detection and quantification of Ancylostoma duodenale, Necator americanus, and Oesophagostomum bifurcum in fecal samples using multiplex real-time PCR. Am J Trop Med Hyg 2007; 77: 685–90. Schaub B, Liu J, Höppler S, et al. Maternal farm exposure modulates neonatal immune mechanisms through regulatory T cells. J Allergy Clin Immunol 2009; 123: 774–82. von Mutius E, Vercelli D. Farm living: effects on childhood asthma and allergy. Nat Rev Immunol 2010; 10: 861–68. Elliott AM, Mpairwe H, Quigley MA, et al. Helminth infection during pregnancy and development of infantile eczema. JAMA 2005; 294: 2032–34. Schwarz NG, Adegnika AA, Breitling LP, et al. Placental malaria increases malaria risk in the first 30 months of life. Clin Infect Dis 2008; 47: 1017–25. Malhotra I, Dent A, Mungai P, et al. Can prenatal malaria exposure produce an immune tolerant phenotype? A prospective birth cohort study in Kenya. PLoS Med 2009; 6: e1000116.
Pathogenic C difficile is here (and everywhere) to stay Published Online November 16, 2010 DOI:10.1016/S01406736(10)61885-5 See Articles page 63
Kudos to Martijn Bauer and colleagues1 for surveying in The Lancet the Clostridium difficile landscape across Europe. In the USA, Canada, and Europe, the recent epidemic BI/NAP1/027 strain of the toxinotype III infection has placed C difficile infection in the limelight,
Panel: Pathogenetically and epidemiologically important information about C difficile isolates Molecular typing Toxinotyping3 • A+ B+ CDT+ (IIIa-c, IV, V, VI, VII, IX, XIV, XV, XXII, XXIII, XXIV, XXV, XXVIII). • A+ B+ CDT- (0, I, II, XII, XIII, XVIII, XIX, XX, XXI, XXVI, XXVII, XXIX). • B+ A- CDT+ (X, V-like, XVI, XVII, XXX, XXXI) • B+ A- CDT- (VIII). • CDT+ A-B- (XIa, XIb). Restriction endonuclease analysis (eg, BI). PCR ribotyping (eg, 018, 027, 056, 078). Pulse-field gel electrophoresis (eg, NAP1). Multilocus variable-number tandem-repeat analysis or others. Antimicrobial susceptibility pattern Clindamycin, broad spectrum penicillins, cephalosporins, fluoroquinolones, and others (eg, macrolides or carbapenems). Other Geographical location, hospital acquired versus community acquired, outbreak-related or not, severity of disease association, degree of sporulation. A+B+ refers to production of TcdA (toxin A) and TcdB (toxin B). CDT=C difficile binary toxin. CDT+ refers to presence of complete cdt locus (gene encoding for binary toxin). Toxinotype is defined by changes in PaLoc (19·6 kb region of chromosome encoding TcdA and TcdB) and cdt locus.
8
with striking increases in infection rates, in the need for colectomy and intensive care, and in mortality during the past decade.2 Key traits of C difficile isolates are listed in the panel. Increases in rates and severity of disease were attributed to the BI/NAP1/027 type III strain, which has a deletion in tcdC (a negative-regulatory gene) resulting in increased production of toxins A and B, production of binary toxin, fluoroquinolone resistance, and increased numbers of spores.4,5 However, reports have not yet found the 027 strain to be associated with severe disease, especially in non-epidemic settings.6 Just as others have reported, Bauer and colleagues1 have shown a diversity of PCR ribotypes causing C difficile infection, with more non-027 (018 and 056) strains associated with deaths. Some studies have shown no association of C difficile strain type with disease severity.7 In our experience, the “BI-ness” of an isolate is not what determines the degree of intestinal inflammation, but rather, its resistance to fluoroquinolones.8 What then is driving the continued increases in C difficile infections and their severity? C difficile infection has been repeatedly referred to as an emerging infection; however, the disease might have been around for as long as we have been exposed to antibiotics, or even before. Although pseudomembranous colitis after gastrointestinal surgery was described by John Finney in 1893, it was not until the 1950s that pseudomembranous colitis became associated with antibiotic use. Since toxigenic C difficile was recognised as the cause of www.thelancet.com Vol 377 January 1, 2011
Comment
antibiotic-associated colitis and pseudomembranous colitis in the late 1970s,9 many outbreaks and cases have been documented. Antibiotic use has been implicated as a key risk factor for the development of C difficile infection. Although almost all antibiotics have been associated with C difficile infection, the most prominent ones have been clindamycin (1970s), cephalosporins and penicillins (1980s–90s), and fluoroquinolones (2000s). Our data and others’ have linked fluoroquinolone use to the recent increases in C difficile infections.10,11 Selective pressure from antimicrobial agents has long been known to drive the emergence of resistant pathogens, including meticillin-resistant Staphyloccocus aureus, penicillinresistant Streptococcus pneumoniae, extendedspectrum β-lactamase-producing Escherichia coli, and carbapenem-resistant enterobacteriacae. Severe or even fatal disease develops in the often already debilitated hosts. However, none of the leading resistant pathogens that cause severe infections has been as closely associated with the disruption of the host’s normal flora as has C difficile. The rich and diverse intestinal microbiome inhibits C difficile growth.12 This protection is compromised in the presence of broadspectrum antibiotics, especially those to which the pathogen is resistant. With the development of more potent and broad-spectrum antibiotics for other resistant pathogens, we can only expect the problem of C difficile infection to continue. Thus, although it is important to know where C difficile lurk and what types of bacteria these are (whether or not they are directly related to virulence or severity of disease), we also need to know the antibiotic susceptibilities of these isolates. Knowledge of the susceptibilities might not be relevant to treatment of the infection itself, but will probably identify the predisposing and prevailing antibiotic pressure to which we are subjecting our crucial microbiome. Bauer and colleagues’1 data give a snapshot of the relative distribution of the different types of clinically relevant C difficile isolates in Europe. An important follow through would be the genotypic and phenotypic resistance patterns of these isolates. These are the isolates that will probably spill over into the community, the food supply, and animal husbandry— which can also be another source of antibiotic pressure and potential reservoirs for C difficile—and cause www.thelancet.com Vol 377 January 1, 2011
C difficile infection in those who otherwise have no risk factors.13–15 To stay ahead of these costly and deadly outbreaks, we need to know what is out there and to identify and ameliorate what is driving their increasing frequency and severity. One thing is certain: antibiotic-resistant C difficile and other resistant pathogens are here to stay, as long as antibiotics are around. Cirle A Warren, *Richard L Guerrant Center for International Health (CAW) and Center for Global Health (RLG); Division of Infectious Diseases, University of Virginia, Charlottesville, VA 22908, USA
[email protected] We are partly supported by NIH grants (grant number U01 AI075526). RLG is on a scientific advisory board for Probiotics and is a cofounder of AlGlutamine, LLC. CAW declares that she has no conflicts of interest. 1
2
3 4
5 6
7
8
9
10
11
12 13
14
15
Bauer MP, Notermans DW, van Benthem BHB, et al, for the ECDIS Study Group. Clostridium difficile infection in Europe: a hospital-based survey. Lancet 2010; published online Nov 16. DOI:10.1016/S01406736(10)61266-4. Rupnik M, Wilcox MH, Gerding DN. Clostridium difficile infection: new developments in epidemiology and pathogenesis. Nat Rev Microbiol 2009; 7: 526–36. Rupnik M. Clostridium difficile toxinotypes. August, 2010. http://www.mf.uni-mb.si/mikro/tox (accessed Sept 13, 2010). Akerlund T, Persson I, Unemo M, et al. Increased sporulation rate of epidemic Clostridium difficile type 027/NAP1. J Clin Microbiol 2008; 46: 1530–33. McDonald LC, Killgore GE, Thompson A, et al. An epidemic, toxin genevariant strain of Clostridium difficile. N Engl J Med 2005; 353: 2433–41. Cloud J, Noddin L, Pressman A, Hu M, Kelly C. Clostridium difficile strain NAP-1 is not associated with severe disease in a nonepidemic setting. Clin Gastroenterol Hepatol 2009; 7: 868–73. Wilson V, Cheek L, Satta G, et al. Predictors of death after Clostridium difficile infection: a report on 128 strain-typed cases from a teaching hospital in the United Kingdom. Clin Infect Dis 2010; 50: e77–81. Pawlowski SW, Archbald-Pannone L, Carman RJ, et al. Elevated levels of intestinal inflammation in Clostridium difficile infection associated with fluoroquinolone-resistant C difficile. J Hosp Infect 2009; 73: 185–87. Bartlett JG, Chang TW, Gurwith M, Gorbach SL, Onderdonk AB. Antibiotic-associated pseudomembranous colitis due to toxin-producing clostridia. N Engl J Med 1978; 298: 531–34. Archbald-Pannone L, Alcantara-Warren C, Lyman J, Guerrant RL. Fluoroquinolone use and increased incidence of C difficile associated diarrhea. UVA J Med 2009; 6: 45–48. Muto CA, Pokrywka M, Shutt K, et al. A large outbreak of Clostridium difficile-associated disease with an unexpected proportion of deaths and colectomies at a teaching hospital following increased fluoroquinolone use. Infect Control Hosp Epidemiol 2005; 26: 273–80. Borriello SP. The influence of the normal flora on Clostridium difficile colonisation of the gut. Ann Med 1990; 22: 61–67. Centers for Disease Control and Prevention (CDC). Surveillance for community-associated Clostridium difficile—Connecticut, 2006. MMWR Morb Mortal Wkly Rep 2008; 57: 340–43. Bakker D, Corver J, Harmanus C, et al. Relatedness of human and animal Clostridium difficile PCR ribotype 078 isolates based on multilocus variable-number tandem-repeat analysis and tetracycline resistance. J Clin Microbiol 2010; 48: 3744–49. Jhung MA, Thompson AD, Killgore GE, et al. Toxinotype V Clostridium difficile in humans and food animals. Emerg Infect Dis 2008; 14: 1039–45.
9
Comment
The white plague returns to London—with a vengeance In 1660, John Bunyan (1628–88), an English Christian writer and preacher, described tuberculosis as “The Captain among these men of death” when tuberculosis case rates in London had reached a phenomenal 1000 per 100 000 population per year,1 far more than current rates of 340 per 100 000 in sub-Saharan African countries.2 During the 19th century, the white plague, as tuberculosis was named in Victorian Britain (due to the loss of skin colour seen in London tuberculosis patients), continued to ravage Britain, and up to 25% of deaths in Europe were caused by this disease. The death toll from tuberculosis began to fall in London at the start of the 20th century, as living standards (better housing, nutrition, and economic status) improved; subsequent tuberculosis control was achieved by the introduction in the early 1960s of antituberculosis drugs, improved health services, and BCG vaccination. By the early 1980s, tuberculosis was considered to be conquered in the UK and National Health Service (NHS) tuberculosis services were scaled down considerably. Presently, 1·7 million people die of tuberculosis globally each year and the disease is out of control in 22 high-burden countries worldwide.2 Easy travel and migration, and poor socio-economic and living conditions in certain groups, have allowed the disease to re-surface as a public health problem in all European countries. The incidence in the UK has gradually increased over the past 15 years. In 2009, over 9000 cases were reported, a rate
The Bridgeman Art Library
Published Online December 17, 2010 DOI:10.1016/S01406736(10)62176-9
Poor housing conditions in Victorian London
10
of 14·6 per 100 000 population.3,4 This pattern is striking when compared with the general decline in other western European countries; the UK is the only European country where tuberculosis incidence rates continue to rise.3 Tuberculosis has returned to London in force with an increase in the number of cases by nearly 50% since 1999, from 2309 in 1999 to 3450 in 2009,3 accounting for almost 40% of all tuberculosis cases in the UK.3,4 Because the current gold standard for diagnosis (sputum microscopy and culture) only detects up to 70% of active cases, and London general practitioners require education and heightened awareness to improve diagnosis,5 the number of cases reported is underestimated. Drug-resistant tuberculosis is also becoming an important problem in London. During the past 10 years, an ongoing outbreak of isoniazid-resistant tuberculosis in hard-to-reach groups is spreading, with 172 isoniazid-resistant cases reported in London in 2009.6 Ominously, there were a further 58 cases of multidrug-resistant tuberculosis in 2009.7 The increase in the number of tuberculosis cases in the UK has largely been in non-UK born groups; in 2009, these were black African (28%), Indian (27%), and white (10%).3 Interestingly many of these cases were not in new migrants; 85% of individuals born overseas had lived in the UK for 2 or more years, and tuberculosis was common in London boroughs that are relatively deprived.3 Poor housing, inadequate ventilation, and overcrowding—conditions prevalent in Victorian Britain—are causes of the higher tuberculosis incidence rates in certain London boroughs. In all European countries, the disease is mainly concentrated in high-risk groups, such as migrants, refugees, homeless people, drug users, prisoners, and HIV-infected groups.6,8 Prisons provide ideal breeding grounds for tuberculosis and development of drug resistance. The spread of tuberculosis in prisons to prisoners and staff, and resultant spillover into the community when inmates are released or when they need to be transferred for specialist hospital care, are emerging problems in London. A retrospective 4-year (2004–07) study9 of 205 prisoners with newly diagnosed tuberculosis showed that prisoners were more likely to be UK born (47% vs 25%), to be white (33% vs 22%), and to have pulmonary tuberculosis (75% vs 56%) compared www.thelancet.com Vol 377 January 1, 2011
Comment
with all other patients in the UK during that period (29 340 cases aged 16 years or older). Over one-third of culture-confirmed cases (48/139) in prisoners were resistant to isoniazid. Importantly, only 48% of prisoners diagnosed with active disease completed treatment and 20% were lost to follow-up. The ominous tuberculosis situation in London is reminiscent of the unexpected outbreaks of multidrug-resistant tuberculosis in New York and Californian prisons in the early 1990s,10 which arose as a result of complacency of the respective states’ tuberculosis surveillance and control programmes. A large financial investment, with political and legal support, was required to re-establish effective tuberculosis control. An extensive standardised manual of clinical polices and protocols, with a wide outreach service,11 has maintained control to date. Lessons need to be learnt from that situation to reverse the increasing tuberculosis trends in London. A recent London tuberculosis service review assessment has important recommendations,12 which will, if implemented by the UK’s NHS, allow standardisation of tuberculosis clinical policy and practice and improve responsiveness of London’s tuberculosis services’ needs. An immediate and serious long-term political and financial commitment is needed from the UK Government through the NHS if the tide is to be turned against the return of the white plague in London, and if tuberculosis is to be controlled. Such measures will erase London’s reputation as the tuberculosis capital of Europe.
Alimuddin Zumla Department of Infection, University College London Medical School, Windeyer Institute of Medical Sciences, London W1T 4JF, UK
[email protected] I declare that I have no conflicts of interest. 1
2
3
4 5
6
7
8
9
10 11
12
Daniel TM. The history of tuberculosis: past, present and challenges for the future. In: Schaaf S, Zumla A, eds. Tuberculosis–a comprehensive clinical reference London: Saunders Elsevier, 2009: 1–7. WHO. Global tuberculosis control—epidemiology, strategy, financing. http:// www.who.int/tb/publications/global_report/2009/en/index.html (accessed Nov 6, 2010). Health Protection Agency (UK). Tuberculosis in the UK: annual report on tuberculosis surveillance in the UK 2009. http://www.hpa.org.uk/ Publications/InfectiousDiseases/Tuberculosis/0912TuberculosisintheUK/ (accessed Nov 6, 2010). Crofts JP, Gelb D, Andrews N, Delpech V, Watson JM, Abubakar I. Investigating tuberculosis trends in England. Public Health 2008; 122: 1302–10. Griffiths C, Sturdy P, Brewin P, et al. Educational outreach to promote screening for tuberculosis in primary care: a cluster randomised controlled trial. Lancet 2007; 369: 1528–34. Mitchell SL, Seoudi N, Hutchison DC, Drobniewski FA. Multidrug-resistant tuberculosis: resistance rates to first and reserve antituberculosis drugs in the UK in 2008/9 and the role of rapid molecular tests for drug resistance. Thorax 2010; published online Sept 29. DOI:10.1136/thx.2010.148866. Gilbert RL, Antoine D, French CE, Abubakar I, Watson JM, Jones JA. The impact of immigration on tuberculosis rates in the United Kingdom compared with other European countries. Int J Tuberc Lung Dis 2009; 13: 645–51. Hollo V, Amato-Gauci A, Kodmon C, Manissero D. Tuberculosis in the EU and EEA/EFTA countries: what is the latest data telling us? Euro Surveill 2009; 14: Pii=19151. Anderson C, Story A, Brown T, Drobniewsky F, Abubakar I. Tuberculosis in UK prisoners: a challenge for control. J Epidemiol Community Health 2010: 64: 373–76. Drobniewski, F. Tuberculosis in prisons—forgotten plague. Lancet 1995; 346: 948–49. New York City Department of Health and Mental Hygiene, NY Health Department. Clinical policies and protocols, 4th edn. March, 2008. http:// www.nyc.gov/html/doh/downloads/pdf/tb/tb-protocol.pdf (accessed Nov 6, 2010). Hayward JA, Murray D, Iny I, et al. London TB service review and needs assessment. 2010. http://www.brit-thoracic.org.uk/Portals/0/Clinical%20 Information/Tuberculosis/P263%20PHAST%20London%20TB%20 Project%20Final%20report%20I.pdf (accessed Nov 6, 2010).
Health and philanthropy—the tobacco connection On June 14, the world’s two richest men, Mexico’s Carlos Slim Helú and the USA’s Bill Gates, jointly announced that they would each contribute US$50 million to the Latam health project to increase vaccinations and improve child nutrition and natal health in central America.1 Slim already contributes reputedly $2·5 billion annually to his Instituto Carlos Slim de la Salud, which runs a large variety of health programmes in Latin America.2 The latest announcement will naturally attract widespread acclaim as an outstanding example of philanthropy. But it also invites important questions about consistency and competing interests. Any assessment of Slim’s net contribution to public health must balance the impact of his philanthropic www.thelancet.com Vol 377 January 1, 2011
contributions as well as the indirect health consequences that flow from his wealth generation with a less appreciated source of his wealth. Descriptions of Slim’s vast fortune generally concentrate on his telecommunications empire.3 Relatively little is mentioned about his long-standing majority ownership of the Mexican tobacco company Cigatam,3 which has since 2007 been 80% owned by Philip Morris.4 Slim’s website acknowledges that Cigatam “turned out to be the first and most important because of its cash flow, providing the Group with sufficient liquidity to capitalize on available opportunities and thereby increase its acquisitions of big companies”.5 Nor is it
Published Online August 25, 2010 DOI:10.1016/S01406736(10)61036-7
11
Bloomberg via Getty Images
Comment
Carlos Slim and Bill Gates
as widely publicised that he has a continuing role as a non-executive director6 of the world’s largest tobacco company, Philip Morris International (PMI). The company’s shareholders doubtless expect him—like all directors—to make a major contribution to maximise investment returns and the company’s bottom line. PMI’s website6 notes that Slim serves on its finance, product innovation, and regulatory affairs committees. The purposes of these committees include to “monitor the financial condition of the Company, oversee the sources and use of cash flow, capital structure and resulting financial needs”,7 to “monitor and review the development of new product strategies, key legislative, regulatory and public policy issues and trends affecting the Company”, and to “anticipate, respond to, and challenge where appropriate, regulatory and fiscal proposals”.8 This must include responding to international efforts at tobacco control. It is inconceivable that Slim would not have known of the action Philip Morris is currently taking against Uruguay in the International Centre for the Settlement of Industrial Disputes objecting (among other things) to that nation’s new requirement for large graphic healthwarnings on cigarette packs.9 The tobacco industry has long suffered ethical bottom-feeder status with both the public and the corporate world. The Reputation Institute’s 2010 report,10 which involved over 80 000 respondents in 12
32 countries, saw the tobacco industry ranked a distant last of 25 industries on “reputation”. Why? This is an industry whose products kill over 5·5 million people each year, on average 15 years earlier than normal life expectancy.11 It is an industry which has engineered the chemistry and design of its products to, as one infamous 1984 Philip Morris internal memorandum put it, “make it harder for existing smokers to leave the product”.12 It is an industry whose product is responsible for the inexorable rise of lung cancer, the world’s leading cause of cancer death and a disease that was very uncommon before the mass production and marketing of cigarettes.13 Slim’s massive contributions to Latin American health undoubtedly do much good. But the consequences of his continuing history of high-level regional and global involvement in the tobacco industry are hardly trivial in any assessment of his public health footprint. Gates’ philanthropy is unmatched this century. His commitment to reducing some of the world’s worst infectious diseases has poured unprecedented capital into health projects in many of the world’s poorest and unhealthiest nations. His contributions will have already saved uncounted lives. Gates has recently begun to fund tobacco-control projects in low-income and middle-income nations, joining New York’s mayor Michael Bloomberg in injecting an estimated $500 million to try and curb tobacco use in the world’s poorest nations.14 In April this year,15 the Bill & Melinda Gates Foundation withdrew a grant of $5·2 million to Canada’s International Development Research Centre (IDRC), after it emerged that the IDRC’s chair, Barbara McDougall, was a very recent board member of Imperial Tobacco Canada.16 The Gates Foundation statement17 said: “The foundation was recently informed that the chair of the board of our partner, the International Development Research Centre (IDRC), has until recently also been a Director of Imperial Tobacco Canada, Ltd. We are deeply disappointed by this revelation and feel this conflict is unacceptable as we work to support meaningful tobacco control programs in Africa. Therefore, we are terminating our tobacco control grant to IDRC, effective immediately. We remain committed to tobacco control work and look forward to continuing to partner with the anti-tobacco community to reduce tobacco use in Africa.” www.thelancet.com Vol 377 January 1, 2011
Comment
Gates’ decision just 2 months later to partner with Slim is plainly inconsistent. He apparently did not know of McDougall’s appointment when he funded the IDRC. He might well not have known about Slim’s tobacco connections when he joined with him in the Latam project. He must know now. His subsequent actions with IDRC were an outstanding example of principled philanthropy. Let us hope he makes the same call again.
7
Simon Chapman
11
School of Public Health, University of Sydney, Sydney, NSW 2006, Australia
[email protected] I declare that I have no conflicts of interest. 1
2 3 4
5 6
Anon. World’s richest want better health for Latam poors. Reuters June 14, 2010. http://www.reuters.com/article/idUSN1415624220100614 (accessed June 22, 2010). Chapman S. International tobacco control should repudiate Jekyll and Hyde health philanthropy. Tob Control 2008; 17: 1. Anon. Carlos Slim Helú: business activity. http://www.carlosslim.com/act_ empresarial_ing.html (accessed June 22, 2010). Philip Morris International. Philip Morris International announces agreement in principle to acquire additional 30% stake in Mexican tobacco business from Grupo Carso. July 18, 2007. http://www.pmi.com/eng/media_center/press_ releases/Pages/200707180000.aspx (accessed June 22, 2010). Anon. Carlos Slim Helú: biography. http://www.carlosslim.com/biografia_ ing.html (accessed June 22, 2010). Philip Morris International. Board of Directors. http://www.pmi.com/eng/ about_us/corporate_governance/pages/board_of_directors.aspx (accessed June 22, 2010).
www.thelancet.com Vol 377 January 1, 2011
8
9
10
12 13 14
15
16
17
Philip Morris International. Finance Committee Charter http://www.pmi.com/ eng/documents/finance_committee_charter.pdf (accessed June 23, 2010). Philip Morris International. Product Innovation and Regulatory Affairs Committee Charter. http://www.pmi.com/eng/documents/product_ innovation_and_regulatory_affairs_committee_charter.pdf (accessed June 23, 2010). Dias FC. Philip Morris initiates arbitration against Uruguay over new labeling requirements, taxes. May 11, 2010. http://www. investmenttreatynews.org/cms/news/archive/2010/05/11/philip-morrisinitiates-arbitration-against-uruguay-over-new-labeling-requirementstaxes.aspx (accessed June 22, 2010). Reputation Institute. Global reputation PulseTM reports. http://www. reputationinstitute.com/knowledge-center/global-pulse (accessed June 22, 2010). Mathers CD, Loncar D. Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med 2006; 3: e442. Webb WH. Status of Marlboro Development Programme. July 12, 1984. http://legacy.library.ucsf.edu/tid/gmr98e00 (accessed June 22, 2010). Oschner A. Bronchogenic carcinoma: a largely preventable lesion assuming epidemic proportions. Chest 1971; 59: 358–59. Goldman H. Bloomberg, Gates set $500 million anti-tobacco fight (update1) 23 July 2008. http://www.bloomberg.com/apps/news?pid=news archive&sid=atfHVx2DGsqE (accessed June 22, 2010). Anon. Gates Foundation pulls funding from IDRC due to ties to Big Tobacco. April 12, 2010. http://www.marketwire.com/press-release/ Gates-Foundation-pulls-funding-from-IDRC-due-to-ties-to-BigTobacco-1146077.htm (accessed June 22, 2010). Todkill A. Tobacco control and the collateral damage of conflict of interest. Open Med 2010; 4: http://www.openmedicine.ca/article/view/411/325 (accessed June 22, 2010). Bill & Melinda Gates Foundation. Statement regarding IDRC tobacco control grant. April 12, 2010. http://www.gatesfoundation.org/ press-releases/Pages/statement-on-idrc-grant-100412.aspx (accessed June 23, 2010).
13
Comment
Offline: Revising our expectations
Photolibrary
The Bridgeman Art Library
What will 2011 bring us? In the western world, thanks to ruthless political ideology and cruel opportunism, we will see the biggest roll back of the welfare state since its creation. Millions of people will be plunged into poverty. Their health, and the health of their children, will suffer— and not only in the short term, but for a generation to come. It is fashionable today to hate the state. But we forget the contribution the state has made to our overall wellbeing. Material deprivation and mortality are closely linked. The idea that a classe dangereuse threatens our polity—an undeserving poor that needs to be kicked and harassed into work—is a myth that has gained currency in every era, fuelled by fear and mistrust among an insecure middle-class and fanned by governments desperate to maintain their tenuous grip on power. But the state has been critical to human survival. The connection between the notion of human welfare and state responsibility brought peace, competent administration, compulsory schooling, good housing, a public health system, nutrition, hygiene laws, investment in science, reliable health statistics, safe water, sanitation, and a new compassion that fostered solidarity and social protection. State and society are interdependent. Doctors should energetically reject the arguments of those who ignore this history. We must be the vanguard of popular resistance to check this brutal philistinism. *
Photolibrary
Corbis
Not only will the state become quantitatively smaller, it will also change qualitatively. Look at our universities. No longer will universities be places for independent critical inquiry. Most have secured a decent settlement in science from government in exchange for throwing humanities departments to the wolves and for accepting a new role as handmaidens to the economy. In the UK, some of our best scientific leaders have been co-opted into this collusive betrayal. They say: the science community must justify itself to the political class by delivering economic growth. But if we define impact in this one-dimensional way, we will strip any remaining moral purpose from the mission of our universities. Perhaps this is the goal—to weaken, once and for all, one of the few remaining sources of socially progressive analysis and thought. *
14
Global health will undergo a period of reassessment, even retrenchment. During 2010, there was much optimism that enormously successful vertical approaches to disease—the Global Fund and GAVI, for example—would evolve into broader alliances. The Global Fund might expand its mandate to include maternal, newborn, and child health. GAVI might develop a larger healthsystems window to go beyond simply funding vaccine programmes. These hopes, under huge financial pressure, have died in a fire of recrimination. GAVI deposed its leadership, unhappy that the core mission was being tampered with. Interviews for the new Executive Director of GAVI will take place next month. And the Board of the Global Fund last month rejected attempts to make it more than an organisation dedicated to HIV, tuberculosis, and malaria. Silos will be back in fashion for 2011. * Except, perhaps, for one: UNAIDS. Those who have suggested the time is right to close UNAIDS have usually been squashed by the ardent advocacy of the AIDS movement. But there are voices inviting more mature reflection about the future of this UN initiative. When one visits UNAIDS in Geneva, one finds an elegant new building sitting opposite WHO’s ageing headquarters. It is tempting to think that UNAIDS is like WHO—an intergovernmental body, constituted to exist forever. Not so. UNAIDS is a programme—a very successful programme—but not an agency. On June 25–27, 2001, nations came together in a Special Session at the UN General Assembly. There, they made an unprecedented Declaration of Commitment on HIV/AIDS. This coming June will be the tenth anniversary of that commitment. What should UNAIDS do? It should certainly seek to reaffirm that declaration. But it should also invite the wider global community to consider an incremental shutdown of UNAIDS over the next decade, reallocating responsibilities to organisations that may be better equipped to deal with the next phase of the epidemic. The success of UNAIDS—eg, on access to antiretrovirals and human rights—puts it in a strong position to call on its co-sponsors to step up to the future challenges the AIDS epidemic presents. Such would be a fitting sign off. Richard Horton
[email protected]
www.thelancet.com Vol 377 January 1, 2011
World Report
Myths and realities about drug addiction in Mexico As the death toll from Mexico’s war on drugs continues to soar, a spate of attacks on recovering addicts has focused attention on the country’s domestic drug consumption. Mary Cuddehe reports. One Sunday in early December, gunmen burst into two drugrehabilitation centres in Ciudad Juarez, Mexico, killing four patients and wounding five. It was the latest of a series of attacks against drugtreatment facilities in the north of the country that began in the summer of 2008. In October, 13 people were killed in this way in Tijuana. And in a single grim episode in July, 19 were marched out of a clinic, lined up against a wall, and slaughtered. The massacres, cruel and perplexing, have shocked Mexicans, who fail to see the logic of assassinating powerless drug addicts. And while authorities have developed theories about the attacks—they now claim that criminals use the centres as recruiting grounds—the violence has also reminded the nation of the issue of addiction itself. It is an interesting issue. For a while now, Mexico’s explosive addiction rate has been a feature of political speeches and media reports—part of the larger rationale for the government’s battle against drug organisations that has plunged the country into chaos and resulted in 30 000 deaths. The notion links back to the 2008 release of a national household drug-use survey that showed 51% more Mexicans self-reporting as addicts than did in the previous study from 2002. “It is clear to everyone that our nation has stopped being a transit country for drugs going to the United States and become an important market as well”, Eduardo Medina Mora, then the attorney general, noted. “Mexico confronting a drug addiction epidemic”, read a Houston Chronicle headline. “Mexico struggles with soaring drug addiction rate”, read another, in the Toronto Globe and Mail. Even the USA has acknowledged the problem by pledging funding for treatment and www.thelancet.com Vol 377 January 1, 2011
prevention programmes through the Merida Initiative. In fact, the widely reported data that fuelled the panic—460 000 addicts and 4·5 million citizens who have tried illicit drugs at least once—actually represent a trend that is closer to “slow and steady” than “soaring”. In all, the percentage of the population that has experimented
“...Mexico’s explosive addiction rate has been a feature of political speeches and media reports—part of the larger rationale for the government’s battle against drug organisations...” with illegal drugs rose from 4·6 to 5·2. For a country with 110 million people, and arguably more narcotics passing through it at any given time than anywhere in the world, this hardly makes for an epidemic. By contrast, a study assessing WHO data that was published the same year as Mexico’s showed that 42% of Americans had smoked marijuana and 16% had used cocaine. “Mexico doesn’t compare with other countries”, says Luis Solis, director of the National Council Against Addictions (CONADIC), an arm of the Health Ministry. “The problem isn’t as serious as in other countries”, echoes Victor Marquez, the director of treatment at the Youth Integration Center, a non-profit drug-rehabilitation group. No one is discounting the seriousness of addiction as a threat to public health— or that the Mexican rate is indeed on the rise. But viewed in the context of the cartel war that is currently unravelling the social fabric of Mexico, a curious fact is the relatively small effect on domestic consumption. Up until the 1990s, South American cocaine found its way to the USA
primarily via Caribbean maritime paths. When US agencies shut down those routes, they effectively diverted the transitory trade to overland corridors in Mexico. As Medina Mora himself has said: “Drugs are like water. They always find their course.” More than 90% of cocaine that enters the USA now travels through its southern neighbour, according to US State Department estimates. This figure has been reported at somewhere between 300 and 500 tons per year. Additionally, thousands of pounds of marijuana are grown inside the country, largely in the northwestern region known as the Golden Triangle, where the states of Sinaloa, Durango, and Chihuahua, meet and where the poppy that is the base for opium and heroin is also cultivated. Targeting the traffickers who run these operations has done little to change the numbers. On the contrary, according to the UN Office on Drug and Crime’s 2010 World Drug Report, opiate production increased 120% in 2008, ranking Mexico as the world’s third largest producing country (although it is still insignificant measured against Afghanistan, which accounts for 90% of the global supply). In 2009, cannabis cultivation spiked 35% to 12 000 hectares—the highest since 1992. Methamphetamine manufacturing, an important source of income for the cartels, has also increased. In August, 2009, authorities raided a crystal-meth laboratory so enormous and complex it took a reporter for a national newspaper 6 hours to cross. The factory, which the press nicknamed “Crystal City”, was said to turn out about 100 kg of methamphetamine a day. Given the staggering quantities, and the unchecked power that the drug gangs exert in certain regions— 15
AFP/Getty Images
World Report
Mexican soldiers guard 4·5 tons of marijuana seized in November last year
residents of a small town in the border state of Tamaulipas recently abandoned their homes en masse, taking shelter as refugees in a neighbouring city, after criminals took over—it would not be unreasonable to expect the cartels to force their product on the local population. But while reported estimates have put the yearly domestic drug sales figure as high as $5 billion, this is small change compared with the $30 billion market in the USA. There are various theories about why, statistically speaking, Mexicans are justnot that into drugs. One has to do with the government’s substanceabuse prevention programmes. Treatment as a method for drug addiction was written into the 1917 Constitution, but it was not until the 1970s that Mexico’s demandreduction efforts truly began to take shape. In 1969, the governmentfunded non-profit Youth Integration Centers opened. The country’s most established organisation, it operates 110 clinics nationwide and gets about 80% of its financing—about $5 million a year—from the Health Ministry. A few years later, the Health Ministry created its own Center for Drug-Dependence Studies (now the National Institute of Psychiatry). By the 1980s, the private rehabilitation industry was expanding as well. The era, it turns out, was pivotal for drugs in Mexico in more ways than one. 16
Today there are thousands of private and public institutions dealing with all forms of substance abuse, mostly alcohol, according to Solis. Costs and quality range widely. Public outpatient programmes, which have been expanded under President Felipe Calderon, can be as inexpensive as $5 a week—the highest-priced private options cost as much as $10 000 a month. Since Calderon took office in 2006, the new Centers for New Life programme has opened more than 300 treatment centres for at-risk communities. Also, in April, 2009, the law the president had proposed to decriminalise possession of illicit substances was passed. Designed to combat what the government considers to be the rapidly expanding street trade, known as narcomenudeo, the law established personal-use limits. Anyone caught with up to a 0·5 g of cocaine, 50 mg of heroin, or 40 mg of methamphetamine is encouraged to seek treatment, and third-time offenders are required to do so. To many critics, the law was largely pointless because police officers, who are poorly paid and prone to bribes, rarely make small-possession arrests. But it also required the development of a national strategy for prevention and treatment programmes, and the Health Ministry has since taken steps to train and certify drug counsellors. Calderon “has done much
more than previous administrations”, says Marquez. The importance of the Mexican family, the country’s social conservatism, and its deep Catholic roots also help explain the traditionally low incidence of substance abuse. Marcela Lopez, a health researcher with Mexico’s Autonomous University, says “Mexico has more factors of protection than risk”, and this is especially so when compared with the USA. Also important is the fact that fewer Mexicans have the time or financial freedom for the sort of recreational use that fuels much of the drug trade in western Europe and the USA. Roughly half the population lives in poverty. Ultimately, though, it all comes down to the bottom line—the fact that a pound of marijuana can fetch a higher price in San Francisco than in San Luis Potosi. The Mexican attorney general’s office recently reported that $10·9 billion worth of drugs had been confiscated since the beginning of the war in 2006, and government spokespeople on both sides of the border point out how their bi-national interdiction efforts have made smuggling more difficult for the cartels. But the ever-increasing drugproduction efforts and the endless raids on vast, air-conditioned border tunnels, specially designed semisubmersibles, and even shipments of frozen sharks stuffed with cocaine tell a different story. No matter how many seizures are made, the lucrative US market is worth the risk. Mexico’s cartels operate a wildly profitable, tax-free business. But the fact that most of their customers live in another country is some scant consolation for Mexican public health officials. If the domestic market were anywhere near as enticing, they would surely focus on their home turf instead of the USA. In which case, the government really might have to face an epidemic.
Mary Cuddehe www.thelancet.com Vol 377 January 1, 2011
World Report
Pulse oximeters breathe life into surgery in poorer nations Anaesthetists worldwide have joined forces with safer-surgery advocate Atul Gawande to enable low-income nations to buy vital pulse oximeters. Tony Kirby reports. With worldwide demand for surgery reaching an all time high, so attention has focused on outcomes of surgical procedures. Among the many things taken for granted in the operating theatre in high-income countries is the availability of good quality pulse oximeters—devices that monitor the level of oxygenation in a patient’s blood and alert the physician if oxygen concentrations drop below safe levels, allowing rapid intervention. The devices are essential in any setting in which a patient’s blood oxygenation is unstable, including routine operations, emergency and intensive care, and also hospital wards. Unsurprisingly, lowincome and middle-income countries have little or no access to such devices, and poor or non-existent training for the often low-quality devices they do have. Pulse oximeters use a finger probe to measure the oxygen saturation of haemoglobin in the peripheral circulation. The technology in its current form became available in the early 1980s, and by the early 1990s pulse oximeters were adopted as a standard of care in international anaesthesia. Today, 58 countries have established anaesthesia-monitoring standards, and all include pulse oximetry as a minimum requirement. The World Federation of Societies of Anaesthesiologists (WFSA) includes pulse oximeters as a minimum standard of care in all hospitals where surgery is undertaken. Introduction of these devices into developed countries such as the UK reduced death rates by 20 times, from one in 10 000 operations to one in 185 000. In stark contrast, anaesthesia mortality in low-income countries today has been reported to be as high as one in 133. A study published in 2010 by Gawande and colleagues in The Lancet www.thelancet.com Vol 377 January 1, 2011
showed that across sub-Saharan Africa, between 60% and 70% of operating theatres have no pulse oximeter, compared with a global average of 20%. Almost 100% of highincome country operating theatres possess these vital devices. This followed a 2009 study that showed
“Introduction of pulse oximetry is one of the most urgent improvements that could be made to anaesthesia safety.” substantial gaps in pulse oximetry availability and training in pilot sites in Uganda, Vietnam, India, and the Philippines. But the introduction of oximetry led to a sustained change in anaesthesia practice in these settings; with early detection of hypoxia that was managed appropriately, and, vitally, the characteristics of the ideal oximeter for low-resource settings were defined. Stephen Ttendo, President of the Uganda Society of Anaesthetists, says that the reality faced at present by many anaesthetists in resourcepoor operating theatres is that they can only monitor patients by feeling their radial artery or watching for their blood going dark. “Use of pulse oximeters will fundamentally improve the way we look after our patients in developing countries”, he says. “Helping these countries to reach a critical mass of trained anaesthetists is essential. If these numbers are in place, advocacy and sustainability of essential health services will become a reality because the voices of these professionals will be so loud that they cannot be ignored by policy makers.” The Association of Anaesthetists for Great Britain and Ireland (AAGBI), WFSA, Harvard University, and other partners are helping Uganda to
increase its numbers of anaesthetists from the current 20 to 40 by 2013, with hopes of more than 100 by 2020. About 7 million people die or have major complications from surgery worldwide every year. The introduction of WHO’s Safe Surgery Checklist—created by surgeon Atul Gawande (Brigham and Women’s Hospital, Boston, MA, USA)—in 2008 has been changing the landscape of surgery in many settings, including the UK. The checklist enables medical teams to eliminate errors that can lead to increased morbidity and mortality. The strength of the checklist comes through its simplicity and nearuniversal applicability in all countries, irrespective of income. But one item on the checklist prevents its complete rollout in poor countries: the requirement for a pulse oximeter. Anaesthetists, including Ian Wilson, President of AAGBI, raised concerns during the checklist drafting that without the introduction of pulse oximeters, mortality and complications would remain unacceptably high. Wilson and WFSA colleagues then worked with Gawande and partners to develop the Lifebox project—to bring cheap, good quality pulse oximeters to low-income countries. The result is that early this year, the Lifebox oximeter devices (Lifeboxes) will be available for US$250 through the Lifebox website, with vital replacement probes (needed on average every 6 months) available for just $25. In the USA, pulse oximeters can cost $1000. Gawande has seen first-hand the dire need for pulse oximeters during a trip to India. “The man had been seen in the emergency ward for a tubercular empyema and left to wait in the emergency ward more than an hour for an operating room”, he recalls. “Even in the operating room
For more on Lifebox see http://www.lifebox.org
See Articles Lancet 2006; 376: 1055–61
17
Isabeau Walker
World Report
An oximeter in use during anaesthesia at Kayunga Hospital in Uganda
it was not recognised that he needed immediate drainage until he had a cardiac arrest and in the time it took to act he died. If this ward had had pulse oximeters, the doctors would have been able to tell immediately that despite his relatively good appearance compared with other patients, this man was one of the sickest on the ward”, says Gawande. “Introduction of pulse oximetry is one of the most urgent improvements that could be made to anaesthesia safety”, Wilson told The Lancet. “Many of our members have worked in developing countries where there continues to be extremely high anaesthesia related mortality.” As part of the Lifebox initiative, the pulse oximeter design was put out to tender, drawing on the earlier work by Wilson and colleagues. Their message was simple: the oximeter had to meet International Organization for Standardization requirements and be suitable for use in low-resource settings—robust, resistant to damage, with probes that were durable and reusable. Furthermore, the oximeter had to be suitable for all ages. And of course, most importantly, be cheap. WHO and WFSA endorsed these specifications. From eight competing manufacturers, Acare Technology, Taiwan, won the tender. Their oximeter costs $199 ($250 including delivery), with replacement probes available for $25. 18
Concerns have been raised that these low-cost devices could be bought or sold on to more wealthy nations. “The contract between the project and the manufacturer stipulates that the Lifebox oximeters will only be available to not-forprofit hospitals in low-income, or lower middle-income countries”, says Wilson. “The project will monitor purchasers who have to agree not to re-sell the oximeters. Clearly there is some risk that oximeters will end up in the wrong places, but we hope strict monitoring will prevent this.” A key part of the project is that each oximeter is supplied with educational materials including videos, tutorials, and a manual—all available free on a CD. Part of the project is to run training workshops, so that the technology is accompanied by the knowledge of how to use and look after the oximeters to make them last longer and sustain improved standards of operative care. Gawande and Wilson hope that donors will step in to bulk-buy the oximeters for distribution to those countries most in need. Such efforts will, says Gawande, save vital time in getting the devices rolled out. “A donor taking on the purchase and distributing free devices to countries could happen much more quickly than lengthy negotiations with health ministries of low-income countries such as many found in Africa. Such negotiations could take a year or more”, he says. Ongoing WHO pilot projects in Moldova and Zambia are testing the effects of pulse oximeters on outcomes of surgery in low-income nations, as well as the other aspects of the Safe Surgery Checklist. Developing effective procurement systems for hospitals in low-income and middle-income countries will make a major contribution to the quality and safety of anaesthesia and surgical care. Wilson and Gawande agree that there is potential to increase access to devices such as
combined anaesthesia monitors (non-invasive blood pressure, capnography, and electrocardiogram) and essential items of disposable equipment such as spinal needles, sutures, and sterile gloves. Currently, items as basic as alcohol gel rub for hand-washing can cost nine times as much in Africa as in developed countries. The Lifebox project has now, with the help of the global children’s charity Smile Train, ordered 2000 Lifeboxes for delivery in spring 2011 to various countries. But the Lifebox team are now appealing for extra help to get past the finishing line for this initial phase. “We are seeking other partners and funds, and those will be crucial to our success”, says Gawande. “This is a gamble and an experiment in providing market power for poor hospitals and using it to drive the checklist into place as well. The two go hand in hand. The oximeters are an essential component of dramatically improving the safety of surgical care. The surgery checklist is another. And local training and monitoring to drive a culture of increasing safety is important as well. This together is the Lifebox mission.” Before this large-scale roll out, 200 oximeters are being shipped out to test and refine the best strategies for ensuring reliable delivery, learning, and governance. The ultimate aim is to update the more than 70 000 operating theatres worldwide without the monitors, with an expected order of more than 10 000 monitors in the first 2 years. Like Wilson, Gawande is excited about the precedent that Lifebox could be setting: “If this works, imagine being able to further build the market power of low-income hospitals to lower the price barrier for all kinds of essential technologies for safety and basic health-care services, while also helping us bring quality and safety practices.”
Tony Kirby www.thelancet.com Vol 377 January 1, 2011
Perspectives
Book Eyewitness accounts from surgeons in Gaza Was it a necessary invasion undertaken as humanely as possible to eradicate a potent source of terrorism? Or a brutal attack on a civilian population designed to create fear? These two narratives are replayed in relation to the Israel–Gaza conflict of 2008–09. Israeli doctors remind me about the reality of the attacks by rockets on Israeli cities, the fears induced by suicide bombings, and the threats of outside invasion— together with the absolute necessity of a national home for the Jewish people who are still under threat of harassment in many countries. At the same time, Palestinians speak to me of their history of dispossession and discrimination, continuing violence against property and land, and their right to resist occupation and oppression. Both these perspectives must be understood, and acts of violence condemned wherever they originate. And yet is there not an issue of justice at stake? These matters are explored in Eyes in Gaza, a disturbing description of the conflict in Gaza by two experienced Norwegian surgeons, Mads Gilbert and Erik Fosse, who were among the very few westerners allowed into Gaza after the start of the war. The book is a day by day account of the incredible work performed by these two dedicated surgeons, side by side with their Palestinian colleagues, during 16 days at the height of the conflict. It is a hugely moving work. Much of it is about children—according to Gilbert and Fosse “The attack hit the children of Gaza particularly hard: more than 300 children killed and over 1600 injured.“ The two surgeons write alternate chapters, describe their cases, and evoke the atmosphere in the hospital as child after child was admitted with life-threatening injuries. They wonder at the superb emergency www.thelancet.com Vol 377 January 1, 2011
action taken by their Palestinian colleagues in the challenging working conditions of a packed intensive care unit with erratic electricity supplies and the continual stress of possible injury to their own families. Reading the book, one gets to know these fine colleagues, and their patients.
“It is now accepted that doctors who do not speak out against torture are complicit. Should we not apply the same argument in respect of any war that devastates civilians?“ 12-year-old Ahmed, the nephew of a Palestinian surgeon, who died after both legs and much of his genitals were blown off. Mahmoud, aged 11 years, who was sent by his parents to play on the roof since they thought it would be safer than playing on the street, and returned with 20–30 small dotted holes in his chest wall, and died shortly after, hit by an accurately targeted drone rocket. 9-month-old Jumana, whose hand was amputated after the house in which her family were sheltering was bombed, killing 11 family members. In total during 2 weeks Gilbert and Fosse performed 270 operations, which included 53 amputations, 36 brain operations, and 43 fracture treatments. When the time came to leave, the horrors did not end. As part of a convoy of 16 ambulances taking seriously ill and ventilated patients for transfer to Egypt, Gilbert and Fosse describe how they were stopped before reaching the border and subjected to warning tank fire, forcing their return to Al-Shifa hospital. It was another 3 days before they could finally leave, again subject to bombings as they passed into Egypt. Eyes in Gaza is also a political book that made me reflect on the wider context of these events. As
the conflict progressed, there was increasing unease in certain segments of the Israeli population, particularly among medical professionals active in Physicians for Human Rights, and many assisted with relief efforts. However, this did not turn into mass political opposition, and indeed I received a letter from the President of the Israeli Medical Association as follows: “Regarding your request that we call on the government to stop further bombardment, I am sure you are aware that we are not a political organization.” So, it becomes political to call for a halt to a humanitarian disaster. The same motivation was perhaps behind what I regard as the failure of UK medical royal colleges and the British Medical Association to make statements beyond bland words about this humanitarian disaster. For me, this is a sad reflection on a failure to understand the social determinants of health. The two Norwegian surgeons were, therefore, vitally important as reporters when no western or Israeli reporters were allowed into Gaza. Their message did get out. In a special edition of The Lancet in January, 2009, Gilbert and Fosse gave a graphic account of the hospital conditions that could leave no one unaware of the tragedy unfolding in Gaza. An accompanying Editorial stated “We are disappointed by the silence of national medical associations and professional bodies worldwide. Their leaders, through their inaction, are complicit in a preventable tragedy…” It is now widely accepted that doctors who do not speak out against torture are complicit. Should we not apply the same argument in respect of any war that devastates civilians?
Eyes in Gaza Mads Gilbert, Erik Fosse. Quartet Books Ltd, 2010. Pp 332. £20·00. ISBN 070437191X.
Tony Waterston
[email protected]
19
Perspectives
Exhibition Journey through the afterlife
The Trustees of the British Museum
A dead man’s heart is weighed in a balance against a feather that represents truth, while he affirms that, in life, he had not committed any one of 42 sins, which include murder, theft, blasphemy, or stealing the gods’ food from the temples. His innocence, as judged by the gods, will mean that the deceased either attains eternal life or his body will be eaten by the Devourer. This judgment was one of the important stages in the ancient Egyptians’ journey from death to eternal life and is one of the scenes depicted in the British Museum’s major exhibition Ancient Egyptian Book of the Dead. The visitor is taken on a unique journey, through a series of thematic sections that encompass the Egyptians’ preparation and burial of the mummy, through dangers encountered in passing from this world, to the ultimate joys experienced in the land of eternity. An essential possession for this journey was the Book of the Dead—a set of spells, often accompanied by exquisite illustrations, and customarily inscribed on papyrus or coffins and linen wrappings.
Ancient Egyptian Book of the Dead The British Museum, London, UK, until March 6, 2011. http://www. britishmuseum.org/whats_on/ all_current_exhibitions/book_of_ the_dead.aspx
The Trustees of the British Museum
See Online for webvideo
The Book of the Dead was essentially a powerful tool of Egyptian magic, and the specific positioning of spells on one coffin on display was clearly intended to provide magical protection for the relevant body parts of the mummy inside. A mummy shroud provides an intriguing glimpse into practices in an embalmer’s workshop: clearly an item of “undertaker’s stock”, it has been decorated with the relevant magical spells, but the text contains spaces for the names of a future purchaser and his wife to be inserted later. Various physical and magical procedures were deemed necessary for a safe passage to the next world. The “Day of Burial” part of this exhibition looks at mummification and “Opening of the Mouth”, when the priest used magical tools to touch the mummy, coffin, and other tomb goods to enable them to “function” for their owner in the next world. Ancestral statues and a letter to a deceased relative asking her to take action to cure the writer’s illness, illustrate poignant attempts to retain a link between the dead and the living.
Rosalie David Weighing of the heart, detail from the Book of the Dead of Ani
20
Among the many spells in the Book of the Dead are those to activate magical power in protective jewellery (amulets) placed on the mummy, while others provide a kind of “travel guide”, detailing the various features that the deceased would encounter on the journey to the afterlife. Other spells empower the dead to repel or avoid threats posed by snakes, crocodiles, insects, slaughter, and a diet of excrement and urine encountered in the afterlife. I found the exhibits relating to the role of scribes, their equipment, and the production methods and evolution of the Book of the Dead especially interesting. It also demonstrates how ongoing research by Egyptologists and conservators can shed light on possible regional traditions, previously unsuspected links between fragments held in different museums, and how infrared techniques have been used to identify the hitherto unknown name of one papyrus owner. The final display contains the unravelled Greenfield Papyrus, the greatest in length of any known copy of the Book of the Dead, which belonged to Nesitanebisheru, daughter of the High Priest Pinudjem II, who virtually usurped the king’s role throughout southern Egypt in about 930 BC. This papyrus provides a dramatic and fitting conclusion to an exhibition that effectively explores Egyptian mythical and spiritual ideas of life and death. The wide range of objects on display, several short films, and an excellent multimedia guide featuring curators’ and conservators’ comments about the exhibits and their personal involvement in the project all help to reveal something about the aspirations, fears, and moral dilemmas present in one of the world’s oldest civilisations.
[email protected]
www.thelancet.com Vol 377 January 1, 2011
Perspectives
Clare Gerada is the first female chair of the UK’s Royal College of General Practitioners (RCGP) for more than half a century, and enters the role at one of the most challenging times in the history of the profession. The UK Government’s plans to hand over commissioning of services and accompanying large budgets to GPs have been denounced by some, and cautiously welcomed by others. As RCGP chair, Gerada will be at the forefront of the profession’s responses to these major challenges. Gerada was absorbed into the profession from her earliest years, thanks to her father’s role as a family doctor for a large Italian community in the UK town of Peterborough. The young Gerada would peer over the balcony of her home—that doubled as the surgery—and also accompany her dad on his home visits. She completed her medical degree at University College Hospital in London. Among the areas that fascinated her most during her training were emergency medicine and mental health, and she completed a rotation at the Maudsley Hospital in London to hone her psychiatric interests. In the late 1980s, Gerada made some pioneering steps in the field of drug addiction and helped set up a “barefoot” style clinic for drug addicts in London who found themselves isolated from the conventional general practice model of care. “It was a big relief for these patients to talk to a doctor about regular health matters separate to their addiction problems. Such basic things as diet and blood pressure advice. I even detected several cases of diabetes”, Gerada told The Lancet. Part of the clinic’s remit was to improve access of drug users to primary care. Gerada was able to maintain the clinic through her GP training and her first years in practice. She then had to manage an addiction of her own—the desire to combat health inequalities at the coal face. Gerada decided to combine the best of both worlds, by becoming a GP at a multidisciplinary practice at the Hurley Clinic in the Lambeth Borough of London, where she remains a partner today. In this practice, care of families, ministers, and civil servants takes place alongside that of drug addicts and people with severe mental health disorders. Gerada’s first night on call saw her visit a Member of Parliament and a patient living in a squat in successive home visits. Those early years as a GP also saw her set up a specialist service for pregnant drug users and an inter-practice liaison service for mental health issues. “I think we changed the way substance misuse was managed in this country”, says Gerada. Mark Ashworth, a partner at the Hurley Clinic, describes Gerada as an innovator who is now a household name in the world of primary care substance abuse. As an example of one of Gerada’s many innovative projects for the practice, Ashworth explains that www.thelancet.com Vol 377 January 1, 2011
“We now employ six teachers in a project called the Health Education Links Team. This team goes into local schools with a health education message, a project that started from Clare’s observation of how little professional exchange there was between local teachers and GPs.” Since becoming a GP in 1990, Gerada has hailed the job as “everything I had ever wanted it to be and much more”. She started just after a controversial new contract for GPs in the UK. Out-of-hours care was then the responsibility of GPs, but this meant that many doctors would need to work through the night and into the next day, leaving some of them exhausted. The new contract in 2004 saw local primary care trusts take over responsibility for this service using co-operative models with certain practices on call, but although some areas had enough doctors many others were underserved. Gerada hopes that new health reforms proposed by UK Health Secretary Andrew Lansley “will return responsibility for commissioning out-of-hours care back to GPs but in a sustainable way”. As for the other reforms, Gerada is positive about GPs commissioning local services and increased involvement of patients in decision making. She also favours reduced bureaucracy, and hails the advent of the paperless GP surgery earlier this century as a real watershed moment in the profession. But introducing a direct performance-related incentive—for example, to keep patients out of hospital— would immediately, Gerada believes, introduce a conflict of interest into patient consultations. She also thinks that “too much will be changing too fast, whatever the merits of the individual changes”, and that GP training should be extended from 3 years to 5 years to help cover all the skills required. “If we really want to make a difference to the communities we serve, we need to build in time to go out into the community and find the problems before they find us.” She gives the examples of targeting high hypertension rates among Afro-Caribbean men and high diabetes rates in Asian populations, as well as breakfast clubs, parenting classes, and school clinics to keep people healthy, all this as the demands of an ageing population also mount on general practice. “The future cannot just be about quicker access to tests”, says Gerada. “A mother being given parenting and cookery classes to prevent obesity in her family could be just as important.” She hopes the forthcoming Health Bill from the UK Government will allow the seeds for these issues to be sown. “Ultimately, these new GP commissioners, when appointed, will need to be brave and look beyond the surgery for the future of general practice”, says Gerada.
RCGP
Profile Clare Gerada: Chair of UK’s Royal College of General Practitioners
See Online for webvideo
Tony Kirby
[email protected]
21
Perspectives
The art of medicine Of wandering doctors, cities, and humane hospitals
Marek H Dominiczak
Legend has it that on the island of Kos in Greece, Hippocrates taught his pupils under a plane tree. There is something symbolic in the simplicity of such a depiction. It reminds one that doctors have often been wanderers, not bound to particular locations or buildings. They went out to see the sick in their own surroundings. Of course, the sick have always needed shelter and respite space. However, as more complex medical procedures developed, specific space was also required to perform them. Thus the respite and treatment strands converged on an architectural platform and, with time, led to the development of modern hospitals. The contemporary debate about what constitutes an optimum healing space to a large extent relates to the balance between the shelter-respite and the active treatment-related functions. This essay attempts to trace key historical developments of healing spaces. While early doctors were wanderers, spiritual healers often were not. A temple had been the dominant architectural form in human communities long before Hippocratic medicine, and indeed before the rise of European culture. In ancient Greece, healing was linked to worship: the centres of healing were the temples of the god of health Asklepios, known as Asklepeions. The largest of these were in Epidauros on the Greek mainland, and on the Hippocratic island of Kos. The monumental three-level temple on Kos stood on a hill facing the distant sea. Visitors were admitted at the lower level. They were subsequently taken to the higher terrace with open views, where they were to spend the night and be visited by Asklepios. The Kos Asklepeion very well illustrates the role
Paimio Hospital, Finland, where Alvar Aalto’s concept of total design included interiors, furniture, and utensils
22
of architecture as a framework for a healing ritual, and shows how spectacular blending with surrounding nature enhanced its impact on the “treated”. In the Middle Ages, Christianity, very much in the GraecoRoman tradition, associated worship with monumental architecture. Caring for the sick and the destitute was part of the Christian message of charity. The early “hospitals” sheltered the poor, the pilgrims, and the abandoned children. For the sick, they were places of terminal care rather than of active treatment. They were appended to churches, and particularly to monasteries. The St John Hospital in Bruges founded in 1188 (probably by the city rather than the Church) was such a church-like building: by the 15th century it had three halls containing “patient” cubicles and the church space, which was an integral part of the hospital. The main altar painted by Hans Memling was visible from the wards. Prayer was a dominant part of the healing process. Subsequently, the ownership of more institutions became secular. In the Catholic city states, a city often owned hospitals jointly with the Church. In England, dissolution of the monasteries during the Reformation of 1536–41 forced a change of ownership. In London, St Bartholomew’s Hospital (founded in 1123) and three others were granted by King Henry VIII to the City of London in 1546. City hospitals were often prominent public buildings. In Siena, the Santa Maria della Scala Hospital faces the cathedral across the main square. In Florence, the Foundling Hospital designed by Filippo Brunelleschi in 1419 is an outstanding example of renaissance architecture. The next major change in thinking about hospitals took place in the 18th century. It followed the Enlightenment emphasis on rationality, science, and nature. Later, during a reform of hospitals in France, their improvement amounted largely to the architectural redesign, particularly the introduction of pavilion structure, first in the Hôpital Lariboisiere, opened in Paris in 1854. This created a more open space that improved ventilation and access. Interestingly, the pavilion design later provided a framework for separate housing of emerging medical specialties. A true revolution in the concept of the hospital was caused by the accelerating progress of medical science in the late 18th and the 19th century. Hospital-based care maximised the medical experience by bringing together patients with similar conditions, facilitating statistical assessment, and—importantly—making clinical teaching possible. Thus, scholars entered hospitals, which became major sites of medical research on a trajectory rising until the late 20th century. A parallel, and complementary contribution to hospital design in the 19th century was the work of Florence Nightingale (1820–1910) and her nursing colleagues who www.thelancet.com Vol 377 January 1, 2011
Perspectives
reorganised hospital wards by applying the concepts of cleanliness and antisepsis, after spectacular results achieved at a field hospital in Scutari in 1854, during the Crimean War. The new science needed specialised spaces: laboratories gradually became part of hospitals, and the main dwelling places of scientists. This had a major influence on the architectural shape of hospitals—they became buildings with wards at the core, surrounded by increasingly complex support structures. In the early modernist period architecture gained increasing importance as a tool for social change. Architects such as Charles Edouard Jeanneret (1887–1965), better known as Le Corbusier, developed comprehensive plans for cities with new “division-of-function” layouts. Le Corbusier was fascinated by industrial buildings, processes, and mass production. He famously named his Unité d’Habitation (housing unit) in Marseille built in 1947–52 a “machine for living”. In addition, the “organic” architecture of Charles Rennie Mackintosh (1868–1928), Frank Lloyd Wright (1867-1959), and Antoni Gaudi (1852–1926) achieved new sophistication in manipulating textures and materials, and fusion of architecture with the natural environment. The emphasis on access to natural light and to surrounding nature became the hallmark of modernism— neutralising to an extent the effects of stern functionality. Within health care, this was particularly applicable to tuberculosis sanatoria built in the early 20th century, to allow patients ample access to fresh air, which was at that time regarded as a therapeutic measure. An extraordinary healing space implementing these principles was designed by a Finnish architect Alvar Aalto (1898–1976), in Paimio near Turku. His building was integrated with the surrounding forest but in addition Aalto applied the concept of total design that included the building, the interiors, and even small utensils. The hospital still functions as a health-care unit; it recently became a UNESCO World Heritage Site. In contrast to the sanatoria, the architecture of urban hospitals became increasingly determined by limited space and rising maintenance costs. The minimalist skyscraper, which turned out to be the most economical form of a building, was eventually widely used in hospital architecture. A “tower and podium” hospital evolved, with a ward tower and the podium containing support services. In these technically complex structures, the therapeutic functions, in a continuation of modernist zeal, had had unqualified priority, while the respite function withered. Architecturally, the undiluted functionality often created rather sinister structures, which the architect Markus Schaefer has said “neglected basic human needs”. Schaefer has suggested that “the hospital building stripped the patient of her privacy and individuality, the healing machine of her body”. Hospital space became increasingly perceived by patients and visitors as overwhelming, incomprehensible, and even threatening. Retrospectively, it was an astonishing lack of cultural balance. One could argue that the neglect of humane aspects www.thelancet.com Vol 377 January 1, 2011
of hospitals had roots in the rather insensitive nature of modernist science. One also wonders whether the postFlexnerian, strongly focused view of medical science that became a mantra of medical education in the 20th century had an effect here. Or was it the disengagement of the arts and humanities from the medical landscape? The end of the 20th century saw some change, with increasing emphasis on individual rights and the contextualisation of medicine. These ideas influenced the design of a new generation of hospital spaces—all in the context of consumer culture and aesthetics. Also at that time, an application of scientific method to hospital architecture emerged in the form of evidence-based design. A seminal paper by R S Ulrich, published in Science in 1984, showed that access to natural light improves patients’ recovery after surgery. There is now a substantial body of evidence that relates aspects of design to outcomes such as patient safety, infections, medical errors, falls, pain, sleep, depression, and length of hospital stay. Thus, in the context of individual rights, privacy, openness, and also aesthetics sparked by consumerism, architecture returns again to a prominent role: it becomes an advocate of cultural trends, which require an adjustment of architectural spaces beyond their functionality. Openness and clarity of structure become paramount for a hospital, not only from a functional but also from a cultural point of view. The challenge is to achieve this in a sustainable way while also allowing for future morphing of hospital spaces in response to medical science-driven requirements. There is a serious cultural role for a health-care facility in the contemporary world. A modern hospital can potentially enhance inclusiveness and social cohesion. The Philips Index: America’s Health and Well-Being Report 2010 has shown that inhabitants rank the access to local hospitals as the most important factor related to wellbeing after safety and crime rate. The quality of health care was also a factor in the recent international Livable Cities survey—therefore it has an impact on the global “status” of cities, their attractiveness, and consequent investment prospects. What about the medical world? Since hospital spaces have both a functional and cultural meaning, they deserve more mention in medical education. At the University of Glasgow Medical School we have recently introduced an interdisciplinary teaching module on hospital spaces as part of the medical humanities teaching programme. Doctors started as wanderers—however, today, many wander no more. And they need to be aware of their dwellings. This essay is based on the Presidential Address I gave to the Royal Medico-Chirurgical Society of Glasgow in October, 2010.
Marek H Dominiczak College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow G12 8QQ, UK
[email protected]
Further reading Cox M, Irby DM. American Medical Education 100 years after the Flexner Report. N Engl J Med 2006; 355: 1339–44. Dominiczak MH. Medicine, architecture and the arts. Opportunities for dialogue. In: Law A, ed. Space to heal. Edinburgh: Sleeper Publications, 2009, 5–21. Schaefer M. Building hospitals— hospital buildings. Rotterdam: The Berlage Institute, 2005. Ulrich RS. View through a window may influence recovery from surgery. Science 1984; 224: 420–21. Ulrich RS, Zimring VC, Zhu X. A review of the research literature on evidence-based healthcare design. Healthcare Leadership White Paper Series. Atlanta GA: Georgia Institute of Technology, 2008. Philips Index: America’s Health and Well-being Report 2010. http://www.newscenter.philips. com/pwc_nc/main/standard/ resources/corporate/press/2010/ Philips_index/The_Philips_ Index_Americas_Health_and_ Well-being_report.pdf (accessed Dec 14, 2010).
23
Corbis
Obituary
Frank John Fenner Australian virologist who had a leading role in the global eradication of smallpox and the use of myxomatosis to control rabbits in Australia. Born on Dec 21, 1914, in Ballarat, Australia, he died on Nov 22, 2010, aged 95 years. It was the summer of 1951, and the Australian media were in an uproar. Myxoma virus, which scientists hoped would control the country’s rabbit plague, had escaped into the wild. Within weeks, millions of rabbits had died across the Murray-Darling Basin. This was good news. The trouble was that an outbreak of human encephalitis had sprung up in the same region at the same time. Was the same virus infecting human beings and rabbits? Scientists working under Sir Frank Macfarlane Burnet, the head of Melbourne’s Walter and Eliza Hall Institute, quickly established that the two diseases weren’t connected, but the message didn’t seem to be getting through to the public. The chairman of a local hospital challenged Burnet to test the harmlessness of the virus on himself, so he turned to an expert in poxviruses who had previously worked at the Hall Institute, Frank Fenner. “Burnet consulted me and we decided that it was absolutely safe”, Fenner wrote in an autobiography. “I prepared a suspension and injected Burnet subcutaneously with one, 10 and 100 rabbit-infectious doses; Burnet inoculated me…none of us showed lesions or an antibody response…and the public was reassured.” This straightforward retelling of a dramatic national moment is classic Fenner. As a major figure in 20th-century Australian science and a prolific writer, he was a man whose style focused more on facts than personal drama. “Frank was a 24
fascinating character”, says Ian Gust, a medical virologist who was the research and development director for the medicinal products manufacturer, CSL Limited. “He was not only a great historian of virology, but he was also the history.” After time as a medical resident at Adelaide Hospital and completing a diploma of tropical medicine, Fenner enlisted in the Royal Australian Army Medical Corps in 1940. During the war, he served in New Guinea as a malariologist and was made a Member of the Order of the British Empire for his important role in controlling the disease among troops. Immediately after the war, he accepted a position at the Walter and Eliza Hall Institute, where he took up Burnet’s suggestion and began studying mousepox. “Frank was a great basic scientist in the field of virology”, says Sir Gustav Nossal, who became the head of the Hall Institute in 1965 when Burnet retired. “His landmark studies were on mousepox. He investigated everything to do with how the mousepox virus behaves in the body of the mouse, the totality of the epidemiology, clinical features, the spread, the way it multiplies. Those were really landmark studies.” After a fellowship at the Rockefeller Institute in New York, Fenner was invited by Nobel laureate Howard Florey to head the department of microbiology at the Australian National University’s newly established John Curtin School of Medical Research. He would work there until 1967, during which time myxomatosis was his main research subject. “What Frank did was to study in meticulous detail the myxomatosis epidemic that ran through Australia and in particular chart the changes in both virus and host—the virus mutating and the rabbits being so decimated by the disease”, said Nossal. In 1967, he was appointed director of the John Curtin School, a post he held until 1973 when he put medical research aside to set up a new Centre for Resource and Environmental Studies. “The thing about Frank was that he was extremely organised, and that’s why he was so effective at setting things up”, said Professor Peter Doherty, who won a Nobel Prize for research conducted at the John Curtin School. “In science, good institutions are very important, and he was important in setting up good institutions.” Fenner’s organisational skills also played a vital part in ending smallpox as a human problem. He served as chairman of the Global Commission for the Certification of Smallpox Eradication, a body set up to confirm the work of WHO’s mammoth eradication programme. In May, 1980, he presented a report to the World Health Assembly stating that the effort had been successful. “He was the prince of Australian virologists”, said Nossal. “He was a modest and self-effacing person, steady and steadfast, and just absolutely focused and systematic.” Fenner is survived by his daughter, Marilyn, two grandchildren, and a great-grandchild.
Stephen Pincock
[email protected]
www.thelancet.com Vol 377 January 1, 2011
Correspondence
that sucrose has long-term negative effects.5 We agree with Slater and colleagues that every effort should be made to minimise nociceptive pathway activation in infants.
Rebeccah Slater and colleagues (Oct 9, p 1225)1 question the benefit of sucrose for alleviating procedural pain in infants. We believe that they might have overstated their conclusions and suggest a more cautious interpretation of the study findings. In agreement with other studies,2 Slater and colleagues report a significant reduction in observed pain scores in infants given sucrose compared with placebo. An intriguing observation was the absence of any difference in nociceptive pathway activity in infants who received sucrose or placebo. On this basis, Slater and colleagues conclude that oral sucrose is ineffective for pain relief in infants, and discourage its routine use in infants. We disagree that the documented cortical measures are a reliable surrogate of perceived pain in infants. The WHO definition of pain refers to the subjective and emotional experience being of primary importance. Our access to infants’ emotions is through their behaviour. The fact that “pain pathway nerves” are active during procedures in infants who are calm after sucrose is not unexpected. We disagree that a reduction in clinical observation scores should not be interpreted as pain relief. Limitations of the study include the small sample size and that electrophysiological monitoring was limited in spatial and temporal domains. The effect of sucrose on neural pathways of the brain relevant to emotional responses to pain was not investigated. Whether a blunting of the behavioural response to pain reduces the adverse effects of nociceptive pathway activation remains to be determined.3 Substantial media interest has been generated, questioning the safety of sucrose administration on the developing brain.4 There is no evidence
We declare that we have no conflicts of interest.
www.thelancet.com Vol 377 January 1, 2011
D Steed, L Port, *T G Connell, J Standish, J Munro, L Takacs, I McKenzie
[email protected] Department of General Medicine (DS, LP, TGC, JS, JM) and Comfort Kids Program, Department of Anaesthesia and Pain Management (LT, IM), Royal Children’s Hopital Melbourne, Parkville, Melbourne, VIC 3052, Australia 1
2
3
4
5
Slater R, Cornelissen L, Fabrizi L, et al. Oral sucrose as an analgesic drug for procedural pain in newborn infants: a randomised controlled trial. Lancet 2010; 376: 1225–32. Stevens B, Yamada J, Ohlsson A. Sucrose for analgesia in newborn infants undergoing painful procedures. Cochrane Database Syst Rev 2010; 1: CD001069. Hohmeister J, Kroll A, Wollgarten-Hadamek I, et al. Cerebral processing of pain in school-aged children with neonatal nociceptive input: an exploratory fMRI study. Pain 2010; 150: 257–67. Campbell D. Newborn babies should not be given sugar as pain relief, says study. The Guardian Sept 2, 2010. Stevens B, Yamada J, Beyene J, et al. Consistent management of repeated procedural pain with sucrose in preterm neonates: is it effective and safe for repeated use over time? Clin J Pain 2005; 21: 543–48.
Practitioners should disregard the advice of Rebeccah Slater and colleagues1 that sucrose “should not be used routinely for procedural pain in infants”. Such a change in practice would be a retrograde step in the struggle to provide optimum care for newborn infants, and is not supported by the study itself nor the references they quote. Slater and colleagues state that there is “absence of evidence for an analgesic action of sucrose in this study”. Analgesia is defined as “absence of pain in response to stimulation which would normally be painful”. Seven of 20 “facial non-responders” in the sucrose group seem to have had an analgesic action, particularly given that none of the 24 participants in the sterile water group exhibited this effect. Slater and colleagues speculate that this effect of sucrose could occur
at the midbrain owing to its ability to “inhibit facial motor activity”. “Absence of evidence of effect in this study”, does not mean evidence of absence of true effect. The study was designed with a power of 80%, meaning that the probability of missing a true effect is 20%. Slater and colleagues define a drop of 30% amplitude during a 450–750 ms section of their EEG data as clinically significant, yet do not explain why. Furthermore Slater and colleagues do not explain adequately why they analysed only a 300 ms section of their data; there does seem to be a visible difference in the trace with sucrose compared with water at 750–950 ms. This might support the notion that there is indeed a nociceptive response with sucrose. The question remains: does EEG nociceptive-specific brain activity at a particular locus during a specific time in the presence of oral sucrose mean that the child had pain but could not express it; or has the child received the signal from the heel but not perceived it as pain and therefore not expressed pain behaviour? A further important question: is it the nociception or the subsequent cortical pain response that leads to long-term sequelae, and how might sucrose modify this response?
Science Photo Library
Oral sucrose for procedural pain in infants
We declare that we have no conflicts of interest.
Paul A Heaton, *Andrew M Fernando, David Herd
[email protected] Yeovil District Hospital, Higher Kingston, Yeovil BA21 4AT, UK (PAH, AMF); and Mater Children’s Hospital, Raymond Terrace, South Brisbane, QLD, Australia (DH) 1
Slater R, Cornelissen L, Fabrizi L, et al. Oral sucrose as an analgesic drug for procedural pain in newborn infants: a randomised controlled trial. Lancet 2010; 376: 1225–32.
Rebeccah Slater and colleagues1 report that oral sucrose did not diminish localised brain activity (electroencephalography [EEG]) and a spinal nociceptive reflex (electromyography [EMG]) after heel lance. Conversely, sucrose significantly reduced behavioural expressions, consistent with a Cochrane review.2 We consider the
Submissions should be made via our electronic submission system at http://ees.elsevier.com/ thelancet/
25
Correspondence
conclusion, “sucrose should not be used routinely for procedural pain in infants without further investigation”1 unjustified. The selected indices did not recognise pain as engaging widely distributed, serial and parallel, dynamic activation in multiple areas of the brain.3,4 EEG activity was reported for only one of 32 recording electrodes, and thus did not address multiple brain regions that might have been activated.5 The suggestion of dissociation between brain activity and the experience and expression of pain is reminiscent of Descartes’ dualism. The intervention clearly affected behaviour, with the neural measures not identifying the biological substrate. Other methodological issues could account for differences between the neural and behavioural measures. The single control procedure was not randomly ordered. Only 44 of 59 infants were included in the analysis of EEG and premature infant pain profile (34 of 59 in the EMG analysis). These 25% and 42% attrition rates, respectively, might have eliminated infants with vigorous facial displays contributing to dissociation between measures. Although infant pain assessment is imperfect and imprecise, caution should prevail concerning conclusions on well supported pain-relieving strategies. Because one selected measure of neural activity was not responsive to sucrose does not preclude other measures from being responsive. Researchers and clinicians have an ethical responsibility to consider oral sucrose for painful procedures, while doing further effectiveness research using validated cognitive and behavioural assessment measures. We declare that we have no conflicts of interest.
*Bonnie Stevens, Kenneth Craig, Celeste Johnston, Denise Harrison, Arne Ohlsson
[email protected] Centre for Nursing, Hospital for Sick Children, Toronto, ON M5G 1X8, Canada (BS); University of British Columbia, Vancouver, BC, Canada (KC);
26
McGill University, Montreal, QC, Canada (CJ); Royal Children’s Hospital, Melbourne, VIC, Australia (DH); and University of Toronto, Toronto, ON, Canada (BS, AO) 1
2
3
4 5
Slater R, Cornelissen L, Fabrizi L, et al. Oral sucrose as an analgesic drug for procedural pain in newborn infants: a randomised controlled trial. Lancet 2010; 376: 1225–32. Stevens B, Yamada J, Ohlsson A. Sucrose for analgesia in newborn infants undergoing painful procedures. Cochrane Database Syst Rev 2010; 1: CD001069. Price DD. Psychological and neural mechanisms of the affective dimension of pain. Science 2000; 288: 1769–72. Tracey I. Imaging pain. Br J Anaesth 2008; 101: 32–39. Hofbauer RK, Rainville P, Duncan GH, Bushnell MC. Cortical representation of the sensory dimension of pain. J Neurophysiol 2001; 86: 402–11.
in the context of conclusions were premature infant pain profile and facial expressions, both of which were affected by sucrose, as shown in many earlier studies.5 Development of objective physiological tools for assessment of neonatal pain is a major endeavour, which calls for careful and solid reasoning. Advanced-looking tools might bring a high-impact audience to old discoveries. However, use of unfounded tools will readily bring unsound conclusions, which might sooner or later become counterproductive to the field itself. I declare that I have no conflicts of interest.
Rebeccah Slater and colleagues1 studied the effect of sucrose in treating neonatal pain, and they conclude that “oral sucrose does not significantly affect activity in neonatal brain or spinal cord nociceptive circuits, and therefore might not be an effective analgesic drug”. The neurophysiological data considered by Slater and colleagues as the primary outcome measure do not really provide a physiologically sensible bridge to their conclusions. Slater and colleagues call their electroencephalography (EEG) response a “specific nociceptive brain activity”, because they earlier showed it to come after heel lance but not after a tactile touch. It is obvious that a heel lance also results in arousal,2 causing a comparable scalp response, which can even be seen after an auditory stimulus.3 Hence, the EEG response seen by Slater and colleagues is not proven to be specific to the noxiousness of the stimulus. Slater and colleagues also measured spinal reflexes as the other index of pain response. As every neurologist would expect, spinal reflexes to noxa are not affected by sucrose, which naturally works at supraspinal level. The immaturity of descending inhibition4 makes this approach particularily poor in neonates, and useless with respect to Slater and colleagues’ conclusions. The sound measures
Sampsa Vanhatalo sampsa.vanhatalo@helsinki.fi Department of Children’s Clinical Neurophysiology, PO Box 280 (Lastenlinnantie 2), 00029 Helsinki, Finland 1
2
3
4 5
Slater R, Cornelissen L, Fabrizi L, et al. Oral sucrose as an analgesic drug for procedural pain in newborn infants: a randomised controlled trial. Lancet 2010; 376: 1225–32. Norman E, Rosén I, Vanhatalo S, et al. Electroencephalographic response to procedural pain in healthy term newborn infants. Pediatr Res 2008; 64: 429–34. Hrbek A, Hrbkova M, Lenard H-G. Somato-sensory, auditory and visual evoked responses in newborn infants during sleep and wakefulness. Electroencephalogr Clin Neurophysiol 1969; 26: 597–603. Fitzgerald M. The development of nociceptive circuits. Nat Rev Neurosci 2005; 6: 507–20. Stevens B, Yamada J, Ohlsson A. Sucrose for analgesia in newborn infants undergoing painful procedures. Cochrane Database Syst Rev 2010; 1: CD001069.
Rebeccah Slater and colleagues1 conclude that “oral sucrose does not significantly affect activity in neonatal brain or spinal cord nociceptive circuits, and therefore might not be an effective analgesic drug”. In the Summary, they state that “59 newborn infants…were randomly assigned to receive 0·5 mL 24% sucrose solution or 0·5 mL sterile water 2 min before undergoing a clinically required heel lance.” In two previous systematic reviews about sucrose,2,3 the single dose commonly used was 2 mL in full-term and preterm infants and 0·5 mL per kg for www.thelancet.com Vol 377 January 1, 2011
Correspondence
We declare that we have no conflicts of interest.
*Maria Beatriz Martins Linhares, Cláudia Maria Gaspardo, Francisco Eulógio Martinez
[email protected] Faculty of Medicine of Ribeirão Preto, University of São Paulo, Ribeirão Preto, São Paulo 14024250, Brazil 1
2
3
4
Slater R, Cornelissen L, Fabrizi L, et al. Oral sucrose as an analgesic drug for procedural pain in newborn infants: a randomised controlled trial. Lancet 2010; 376: 1225–32. Gaspardo CM, Linhares MBM, Martinez FE. The efficacy of sucrose for the relief of pain in neonates: a systematic review of the literature. J Pediatr (Rio J) 2005; 81: 435–42. Stevens B, Yamada J,Ohlsson A. Sucrose for analgesia in newborn infants undergoing painful procedures. Cochrane Database Syst Rev 2010; 1: CD001069. Gaspardo CM, Miyase CI, Chimello JT, Martinez FE, Linhares MBM. Is pain relief equally efficacious and free of side effects with repeated doses of oral sucrose in preterm neonates? Pain 2008; 137: 16–25.
Authors’ reply Given the importance of pain and pain relief in clinical practice, the area is extraordinarily poorly understood— nowhere more so than in the infant population. Noxious procedures are repeatedly performed on infants in hospital, and the short-term and long-term consequences of this are of increasing concern.1 Current clinical assessment of infant pain, such as the premature infant pain profile (PIPP), makes use of indirect www.thelancet.com Vol 377 January 1, 2011
methods that have not been validated against other objective independent measures. Despite this lack of validation, these measures are widely used as a measure of “infant pain” and, on the basis of these measures, application of a sweet solution to the infant’s mouth is thought sufficient to alleviate procedural pain. D Steed and colleagues’ claim that “access to infants’ emotions is through their behaviour” is not supported by mechanistic data or hypothesis. The acceptance of such claims will deflect attention from the profound activity-dependent changes in brain development that noxious stimulation can evoke and from asking how this can be prevented. In our study, we showed that activity in the brain after a noxious procedure is not affected by sucrose, despite the well documented effects of sucrose on facial expression. It is fact, not speculation as Paul Heaton and colleagues claim, that sucrose can reduce pain-related behaviour through brainstem mechanisms.2 We therefore challenge the community to question whether infants who receive sucrose really are adequately relieved of their pain and, more importantly, adequately protected from the potential adverse effects that pain might cause to the developing nervous system. Our electroencephalography (EEG) recordings were not used (as suggested by Bonnie Stevens and colleagues) to localise brain activity but to detect real-time nociceptive-specific activity in the brain. We recorded from 32 electrodes and, because the nociceptive-specific potential is maximal at the midline,3 we used the signal at a single central electrode as an outcome measure. Furthermore, we disagree with Heaton and colleagues that our decision to focus on the EEG activity in the time-window from 450 to 750 ms after stimulation is not justified: this study, as well as our previous publications, show that nociceptive-specific brain activity is present during this time frame.3
The nociceptive-specific potential cannot be compared to the “arousal” responses referred to by Sampsa Vanhatalo: they are not statedependent and have different amplitude, latency, and morphology. They cannot easily be seen by use of frequency analysis of non-timelocked EEG, which, in the study by Norman and colleagues,4 revealed only muscle activity. Thus by use of careful and rigorous (albeit novel) experimental analysis, we have identified a nociceptive potential that can be used to explore analgesic regimens and improve the experience of infants in hospital. Our paper did not show that sucrose was positively harmful and we have no evidence that this is the case (as was erroneously suggested in the press). One might question the wisdom of giving the equivalent of a “half can of Coke Classic per day”5 to infants in the name of analgesia when activation of the brain goes on unabated. There is also a distinct possibility that the routine use of sucrose has led to other potentially effective analgesic strategies (for example “kangaroo care” or pharmacological approaches) being neglected. What our study did show is that we cannot afford to make assumptions about analgesia simply because of an unfounded belief that the PIPP score taps into the infant “pain experience” when direct measurements from the infant brain are telling another story.
Science Photo Library
repeated doses in preterm infants. In our study,4 we tested the efficacy of oral sucrose 2 min before puncture for blood collection in preterm infants using the 0·5 mL/kg dose. Considering that participants in Slater and colleagues’ study were full-term infants with a mean birthweight of 3449 g (SD 453), the question is whether the single dose of 0·5 mL used was sufficient to obtain analgesia. We suggest that this methodological question should be addressed in future studies before concluding that “the ability of sucrose to reduce clinical observational scores after noxious events in newborn infants should not be interpreted as pain relief”.1
We declare that we have no conflicts of interest.
*Rebeccah Slater, Stewart Boyd, Judith Meek, Maria Fitzgerald
[email protected] Nuffield Department of Anaesthetics, University of Oxford, Oxford, UK (RS); Neuroscience, Physiology and Pharmacology, University College London, London WC1E 6BT, UK (RS, MF); Department of Clinical Neurophysiology, Great Ormond Street Hospital for Children, London, UK (SB); and Elizabeth Garrett Anderson Obstetric Wing, University College Hospital, London, UK (JM) 1
Anand KJ, Aranda JV, Berde CB, et al. Summary proceedings from the neonatal pain-control group. Pediatrics 2006; 117 (3 Pt 2): S9–22.
27
Correspondence
2
3
4
5
Anseloni VC, Ren K, Dubner R, Ennis M. A brainstem substrate for analgesia elicited by intraoral sucrose. Neuroscience 2005; 133: 231–43. Slater R, Worley A, Fabrizi L, et al. Evoked potentials generated by noxious stimulation in the human infant brain. Eur J Pain 2010; 14: 321–26. Norman E, Rosén I, Vanhatalo S, et al. Electroencephalographic response to procedural pain in healthy term newborn infants. Pediatr Res 2008; 64: 429–34. Holsti L, Grunau RE. Considerations for using sucrose to reduce procedural pain in preterm infants. Pediatrics 2010; 125: 1042–47.
Corbis
International response to Niger’s hunger crisis
For The Lancet’s Series on maternal and child undernutrition see http://www.thelancet.com/ series/maternal-and-childundernutrition
28
Samuel Loewenberg’s World Report entitled “Niger’s hunger crisis: a legacy of lessons unlearned” (Aug 21, p 579)1 paints an unfair picture of the international community’s response to last year’s food crisis in Niger. It is regretful that Loewenberg was not able to reflect more accurately the extensive international and European mobilisation to mitigate the effect of this crisis. The risk of a food crisis in the Sahel was identified as early as September, 2009, once it was clear that the erratic rains in July and August, 2009, would negatively affect the size of the harvest. The European Commission started working with the local authorities and with international non-governmental organisations, the Red Cross, and UN partners to raise awareness and mobilise support. It made available in late 2009 a first allocation of €10 million to assist partners in mitigation and preparedness action. Our financial support was stepped up as the food crisis worsened and continued high food prices pushed many households over the edge into crisis. In total, the Commission allocated €108·9 million in humanitarian assistance to address the situation and was instrumental in mobilising further international support through a “Friends of the Sahel” network.
Our funding helped to ensure the large-scale availability and distribution of special ready-to-use therapeutic foods, such as PlumpyNut (which had proven to be so effective in 2005), and supported blanket feeding operations to reach as many children at risk as possible. We actively encouraged the large-scale use of direct cash transfers to the most vulnerable, which was instrumental in ensuring that food remained available in local markets, albeit at a high price. We also sought to facilitate food purchases in the countries of the region, where possible, which provided an incentive to local production. In doing so, we have put into practice the lessons learned from the 2005 Niger food crisis—which were also reflected in The Lancet Series on undernutrition, published in 2008. There can be no doubt that many tens of thousands of lives were saved as a result. These actions were additional to the Commission’s existing long-term strategy to promote nutritional security in the Sahel. The Commission had adopted in 2007 a Sahel strategy based on the three pillars of improving the knowledge base; providing support for pilot, innovative, and replicable action to reduce undernutrition; and advocacy to raise awareness in government and development partners on nutrition issues. The effect of this strategy is starting to show in the increased importance given to food and nutrition security by government and development partners in the region and by the speed in which partners responded to the risks caused by last year’s crisis. This is not to say that everything has been perfect, but we have striven to learn from the past and to apply more modern and efficient methods. And this deserves some acknowledgment. I am the European Union’s Commissioner for International Cooperation, Humanitarian Aid and Crisis Response.
Kristalina Georgieva
[email protected] European Commission, 1049 Brussels, Belgium 1
Loewenberg S. Niger’s hunger crisis: a legacy of lessons unlearned. Lancet 2010; 376: 579–81.
The Greek economic crisis: a primary health-care perspective The Greek economic crisis has caused global concern owing to its sideeffects and risks involved for both the eurozone and the global economic community. Analysis of the problem shows that the major sources of inadequacy are the severe structural weaknesses in Greek public administration, economy, and society, which lead to bureaucracy, corruption, low quality of services, and high costs. Greek health care is a typical example of a bleeding economic sector. Although the total health-care expenditure rose from 5·3% of gross domestic product in 1991 to 9·7% in 2008, Greece has actually got worse in terms of global health outcomes.1 Why is that? Administrative barriers affect every aspect of the Greek healthcare system, with perhaps the largest insufficiencies located in primary care.2 In general, Greek primary care is highly fragmented, since there are several different public and private providers involved, with no coordination between them and no gatekeeping system.3 Some of the main reasons for the high primary health-care costs are the repetition of tests and prescriptions due to poor information transfer between providers and the vast induced demand for health-care services. The latter is perhaps the most important factor and can be explained by the large number of specialists—the highest among countries of the Organisation for Economic Cooperation and Development—and the www.thelancet.com Vol 377 January 1, 2011
Correspondence
ineffectiveness of the existing control mechanisms of health insurance funds who incur the costs.4 We strongly believe that the above problems could be addressed by ensuring adequate political will and social consensus. The integration of all primary-care providers, the establishment of the multidisciplinary primary-care team, and the enactment of the family doctor institution are necessary to obtain comprehensive, continuous, and efficient health care.5 The introduction and implementation of diagnostic and treatment guidelines in daily clinical practice could result in a better quality of care and a rationalisation of health-care expenditure. The induced demand for medical services seems an intractable problem, but might be controlled by a reduction in the production of new physicians. Greece is struggling to manage its fiscal problems, overcome recession, and maintain a social welfare state. To achieve these goals and get out of the crisis, it is time for Greece to capitalise on moral, economic, political, and scientific support from international bodies and advanced nations by making drastic organisational reforms in all aspects of the public sector, including health care. We declare that we have no conflicts of interest.
*Nikolaos Oikonomou, Yannis Tountas
[email protected] Centre for Health Services Research, Medical School, University of Athens, 115 27 Athens, Greece 1 2
3
4
5
OECD. OECD health data 2010. Paris: OECD, 2010. Tountas Y, Karnaki P. The ‘‘unexpected’’ growth of the private health sector in Greece. Health Policy 2005; 74: 167–80. Mossialos E, Allin S, Davaki K. Analysing the Greek health system: a tale of fragmentation and inertia. Health Econ 2005; 14 (suppl 1): S151–68. Oikonomou N, Mariolis A. Three trends that undermine the Greek health system: is there a way out? Eur J Gen Pract 2009; 15: 67–68. Lionis C, Symvoulakis EK, Markaki A, et al. Integrated primary health care in Greece, a missing issue in the current health policy agenda: a systematic review. Int J Integr Care 2009; 9: e88.
www.thelancet.com Vol 377 January 1, 2011
When will the sun shine on Cyprus’s National Health Service? It has been 9 years since an official parliamentary law was passed in Cyprus to implement a primarycare-driven health-care coverage system for the entire population. Since then, the Health Insurance Organisation has been set up as the single payer, which, in collaboration with McKinsey consultants, has brought forth a detailed design for the long-awaited Cyprus National Health Service (NHS). However, despite the project reaching its final stages, a cloud of doubt remains over whether the government will take the plunge and implement what it has so carefully planned. Little has changed since a World Report in The Lancet stated in 2005 that “there has not been sufficient pressure on the government to adopt a universal health plan”.1 Cypriot health care is still divided, with 50% of the population using public healthcare services and 50% private health care. Primary care is underdeveloped, with general practitioners (GPs) comprising only 9·8% of all physicians, and with a decreasing trend.2 This proportion compares with an average of 25·5% for the European Union3 and about 50% for the UK. Patients have often not heard of a “GP”, and certainly not registered with one. Thus, there is commonly no continuity or co-ordination of care. Quality-improvement strategies are lacking.4 The isolation between private and public health care leads to duplication of tests, and of resource use.5 Specifically, the public-sector cost is growing at a double-digit rate, thereby reaching a level in the near future at which the government budget will not be able to sustain it. Furthermore, direct access to specialists often leads to the patient
not having a unified health record; additionally, holistic preventive care often slips through the net. By contrast with other EU countries, Cypriots pay 52% of healthcare cost out-of-pocket (43% of the population pay for private prescriptions). The new NHS proposed for Cyprus aims at equity in finance, universal coverage, shifting of service provision from secondary to primary care, efficiency, high quality standards, and containment of cost through a global budget. Every person will be registered with a personal primarycare physician of their choice. The free choice of specialist physician (public or private) after referral as well as hospital (public or private) will also lie with the patient. We urge the government to take concrete steps towards implementing the health-care reform— ie, finalise the NHS information technology system and GPs’ training tender—thus moving Cyprus’s health care forward while helping to control health expenditure growth. We declare that we have no conflicts of interest.
*George Samoutis, Constantinos Paschalides
[email protected] Messinis 3, 2301 Nicosia, Cyprus (GS); and Tiptree Medical Centre, Tiptree, UK (CP) 1
2
3
4
5
Antoniadou M. Can Cyprus overcome its health-care challenges? Lancet 2005; 365: 1017–20. Cyprus Medical Association. Registry of medical specialties in Cyprus. http://www. cyma.org.cy/ (accessed Dec 14, 2010). WHO Regional Office for Europe. European health for all database. http://www.euro. who.int/en/what-we-do/data-andevidence/databases/european-healthfor-all-database-hfa-db2 (accessed Dec 14, 2010). Samoutis GA, Soteriades ES, Stoffers HE, Philalithis A, Delicha EM, Lionis C. A pilot quality improvement intervention in patients with diabetes and hypertension in primary care settings of Cyprus. Fam Pract 2010; 27: 263–70. Andreou M, Pashardes P, Pashourtidou N. Cost and value of health care in Cyprus. Policy paper. Nicosia: University of Cyprus, 2010. http://www.ucy.ac.cy/data/ ecorece/DOP02-10.pdf (accessed Dec 14, 2010).
29
Correspondence
Reporting on the modes of data collection
Science Photo Library
To gather more insight into the methods of data collection currently used in medical research, we analysed research studies published in four high-impact general medical journals and three epidemiological journals in 2008–09.1 Two epidemiologists independently assessed the modes of data collection reported in each research paper. Surprisingly, the proportion of inconsistencies between the two reviewers was high, especially for papers published in general medical journals (about 30%). Further examination revealed that these inconsistencies were mainly due to unclear reporting of the methods used, with phrases such as “Information was collected on [list of variables]”, “Race/ethnicity was assessed by the investigator or study coordinator”, and “Sociodemographic, clinical, treatment (…), and laboratory data are collected” without any specification. Did they use questionnaires or interviews, were any measurements taken, or was it all hearsay? The choice of the method of data collection for a particular study depends on several factors, including, but not limited to, the type of study, sensitivity of the topic of interest, and costs of the measurements.2 Valid measurement of exposures, outcomes, and potential confounders is essential in medical research to prevent biased results.3 Since different methods of data collection yield various amounts of measurement error, detailed reporting on the methods used is of great importance to assess the quality of the study by both readers and reviewers or editors. Additionally, an adequate description of the methods of data collection used enables other research groups to replicate the original study. In our view, researchers should improve the description of the modes of data collection used in their studies. 30
In addition, we encourage medical journals to pay more attention to the way in which the methods are reported to improve the possibilities of critical appraisal. We declare that we have no conflicts of interest.
The Lancet has rightly emphasised that authors of papers must make a substantial contribution.3 It is also important that all major contributors are recognised, either as an author or with an acknowledgment.
*Marleen M H J van Gelder, Reini W Bretveld, Nel Roeleveld
I have received payment for lectures or speakers’ bureaux from Dako, Ipsogen, Hammersmith Hospital, and Netherlands Cancer Institute.
[email protected] Department of Epidemiology, Biostatistics and HTA, Radboud University Nijmegen Medical Centre, 6500 HB Nijmegen, Netherlands 1
2
3
Van Gelder MMHJ, Bretveld RW, Roeleveld N. Web-based questionnaires: the future in epidemiology? Am J Epidemiol 2010; 172: 1292–98. Armstrong BK, White E, Saracci R. Principles of exposure measurement in epidemiology. New York: Oxford University Press, 1992. Rothman KJ, Greenland S, Lash TL. Validity in epidemiologic studies. In: Rothman KJ, Greenland S, Lash TL. Modern epidemiology, 3rd edn. Philadelphia: Lippincott Williams & Wilkins, 2008: 137–46.
Andrew H S Lee
[email protected] Department of Histopathology, Nottingham University Hospitals, City Hospital Campus, Nottingham NG5 1PB, UK 1 2
3
Kang S. Anecdotes in medicine—15 years of Lancet case reports. Lancet 2010; 376: 1448–49. Wilson R, Liston J, Cooke J, et al. Clinical guidelines for breast cancer screening assessment. Sheffield: NHS Cancer Screening Programmes, 2005. The Lancet. The role and responsibilities of coauthors. Lancet 2008; 372: 778.
All important contributions to papers should be recognised In her Comment (Oct 30, p 1448),1 Seema Kang discusses Case Reports in The Lancet according to specialty interest and mentions that “more than one specialty might be involved in managing a case”. She comments on the absence of contributions from general practice, but makes no comment on the contribution of other specialities such as radiology and pathology. Many patients require multidisciplinary investigation or management. For example, in the diagnosis of breast disease, the triple approach of clinical examination, radiology, and pathology is recommended.2 Over the past 2 years (November, 2008, to October, 2010) histopathology had an important role in the diagnosis in 36 Case Reports in The Lancet. A histopathologist was an author in 20. In the other 16 there was no apparent recognition of histopathology in terms of authorship or acknowledgment, although it had an essential role in ten and was contributory in six.
Department of Error Tatem AJ, Smith DL, Gething PW, Kabaria CW, Snow RW, Hay SI. Ranking of elimination feasibility between malaria-endemic countries. Lancet 2010; 376: 1579–91—In this Series paper (Nov 6), the fourth sentence under the heading “Estimation of intensity of endemic P falciparum transmission” should have read: “R0 is a measure of maximum potential transmission, and if R0 is 1 or greater, then endemic malaria transmission can be sustained”. Additionally, the title for figure 1 should have read “Categorical map of Plasmodium falciparum reproductive number, PfRc, indicating the extent to which transmission needs to be reduced for elimination”. These corrections have been made to the online version as of Dec 31, 2010. The InFACT Global H1N1 Collaboration. InFACT: a global critical care research response to H1N1. Lancet 2010; 375: 11–13—In this Comment (Jan 2), the name of the third member of the InFACT Global H1N1 Collaboration was misspelled. The correct spelling is “Neill K J Adhikari”. This correction has been made to the online version as of Dec 31, 2010.
www.thelancet.com Vol 377 January 1, 2011
Articles
Effect of daily aspirin on long-term risk of death due to cancer: analysis of individual patient data from randomised trials Peter M Rothwell, F Gerald R Fowkes, Jill F F Belch, Hisao Ogawa, Charles P Warlow, Tom W Meade
Summary Background Treatment with daily aspirin for 5 years or longer reduces subsequent risk of colorectal cancer. Several lines of evidence suggest that aspirin might also reduce risk of other cancers, particularly of the gastrointestinal tract, but proof in man is lacking. We studied deaths due to cancer during and after randomised trials of daily aspirin versus control done originally for prevention of vascular events.
Lancet 2011; 377: 31–41
Methods We used individual patient data from all randomised trials of daily aspirin versus no aspirin with mean duration of scheduled trial treatment of 4 years or longer to determine the effect of allocation to aspirin on risk of cancer death in relation to scheduled duration of trial treatment for gastrointestinal and non-gastrointestinal cancers. In three large UK trials, long-term post-trial follow-up of individual patients was obtained from death certificates and cancer registries.
See Comment page 3
Results In eight eligible trials (25 570 patients, 674 cancer deaths), allocation to aspirin reduced death due to cancer (pooled odds ratio [OR] 0·79, 95% CI 0·68–0·92, p=0·003). On analysis of individual patient data, which were available from seven trials (23 535 patients, 657 cancer deaths), benefit was apparent only after 5 years’ follow-up (all cancers, hazard ratio [HR] 0·66, 0·50–0·87; gastrointestinal cancers, 0·46, 0·27–0·77; both p=0·003). The 20-year risk of cancer death (1634 deaths in 12 659 patients in three trials) remained lower in the aspirin groups than in the control groups (all solid cancers, HR 0·80, 0·72–0·88, p<0·0001; gastrointestinal cancers, 0·65, 0·54–0·78, p<0·0001), and benefit increased (interaction p=0·01) with scheduled duration of trial treatment (≥7·5 years: all solid cancers, 0·69, 0·54–0·88, p=0·003; gastrointestinal cancers, 0·41, 0·26–0·66, p=0·0001). The latent period before an effect on deaths was about 5 years for oesophageal, pancreatic, brain, and lung cancer, but was more delayed for stomach, colorectal, and prostate cancer. For lung and oesophageal cancer, benefit was confined to adenocarcinomas, and the overall effect on 20-year risk of cancer death was greatest for adenocarcinomas (HR 0·66, 0·56–0·77, p<0·0001). Benefit was unrelated to aspirin dose (75 mg upwards), sex, or smoking, but increased with age—the absolute reduction in 20-year risk of cancer death reaching 7·08% (2·42–11·74) at age 65 years and older. Interpretation Daily aspirin reduced deaths due to several common cancers during and after the trials. Benefit increased with duration of treatment and was consistent across the different study populations. These findings have implications for guidelines on use of aspirin and for understanding of carcinogenesis and its susceptibility to drug intervention. Funding None.
Introduction In the developed world, the lifetime risk of cancer is about 40%, and rates are increasing in the developing world.1 In Europe, about 3·2 million new cancers present each year, with about 1·7 million deaths,2 and there are more than 1·5 million new cases each year in the USA.3 By contrast with treatment of cancer, there has been little progress in use of drugs in prevention of the disease. However, several lines of evidence suggest that long-term use of aspirin might reduce the risk of some cancers, particularly gastrointestinal tumours. Aspirin reduces incidence or growth rate, or both, of several cancers in animal models,4–6 mediated at least in part by inhibition of the cyclo-oxygenase (COX) enzymes and reduced production of prostaglandins and other inflammatory mediators, but these findings might not be applicable to humans. Observational studies in humans also suggest that aspirin reduces risk of certain cancers,5–8 but results www.thelancet.com Vol 377 January 1, 2011
have been conflicting, with more rigorous studies yielding weaker associations.8 Moreover, observational studies have proved to be unreliable in determining risks and benefits of medications in the past,9,10 and there is trial evidence that one antiplatelet drug might have adverse effects on cancer outcomes.11 Nevertheless, long-term follow-up of randomised trials has shown that aspirin does reduce the risk of colorectal cancer after a delay of several years,12,13 probably by reducing precancerous adenomas,14 possibly by inhibition of COX-2.15 However, proof of an effect on other cancers is lacking. 10-year follow-up of the Women’s Health Study, a randomised trial of 100 mg of aspirin on alternate days versus control, showed no reduction in incidence of cancer.16 However, aspirin also failed to prevent colorectal adenomas in this study, which is consistent with observational studies suggesting that daily aspirin is required for
Published Online December 7, 2010 DOI:10.1016/S01406736(10)62110-1
Stroke Prevention Research Unit, Department of Clinical Neurology, University of Oxford, Oxford, UK (Prof P M Rothwell FMedSci); Wolfson Unit for Prevention of Peripheral Vascular Diseases, Centre for Population Health Sciences, University of Edinburgh, Edinburgh, UK (Prof F G R Fowkes FRCPE); The Institute of Cardiovascular Research, Vascular and Inflammatory Diseases Research Unit, University Division of Medicine and Therapeutics, Ninewells Hospital and Medical School, Dundee, UK (Prof J F F Belch FRCP); Department of Cardiovascular Medicine, Graduate School of Medical Sciences, Kumamoto University, Kumamoto City, Japan (H Ogawa MD); Department of Clinical Neurosciences, University of Edinburgh, Western General Hospital, Edinburgh, UK (Prof C P Warlow FMedSci); and London School of Hygiene and Tropical Medicine, University of London, London, UK (Prof T W Meade FRS) Correspondence to: Prof Peter M Rothwell, Stroke Prevention Research Unit, University Department of Clinical Neurology, John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
[email protected]. ac.uk
31
Articles
Deaths due to cancer Aspirin Control BDAT19 UK-TIA18 ETDRS20 SAPAT22 TPT17 JPAD21 POPADAD23 AAA24 Total
75/3429 21/1621 16/1856 10/1009 87/2545 15/1262 25/638 78/1635
47/1710 23/814 14/1855 19/1026 104/2540 19/1277 31/638 90/1675
Odds ratio (95% CI)
0·79 (0·55–1·14) 0·45 (0·25–0·82) 1·14 (0·56–2·35) 0·53 (0·25–1·15) 0·83 (0·62–1·11) 0·80 (0·40–1·57) 0·80 (0·47–1·37) 0·86 (0·63–1·17)
327/14 035 347/11 535 0·79 (0·68–0·92) 0
psig=0·003, phet=0·84
1 Odds ratio (95% CI)
2
Figure 1: Meta-analysis of the effect of aspirin on deaths due to cancer during all eligible randomised trials of aspirin versus control Data are n/N, where n=number of cancer deaths and N=number of trial participants in that treatment group. BDAT=British Doctors Aspirin Trial. UK-TIA=UK transient ischaemic attack trial. ETDRS=Early Treatment Diabetic Retinopathy Study. SAPAT=Swedish Angina Pectoris Aspirin Trial. TPT=Thrombosis Prevention Trial. JPAD=Japanese Primary Prevention of Atherosclerosis With Aspirin for Diabetes. POPADAD=Prevention of Progression of Arterial Disease and Diabetes. AAA=Aspirin for Asymptomatic Atherosclerosis.
prevention of cancer.5–8 Observational studies also suggest that use of aspirin for at least 5 years is required before reductions in risk of cancer are observed,5–8 and the effect of aspirin on risk of colorectal cancer on follow-up of randomised trials was greatest in patients with duration of trial treatment of 5 years or longer.12,13 We therefore determined the effect of aspirin on risk of fatal cancer by analysis of individual patient data for deaths due to cancer during randomised trials of daily aspirin versus control (done originally for primary or secondary prevention of vascular events) in which the median duration of scheduled trial treatment was at least 4 years. We studied fatal cancers only, in the first instance, because cause of non-vascular deaths was reliably determined in most aspirin trials, and we also aimed to determine the effect of any reduction in cancer deaths on overall all-cause mortality. In three trials done in the UK,17–19 we also determined any delayed effects of aspirin on the 20-year risk of death due to cancer by long-term post-trial follow-up.
Methods Search strategy and selection criteria We searched for randomised trials of aspirin versus control that had a mean or median scheduled trial treatment period of at least 4 years and a range extending beyond 5 years. Eligible trials had investigated the effects of randomised allocation to: aspirin (any dose) versus no aspirin in the absence of another agent; or aspirin (any dose) versus no aspirin in the presence of another antiplatelet agent or antithrombotic agent, if the other agent was used in the same way in the aspirin and no aspirin groups. Given the focus on cancer outcomes, no distinction was made between trials of 32
aspirin in primary versus secondary prevention of vascular disease. In view of the availability of published data for all trials of antiplatelet agents from the Antithrombotic Trialists’ (ATT) Collaboration, literature searches were confined to publications after the last ATT search (2002).20,21 Trials were identified by searching for relevant systematic reviews in the Cochrane Collaboration Database of Systematic Reviews and by searches of PubMed and Embase (both last done on March 12, 2010) using the terms “aspirin” or “salicyl*” or “antiplatelet” with the term “randomised controlled trial”. The searches were restricted to studies done in humans, but there was no restriction on language.
Procedure The original investigators were contacted to determine whether anonymised data were available for the number of deaths in which cancer had been regarded as the main underlying cause, the time from randomisation to death, and the primary site of cancer. All cancer deaths had been coded according to the ninth or tenth revision of the International Classification of Diseases (ICD) and the designation of death due to cancer that had been made by the original trialists was used, unless specified otherwise. However, in three trials,17–19 we reviewed the paper case-records of all deaths in patients with known incident cancer to check the designation of cause of death, with the aim of identifying any possible bias resulting from an increase in risk of vascular events due to withdrawal of aspirin treatment after diagnosis of cancer, which might reduce the number of deaths attributed to cancer in the aspirin groups. Three eligible trials, all UK-based, had continued to obtain data for deaths due to cancer after completion of the trials via the national death certification and cancer registration systems—the Thrombosis Prevention Trial (TPT),17 the British Doctors Aspirin Trial (BDAT),19 and the UK transient ischaemic attack (UK-TIA) aspirin trial.18 TPT17 was a 2×2 factorial double-blind randomised trial of aspirin versus placebo and warfarin versus placebo in men aged 45–69 years at increased vascular risk. 135 000 patient records were reviewed in 108 UK primary care practices to exclude ineligible subjects, including those with a recent history of possible peptic ulceration or previous myocardial infarction or stroke. 5085 men with high vascular risk-factor scores were recruited from 1989 to 1992. 2545 were allocated to aspirin (75 mg daily controlled release) and 2540 to placebo. Men were reviewed by their family doctor each year and a research nurse searched their medical records. None were lost to follow-up before the trial end date (October, 1997). All trial participants were flagged in the National Health Service Central Register and notifications of cancer or death were obtained until September, 2009. www.thelancet.com Vol 377 January 1, 2011
Articles
Control Aspirin
6 5 Risk of cancer death (%)
BDAT19 recruited 5139 male doctors (4377 in 1978 and 762 in 1979) who were resident in the UK, born on or after 1900, had no contraindication to aspirin, no regular aspirin use, and no history of peptic ulcer disease, stroke, or myocardial infarction. Randomisation (in a 2:1 ratio) was to daily aspirin (500 mg ordinary, soluble, or effervescent aspirin, as desired, or, if subsequently requested, 300 mg enteric coated aspirin) versus no aspirin or products containing aspirin. Placebo tablets were not used. Treatment was continued until 1984. All participants were asked to complete a questionnaire every 6 months about their health and use of aspirin. Participants were flagged with the National Cancer Registry and the Office of the Registrar General, and all notifications of cancer and death were collected until 2001.12,13 UK-TIA18 recruited 2435 patients with a recent TIA or minor ischaemic stroke from 33 centres in the UK and Ireland between 1979 and 1985. Participants were older than 40 years, with no aspirin intolerance, alcoholism, chronic renal failure, or peptic ulceration. Randomisation was to 1200 mg aspirin daily versus 300 mg daily versus placebo, and treatment was double-blind. Patients were seen by a physician every 4 months until the end of the trial in 1986, and none were lost. Data for deaths and incident cancers notified during and after the trial until 2006 were obtained from national registries, as reported previously.12,13
4
p=0·01
3 2 1 0
1
2
3
4
12 849 10 351
12 371 10 026
11 919 9720
10 964 8881
0
Number at risk Aspirin 13 026 Control 10 509
5 Years to death 9264 7339
6
7
8
9
7385 5933
3384 3438
1676 1671
977 969
Figure 2: Effect of allocation to aspirin versus control on risk of death due to cancer during the trial treatment periods in a pooled analysis of the 23 535 patients in seven trials17–21,23,24 n
0–5 years’ follow-up HR (95% CI)
≥5 years’ follow-up p value
HR (95% CI)
p value
Site of primary cancer* Gastrointestinal Oesophagus
23
0·78 (0·27–2·23)
0·64
0·43 (0·11–1·72)
0·23
Pancreas
45
0·88 (0·44–1·77)
0·73
0·25 (0·07–0·92)
0·04
Colorectal
54
0·78 (0·39–1·56)
0·48
0·41 (0·17–1·00)
0·05
Stomach
36
1·85 (0·81–4·23)
0·14
3·09 (0·64–14·91)
0·16
Other
24
0·67 (0·23–1·99)
0·47
0·20 (0·04–0·91)
0·04
182
0·96 (0·67–1·38)
0·81
0·46 (0·27–0·77)
0·003 0·11
All Non-gastrointestinal
Analysis of deaths during the trial period The effects of allocation to aspirin on risk of death due to cancer and all-cause mortality during each trial were expressed as odds ratios (ORs; with 95% CIs). Pooled estimates were obtained by fixed-effects meta-analysis. After we assessed heterogeneity in effect of aspirin across trials, individual patient data were pooled. The cumulative effect of aspirin on risk of cancer death was estimated with Kaplan-Meier curves and log-rank test (stratified by trial) and by hazard ratios (HRs) obtained from a Cox proportional hazards model stratified by trial. All analyses were by intention to treat on the basis of treatment allocation in the original trials. The following stratified analyses were done: (1) for cancers of the gastrointestinal tract versus other solid cancers versus haematological cancers, given the prior expectation of greatest effects on gastrointestinal cancers (defined as primary site oesophagus, stomach, small intestine, colon, rectum, pancreas, biliary tract, gallbladder, and liver); (2) for the first 5 years after randomisation versus thereafter, given the expectation of greater effects after scheduled treatment and followup for at least 5 years; (3) for common specific solid cancers (oesophagus [with histological type], stomach, pancreas and biliary tract, colorectal, liver, lung [with histological type], prostate, bladder and kidney, and metastases with unknown primary [with histological type]). www.thelancet.com Vol 377 January 1, 2011
Lung
198
0·92 (0·65–1·30)
0·65
0·68 (0·42–1·10)
Prostate
37
0·70 (0·29–1·73)
0·44
0·52 (0·20–1·34)
0·17
Bladder and kidney
31
1·04 (0·44–2·47)
0·93
1·28 (0·36–4·54)
0·70
Other solid
93
0·86 (0·52–1·44)
0·57
1·01 (0·51–1·98)
0·98
359
0·90 (0·69–1·16)
0·41
0·76 (0·54–1·08)
0·12
36
0·56 (0·28–1·15)
0·12
0·56 (0·09–3·38)
0·53
577
0·88 (0·72–1·08)
0·22
0·64 (0·49–0·85)
0·002
Adenocarcinoma
247
0·86 (0·62–1·18)
0·34
0·53 (0·35–0·81)
0·003
Non-adenocarcinoma
224
0·89 (0·65–1·23)
0·48
0·79 (0·50–1·24)
0·30
Unknown
106
0·91 (0·58–1·44)
0·70
0·69 (0·34–1·43)
0·32
Haematological
50
0·82 (0·44–1·54)
0·53
0·34 (0·09–1·28)
0·11
All cancers*
627
0·88 (0·72–1·06)
0·17
0·62 (0·47–0·82)
0·001
All cancers including ETDRS‡
657
0·86 (0·71–1·04)
0·11
0·66 (0·50–0·87)
0·003
All Unknown primary All solid cancers Histological type†
The numbers of cancer deaths included from each trial are those shown on webappendix p 3. n=number of cancer deaths. HR=hazard ratio. ETDRS=Early Treatment Diabetic Retinopathy Study. *Analysis confined to the six trials with site-specific cancer data follow-up.17–19,21, 23,24 †Analysis confined to solid (non-haematological) cancers. ‡Analysis included cancer deaths in ETDRS,20 in which neither primary site nor histological type was known in any case.
Table 1: Pooled analysis of the effect of allocation to aspirin on risk of death due to cancer during the seven trials from which individual patient data were available,17–21,23,24 stratified by type of primary tumour and period of follow-up
Analysis of long-term risk of cancer death For long-term follow-up of the three UK trial cohorts,17–19 all death certificate and cancer registration data relating to events occurring after the trials were also coded according to ICD 9 or 10 (masked to treatment 33
Articles
1–4·9 years Risk of cancer death (%)
25
the effect of aspirin versus placebo was compared with that of warfarin versus placebo.
5–7·4 years
Control Aspirin
Role of the funding source
20 15
p=0·62
The study was unfunded and was independent of any pharmaceutical company or other commercial interest. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
p=0·0003
10 5 0 0
Number at risk Aspirin 1337 Control 820
5
10 Years to death
15
20
0
5
10 Years to death
15
20
1151 733
942 622
732 497
347 199
5426 3383
5028 3135
4528 2814
3871 2390
2274 1134
≥7·5 years
Results
All patients
Risk of cancer death (%)
25 20 15 p=0·003
p<0·0001
10 5 0 0
Number at risk Aspirin 832 Control 861
5
10 Years to death
15
20
0
5
10 Years to death
15
20
788 813
715 731
614 616
360 359
7595 5064
6967 4681
6185 4167
5217 3503
2981 1692
Figure 3: Effect of allocation to aspirin versus control on 20-year risk of death due to any solid cancer stratified by scheduled duration of trial treatment in three trials with long-term follow-up17–19 Continuous variable interaction, p=0·01. See Online for webappendix
34
allocation). Fatal cancers were defined as those in which the cancer had been recorded as the primary underlying cause of death on the death certificate. After checking for heterogeneity between the trials in the absolute risk of cancer death during the trial and posttrial follow-up, which might confound a pooled analysis, and in the effect of allocation to aspirin on cancer death, we pooled individual patient data. Scheduled duration of trial treatment was 5 or 6 years in BDAT and ranged from 1 to 7 years in UK-TIA and from 4 to 9 years in TPT. The effect of scheduled duration of trial treatment on 20-year risk of cancer death was explored within each trial and in the pooled data with an interaction term in a Cox model with duration modelled as a continuous variable, and subsequent analyses were stratified accordingly (≤5 vs 5–7·4 vs ≥7·5 years). All analyses were done on an intention-to-treat basis, with scheduled duration of treatment simply defined as date of randomisation to date of the end of the trial, irrespective of compliance with treatment. The effect of aspirin on 20-year risk of cancer death was also stratified by category of cancers (gastrointestinal tract vs other solid vs haematological), by period of follow-up (0–10 vs 10–20 years), and was determined for deaths due to specific cancers (as defined above) in the 10 502 patients with more than 5 years’ scheduled duration of trial treatment. In TPT,
Trials Of eight eligible randomised trials of aspirin versus control (webappendix p 1) with a mean duration of scheduled treatment before the end of the trial of 4 years or more, two had been done in primary prevention of vascular disease,17,19 one in secondary prevention after recent vascular events,18 and five in groups with increased vascular risk without previous vascular events (type 1 or 2 diabetes;20 type 2 diabetes;21 stable angina;22 diabetes with asymptomatic peripheral arterial disease;23 low ankle brachial index24). Data for the number of deaths due to cancer were available from all eight trials. Individual patient data were available from seven trials, but all records of one trial had been destroyed (Juul-Moller S, University Hospital, Malmo, Sweden, personal communication)..22
In-trial deaths During the eight trials there were 674 deaths due to cancer among 25 570 patients. The proportion of all deaths that were due to cancer varied (p<0·0001), ranging from 4·2% in the young diabetic population in the Early Treatment Diabetic Retinopathy Study (ETDRS),20 12·8% and 15·4% in the patients with symptomatic vascular disease in the 1980s in UK-TIA18 and the Swedish Angina Pectoris Aspirin Trial,22 to 28·7% (Prevention of Progression of Arterial Disease and Diabetes [POPADAD] study),23 29·2% (BDAT),19 45·4% (TPT),17 46·4% (Aspirin for Asymptomatic Atherosclerosis [AAA] trial),24 and 47·9% (Japanese Primary Prevention of Atherosclerosis With Aspirin for Diabetes [JPAD] study)21 in lower vascular risk or more recently recruited cohorts. However, there was very little heterogeneity between trials (figure 1; phet=0·84) in the effect of allocation to aspirin on risk of death due to cancer (OR 0·79, 95% CI 0·68–0·92, p=0·003, overall; 0·81, 0·68–0·97, p=0·03, in trials of aspirin 75–100 mg daily). Reclassification of cause of death in a small number of cases in UK-TIA and TPT had little effect on the pooled estimate (686 deaths; OR 0·80, 0·69–0·93, p=0·004, webappendix p 3). The reduction in cancer deaths on aspirin during the trials resulted in lowered in-trial all-cause mortality (10·2% vs 11·1%, OR 0·92, 0·85–1·00, p=0·047, webappendix p 4), even though other deaths were not reduced (0·98, 0·89–1·07, p=0·63). In our analysis of individual patient data for time to death, which were available for seven trials (657 cancer www.thelancet.com Vol 377 January 1, 2011
Articles
n
0–10 years’ follow-up
10–20 years’ follow-up
HR (95% CI)
p value
HR (95% CI)
0–20 years’ follow-up p value
HR (95% CI)
p value
Solid cancers Gastrointestinal Oesophagus
62
0·53 (0·24–1·18)
0·12
0·36 (0·18–0·71)
0·003
0·42 (0·25–0·71)
0·001
Pancreas
77
0·82 (0·41–1·67)
0·59
0·79 (0·44–1·42)
0·43
0·81 (0·51–1·26)
0·34
Colorectal
179
0·79 (0·49–1·26)
0·32
0·51 (0·35–0·74)
0·0005
0·60 (0·45–0·81)
0·0007
Stomach
71
1·36 (0·64–2·90)
0·43
0·42 (0·23–0·79)
0·007
0·69 (0·43–1·10)
0·11
Other
18
0·68 (0·14–3·36)
0·64
1·97 (0·53–7·27)
0·31
1·33 (0·50–3·54)
0·57
409
0·80 (0·59–1·08)
0·14
0·56 (0·44–0·72)
<0·0001
0·65 (0·53–0·78)
<0·0001
Lung
326
0·68 (0·50–0·92)
0·01
0·75 (0·55–1·02)
0·07
0·71 (0·58–0·89)
0·002
Prostate
210
0·83 (0·47–1·46)
0·52
0·80 (0·58–1·09)
0·15
0·81 (0·61–1·06)
0·12
94
0·75 (0·41–1·37)
0·35
0·90 (0·52–1·57)
0·72
0·83 (0·55–1·25)
0·37
Other solid
128
0·68 (0·39–1·17)
0·16
1·28 (0·80–2·05)
0·31
0·98 (0·69–1·39)
0·91
All
757
0·71 (0·56–0·88)
0·002
0·85 (0·71–1·03)
0·10
0·79 (0·69–0·91)
0·001
89
1·19 (0·58–2·42)
0·63
0·95 (0·56–1·61)
0·84
1·03 (0·67–1·57)
0·90
1251
0·76 (0·63–0·90)
0·002
0·75 (0·65–0·87)
0·0001
0·75 (0·67–0·84)
<0·0001
Adenocarcinoma
648
0·70 (0·54–0·91)
0·008
0·64 (0·53–0·77)
<0·0001
0·66 (0·56–0·77)
<0·0001
Non-adenocarcinoma
302
1·04 (0·72–1·52)
0·83
0·74 (0·55–0·98)
0·04
0·87 (0·70–1·08)
0·21
Unknown
331
0·66 (0·49–0·90)
0·01
1·12 (0·83–1·52)
0·46
0·84 (0·67–1·05)
0·13
126
1·31 (0·69–2·50)
0·41
1·00 (0·65–1·54)
0·99
1·09 (0·76–1·56)
0·65
1378
0·79 (0·66–0·93)
0·005
0·77 (0·67–0·89)
0·0002
0·78 (0·70–0·87)
<0·0001
All Non-gastrointestinal
Bladder and kidney
Unknown primary All solid cancers Histological type*
Haematological cancers All cancers
Analysis limited to patients with scheduled duration of trial treatment or 5 years or longer. n=number of cancer deaths. HR=hazard ratio. *Analysis confined to solid (non-haematological) cancers.
Table 2: Pooled analysis of the effect of allocation to aspirin on the 20-year risk of death due to cancer during and after the trial treatment periods in the 10 502 patients with scheduled treatment duration of 5 years or longer in the three trials with long-term follow-up,17–19 stratified by type of primary tumour and period of follow-up
deaths in 23 535 patients based on data in webappendix p 3),17–21,23,24 aspirin reduced deaths due to cancer (HR 0·82, 0·70–0·95, p=0·01, figure 2), due mainly to fewer deaths after five years (0·66, 0·50–0·87, p=0·003; table 1), but had no effect on other deaths (n=1871; 1·03, 0·94–1·13, p=0·54). Data were available for the site of the primary cancer in six trials (627 cancer deaths in 19 824 patients).17–19,21, 23,24 Aspirin reduced deaths due to gastrointestinal cancers and deaths due to nongastrointestinal solid cancers (table 1), with most benefit again seen after 5 years of scheduled trial treatment (gastrointestinal cancers, HR 0·46, 0·27–0·77, p=0·003; non-gastrointestinal solid cancers, 0·76, 0·54–1·08, p=0·12), and including significant reductions in colorectal and pancreatic cancer deaths (table 1).
Post-trial follow-up Follow-up was obtained to 20 years in TPT, BDAT, and UK-TIA (1634 cancer deaths in 12 659 patients, webappendix p 2).17–19 Aspirin reduced the 20-year risk of death due to all solid cancers (HR 0·80, 0·72–0·88, p<0·0001) and gastrointestinal cancer (0·65, 0·54–0·78, p<0·0001), but not haematological cancer (1·03, 0·74–1·43, p=0·87). However, in both TPT and UK-TIA, in which duration of trial treatment varied, the effect on 20-year risk www.thelancet.com Vol 377 January 1, 2011
of solid cancer increased with duration of scheduled treatment (interaction: p=0·016 in TPT, p=0·08 in UKTIA). This interaction remained (p=0·01) in the pooled analysis with BDAT (figure 3), with no reduction in solid cancers in patients with scheduled treatment for 1–4·9 years (HR 1·06, 0·82–1·39, p=0·62), significant benefit with 5–7·4 years (0·79, 0·70–0·90, p=0·0003), and greatest benefit with 7·5 years or longer (solid cancers, 0·69, 0·54–0·88, p=0·003; gastrointestinal cancers, 0·41, 0·26–0·66, p=0·0001). Results given below therefore refer to the 10 502 patients (1378 cancer deaths) with scheduled duration of 5 years or longer unless otherwise specified. In patients with scheduled duration of trial treatment of 5 years or more (table 2), allocation to aspirin reduced the 20-year risk of death due to both gastrointestinal (HR 0·65, 0·53–0·78, p<0·0001) and non-gastrointestinal solid (0·79, 0·69–0·91, p=0·001) cancers. There was no significant heterogeneity in effect of aspirin across the different gastrointestinal cancers (p=0·26), but effects were greatest for oesophageal and colorectal cancers (table 2, figure 4). As expected, there was a latent period before any effect was observed, with reductions in risk of death due to oesophageal and pancreatic cancer evident from 5 years onwards and reductions in deaths due to stomach and colorectal cancer not evident until about 35
Articles
Oesophagus Risk of cancer death (%)
2
Stomach 2
Control Aspirin
1·5
1·5 p=0·001
1 0·5
0·5
0
0 Pancreas
Colorectal 4
Risk of cancer death (%)
2
3·5 3 2·5 2 1·5
1·5 p=0·34
1 0·5
Lung
Prostate 4
5 Risk of cancer death (%)
p=0·0007
1 0·5 0
0
4
3
3 2
p=0·002
p=0·12
2 1
1
0
0 Kidney/bladder
Risk of cancer death (%)
p=0·11
1
Haematological
2
2
1·5
1·5
1
1
p=0·37
p=0·65
0·5
0·5
0
0 0 Number at risk Aspirin 6258 Control 4244
5
5816 3948
10 Years to death 5243 3545
15
4485 3006
20
0
5
10 Years to death
15
20
2634 1493
6258 4244
5816 3948
5243 3545
4485 3006
2634 1493
Figure 4: Effect of allocation to aspirin versus control on the 20-year risk of death due to the most common fatal cancers in the 10 502 patients with scheduled treatment duration of 5 years or longer in the three trials with long-term follow-up17–19 The eight most common cancer types are shown.
10 years (tables 1 and 2). The effect of aspirin on death due to pancreatic cancer was only significant at 20-year follow-up in patients with scheduled duration of trial treatment longer than 7·5 years (HR 0·28, 0·08–1·00, p=0·04). Overall, the absolute risk of death due to gastrointestinal cancer was reduced by 2·18% (1·14–3·22) at 20-year follow-up. The effect of aspirin on 20-year risk of death due to non-gastrointestinal solid cancer (table 2) was attributable mainly to a reduction in deaths due to lung cancer and a non-significant late reduction in deaths 36
due to prostate cancer (figure 4), particularly in patients with scheduled duration of trial treatment of 7·5 years or longer (HR 0·52, 0·24–1·10, p=0·08). Aspirin also reduced deaths due to primary brain tumours during the first 10 years of follow-up (5/6258 in the aspirin groups vs 12/4244 in the control groups; HR 0·31, 0·11–0·89, p=0·03) and mean time from randomisation to death from brain tumour remained longer in the aspirin group than in the control group at 20 years (p=0·018, webappendix p 5). Overall, the absolute risk of death due to non-gastrointestinal solid cancer was reduced by 1·88% (0·57–3·19) at 20-year follow-up. The effect of aspirin on risk of death due to gastrointestinal cancer did not differ by age at randomisation (figure 5; interaction: relative effect, p=0·44; absolute effect, p=0·96), but the effect on death due to non-gastrointestinal solid cancers increased with age (relative effect, p=0·056; absolute effect, p=0·001). For the 20-year risk of death due to any cancer, the reductions in absolute risk in the aspirin groups were 1·41% (–0·74 to 3·56) at age less than 55 years, 4·53% (2·06–6·99) at age 55–64 years, 7·08% (2·42–11·74) at age 65 years or older, and 3·49% (1·85–5·13) at all ages combined. Relative and absolute effects were similar in smokers and non-smokers (data not shown). Where data for histological type were available, aspirin had no effect on the 20-year risk of death due to smallcell (HR 0·85, 0·52–1·39, p=0·56) or squamouscell (1·26, 0·73–2·18, p=0·49) lung cancers, but reduced the risk of death due to adenocarcinoma of lung (0·55, 0·33–0·94, p=0·04). The reduction in deaths due to oesophageal cancer was also confined to adenocarcinoma (HR 0·36, 0·21–0·63, p=0·0001), although the number of squamous-cell cancers was small (9/6258 in the aspirin groups vs 2/4244 in the control groups). Indeed, across all cancers (tables 1 and 2, webappendix p 6), aspirin only reduced deaths due to either histologically proven adenocarcinomas or primary cancers in which adenocarcinoma predominates (stomach, small bowel, pancreas, bile duct, colon, rectum, breast, uterus, ovary, and prostate). This effect on adenocarcinoma was consistent across the three trials (webappendix p 7) and for different doses of aspirin (webappendix p 6), but was not seen in the comparison of warfarin versus placebo in TPT (figure 6). In patients with scheduled duration of trial treatment of 5 years or longer, all-cause mortality was reduced at 15 years’ follow-up (HR 0·92, 0·86–0·99, p=0·03), due entirely to fewer cancer deaths, but this effect was no longer seen at 20 years (0·96, 0·90–1·02, p=0·37). However, the effect on post-trial deaths was diluted by a transient increase in risk of vascular death in the aspirin groups during the first year after completion of the trials (75 observed vs 46 expected, OR 1·69, 1·08–2·62, p=0·02), presumably due to withdrawal of trial aspirin. www.thelancet.com Vol 377 January 1, 2011
Articles
Discussion
www.thelancet.com Vol 377 January 1, 2011
Gastrointestinal cancers <55 years Control Aspirin
Risk of cancer death (%)
12 10
15 p=0·001
6
p=0·58
10
4 5 2 0
5
Number at risk Aspirin 1796 Control 1352 12 Risk of cancer death (%)
Non-gastrointestinal cancers <55 years
20
8
0
10 Years to death
1743 1314
1681 1259
15
1566 1179
20
944 596
0
0
5
10 Years to death
15
20
1796 1352
1743 1314
1681 1259
1566 1179
944 596
20
55–64 years
10
55–64 years
15
8 p=0·003
6
p=0·001
10
4 5 2 0
0
5
Number at risk Aspirin 2850 Control 1924 12 Risk of cancer death (%)
We showed previously that treatment with aspirin for longer than 5 years reduced the long-term risk of colorectal cancer.12,13 In analyses of nearly 2000 cancer deaths, we now show that aspirin also reduces deaths due to several other common cancers (panel). First, we showed by meta-analysis that aspirin reduced the risk of death due to cancer by about 20% during the trials. Second, by analysis of individual patient data we showed that this benefit was due mainly to a delayed reduction of about 30–40% in deaths after 5 years of treatment. Third, by long-term follow-up of three large trials we showed that the reduction in deaths due to solid cancers was maintained for 20 years, only becoming apparent for some cancers after completion of the trials. Fourth, these effects were consistent across trials, despite the very different populations, suggesting that the findings will be generalisable. Fifth, as shown for colorectal cancer,12,13 the effect of aspirin increased with duration of scheduled trial treatment. Sixth, the effect was limited to certain cancers, most particularly adenocarcinomas. Seventh, the effect did not appear to increase at aspirin doses greater than 75 mg daily. Eighth, the absolute reduction in death due to cancer increased with age, within the range of patients entered into the trials. Finally, the effect of aspirin on risk of fatal cancers resulted in a small reduction in all-cause mortality. Our analyses were conservative in several respects. First, although all but one of the trials we studied were double-blind, there were high rates of drop-outs from randomised treatment. In the trials in which we obtained long-term follow-up, about 40% of patients in the aspirin groups had stopped treatment by the end of the trial periods.17–19 Nevertheless, to reduce bias we restricted our analyses to intention to treat. Second, since the effect of aspirin increased with scheduled duration of trial treatment, but the trials were of finite length, it is likely that we underestimated the benefit of long-term treatment on deaths due to cancer. The difference in aspirin use between the treatment groups was already limited by the end of BDAT, many patients in the TPT control group went on to aspirin after the trial,25 and posttrial aspirin use would not have differed much between the treatment groups in UK-TIA because trial treatment allocation was never revealed. The trials that we studied were randomised, but could our findings have been due to bias? First, the trials were not designed to study cancer. However, cancer deaths were recorded during the trials, and long-term follow-up via UK cancer registration achieves high rates of ascertainment and accuracy,26–28 as we found previously for colorectal cancer.12,13 Attribution of cause of death during the trials was masked to treatment allocation, as was coding of the cause of post-trial deaths. Attribution was usually based on death certification, supported by any previous cancer registration, which has been shown previously to agree well with expert committee review.29–31 Second, lack of
10 Years to death
2680 1808
2468 1617
15
20
2152 1378
1333 721
0
0
5
10 Years to death
15
20
2850 1924
2680 1808
2468 1617
2152 1378
1333 721
20
≥65 years
10
≥65 years
15
8 6
10
p=0·03
p=0·004
4 5
2 0
0 0
Number at risk Aspirin 1612 Control 968
5
10 Years to death
15
20
0
5
10 Years to death
15
20
1393 826
1094 669
767 449
357 176
1612 968
1393 826
1094 669
767 449
357 176
Figure 5: Effect of allocation to aspirin versus control on risk of death due to cancers of the gastrointestinal tract and other solid cancers in the 10 502 patients with scheduled treatment duration of 5 years or longer, stratified by age at randomisation in three trials with long-term follow-up17–19
knowledge among the trial investigators that data might later be used to study the effect of aspirin on risk of cancer will have limited any potential investigator bias. Third, investigation of side-effects of aspirin, such as anaemia and bleeding, might have resulted in earlier diagnosis of cancers and hence a reduction in later deaths. However, analysis of time to incidence of colorectal cancer showed no evidence of earlier diagnosis,13 and the very low cure rates of cancers such as oesophageal cancer would limit any bias due to earlier diagnosis. The only evidence of a possible effect of increased investigation in the aspirin groups was a transient increase in risk of deaths attributed to stomach cancer during the trials, and a transient reduction in deaths attributed to cancers with unknown primary site (table 1). Moreover, the complete lack of any effect of warfarin on cancer deaths in TPT suggests that 37
Articles
Aspirin vs placebo Risk of cancer death (%)
2·5 2
2 1·5
p=0·008
1
1
0·5
0·5
Risk of cancer death (%)
5
10 Years to death
15
20
2402 2404
2197 2189
1913 1883
410 412
0
Number at risk Aspirin 2545 Control 2540
0
8
6
6 p=0·0002
4
p=0·99
0
5
10 Years to death
15
20
2409 2397
2205 2181
1896 1900
407 415
15
20
1896 1900
407 415
15
20
2179 2158
1845 1849
397 400
5
10 Years to death
15
20
2409 2397
2205 2181
1896 1900
407 415
Gastrointestinal adenocarcinoma
p=0·91
4 2
2
0
5
Number at risk Aspirin 2545 Control 2540
10 Years to death
2402 2404
2197 2189
15
20
1913 1883
410 412
0
8
6
6 p=0·09
4
0
5
Number at risk Warfarin 2545 Control 2540
8 Prostate adenocarcinoma
10 Years to death
2409 2397
2205 2181
Prostate adenocarcinoma
p=0·69
4 2
2 0
Lung adenocarcinoma
Number at risk Warfarin 2545 Control 2540
8 Gastrointestinal adenocarcinoma
0
Risk of cancer (%)
2·5
1·5
0
0
Number at risk Aspirin 2545 Control 2540
14 Risk of cancer death (%)
Warfarin vs placebo Control Aspirin
Lung adenocarcinoma
5
10 Years to notification
15
20
2388 2387
2173 2164
1859 1835
400 397
0
14
All fatal adenocarcinoma
12
10
10
10 Years to notification
2394 2381
All fatal adenocarcinoma
8
p<0·0001
6
5
Number at risk Warfarin 2545 Control 2540
12 8
0
p=0·84
6 4
4 2
2
0
0
0
5
10
15
20
1913 1883
410 412
0
Years to death Number at risk Aspirin 2545 Control 2540
2402 2404
2197 2189
Number at risk Warfarin 2545 Control 2540
Figure 6: Comparison of effect of allocation to aspirin or warfarin versus placebo on risk of death due to adenocarcinoma during long-term follow-up of the Thrombosis Prevention Trial17 Analysis includes all patients, irrespective of scheduled duration of trial treatment. Analysis of prostate cancer also includes non-fatal cancers because of the small numbers of fatal cancers in the single trial.
38
www.thelancet.com Vol 377 January 1, 2011
Articles
Panel: Research in context Findings Using individual patient data from all randomised trials of daily aspirin versus no aspirin with mean duration of scheduled trial treatment longer than 4 years, we showed that aspirin reduced risk of death due to cancer by about 20% in the trials, due mainly to a 34% reduction in cancer deaths after 5 years. By long-term post-trial follow-up of patients in three of these trials, we showed that the 20-year risk of cancer death remained about 20% lower in the aspirin groups, and that benefit increased with scheduled duration of treatment in the original trial. The latent period before an effect on deaths was about 5 years for oesophageal, pancreatic, brain, and lung cancer, but was more delayed for stomach, colorectal, and prostate cancer. For lung and oesophageal cancer, benefit was confined to adenocarcinomas. Interpretation These findings provide the first proof in man that aspirin reduces deaths due to several common cancers. Benefit was consistent across the different trial populations, suggesting that the findings are likely to be generalisable.
early diagnosis due to bleeding is unlikely to have been a major source of bias. Fourth, many patients would probably have stopped taking their trial drug if they developed cancer. Stopping aspirin could in some cases have triggered a fatal vascular event that might have resulted in an underlying cancer not being diagnosed or at least not being listed on the death certificate as the underlying cause of death. However, there was no evidence of any excess of non-fatal vascular events during the year before death due to cancer in the aspirin groups (data not shown), and most of the reduction in cancer deaths occurred after the trials. Finally, we had long-term posttrial follow-up from only three of the trials, but this factor was determined simply by the country in which the trials were done and the era. Moreover, the effect of aspirin on in-trial cancer deaths was no greater in these trials than in the others (figure 1). These results therefore provide the first reliable evidence that aspirin prevents non-colorectal cancer in humans, which is consistent with previous predictions of effects on cancers of the oesophagus, stomach, pancreas, lung, prostate,7 and possibly brain.32,33 However, more work is required. Effects of aspirin on incidence of cancer must be determined, both for cancers that are less commonly fatal and to determine whether the latent period before an effect is shorter than for death. More trial data are required for the effect of aspirin on risk of breast and other cancers of women. Follow-up beyond 20 years is necessary to identify any late rebound in cancer deaths. The estimate of effect of aspirin on death due to cancer in the first 5 years of the trials does not exclude a clinically important short-term benefit for cancers that would probably have already been present at www.thelancet.com Vol 377 January 1, 2011
randomisation, and so pooled analysis of trials with shorter follow-up is also required. To address each of these issues, the Non-Vascular Outcomes on Aspirin Collaboration is collating all available data from trials of aspirin (Rothwell, personal communication) and will report further results in 2011. Our study does have several potential limitations. First, we included only trials of daily aspirin. Alternate-day aspirin was used in other trials in prevention of vascular events because aspirin irreversibly inhibits COX-1 in platelets, but this effect would not be irreversible in other tissues, and observational studies have highlighted the importance of daily aspirin in associations with reduced incidence of cancer.5–8,12 10-year follow-up of the Women’s Health Study, a randomised trial of aspirin 100 mg on alternate days versus control, did show a possible reduction in incidence of lung cancer, but there was no reduction in other cancers or in overall cancer incidence.16 Second, although there was no evidence of any sex-related difference in the effect of aspirin on deaths due to cancer during the trials (data not shown), or in previous observational studies, we had too few women in the trials with long-term follow-up to allow us to determine the effects of aspirin on breast or gynaecological cancers. Third, analysis of effect of aspirin on adenocarcinoma overall was data-dependent, although analysis of histological subtype of lung and oesophageal cancers was prespecified. Fourth, we were unable to determine the effect of long-term (eg, 20–30 years) continued aspirin use on cancer death or all-cause mortality because of the finite duration of the trials. The transient increase in risk of vascular deaths in the aspirin groups after the trials, consistent with studies of aspirin withdrawal,34,35 also diluted the effect that we did observe on long-term mortality. Finally, the benefits of aspirin may be less in populations with a high dietary intake of salicylates. Our results have implications for clinical practice. Since other antiplatelet drugs do not reduce risk of cancer death in randomised trials (Rothwell, unpublished data), patients with an indication for long-term antiplatelet treatment are likely, on average, to benefit most from aspirin. Although the reduction in risk of ischaemic vascular events on aspirin in healthy individuals is partly offset by a small increase in risk of non-fatal bleeding complications,36–38 the balance of risk and benefit will now be altered by the reduction in cancer deaths after 5 years’ treatment. Our analyses show that taking aspirin daily for 5–10 years would reduce all-cause mortality (including any fatal bleeds) during that time by about 10% (relative risk reduction). Subsequently, there would be further delayed reductions in risk of cancer death, but no continuing excess risk of bleeding. In terms of cost-effectiveness,39 such benefit would exceed that of established initiatives such as screening for breast or prostate cancer, potentially justifying added costs to reduce bleeding complications, such as co-prescription of a proton-pump inhibitor,40,41 39
Articles
treatment to eradicate Helicobacter pylori infection,42 and further development of potentially more effective derivatives of aspirin.43 Moreover, since the effect of aspirin on risk of cancer death increased with scheduled duration of trial treatment, the roughly 30% reduction in 20-year risk of cancer deaths observed in patients with scheduled trial treatment of 7·5–10 years may well underestimate the benefit that would result from longerterm treatment (eg, from age 50–75 years). Indeed, a late rebound in cancer deaths in the aspirin group at 10–20 years’ follow-up is clearly present for some cancers (figures 4 and 6, webappendix p 5). Finally, our results have implications for understanding of carcinogenesis, particularly for adenocarcinoma, and they demonstrate the potential for drug intervention in the prevention of cancer. Although the effect of aspirin may be mediated in part by inhibition of COX-2, more research is required, other pro-apoptotic effects early in the development of tumours perhaps also being important.43,44 Contributors PMR conceived and coordinated the project, obtained long-term follow-up of the UK-TIA trial, collated all data, planned and performed all analyses, and wrote the report. FGRF was principal investigator on the AAA trial. JFFB was principal investigator on the POPADAD trial. HO was principal investigator on the JPAD trial. CPW was principal investigator on the UK-TIA aspirin trial. TWM was principal investigator on the TPT and obtained long-term follow-up data. All authors commented on drafts of the report. Conflicts of interest This study was completely independent of any pharmaceutical company or other commercial interest. However, PMR has received honoraria for talks, advisory boards, and clinical trial committees from several pharmaceutical companies with an interest in antiplatelet agents, including AstraZeneca, Bayer, Boehringer Ingelheim, Sanofi-Aventis/Bristol-Myers Squibb, and Servier. FGRF has had research support, honoraria, and travel expenses from Bayer and Sanofi-Aventis/Bristol-Myers Squibb. JFFB has received payment for board membership from Roche Pharmaceuticals and Sanofi-Aventis. HO has received speakers’ fees from Astellas, AstraZeneca, Banyu, Bayer, Boehringer Ingelheim, Chugai, Daiichi Sankyo, Eisai, Guidant Japan, Japan Lifeline, Kowa, Kyowa Hakko Kirin, Novartis, Otsuka, Pfizer, Sanofi-Aventis, Schering-Plough, and Takeda. CPW has received speakers’ fees from Bayer for a talk about aspirin. TWM has received an honorarium and travel expenses from Bayer. Collaborators Emily Chew (ETDRS; National Eye Institute, Bethesda, MD, USA); Takeshi Morimoto (JPAD trial; Kyoto University Graduate School of Medicine, Kyoto, Japan); Richard Peto (BDAT; University of Oxford, Oxford, UK). Acknowledgments The study received no specific funding. Funding of the original trials was as reported previously.17–19,22–26 The cost of coordination of the project and collation and analysis of data was met by unrestricted research funds from the Stroke Prevention Research Unit, Oxford. PMR is in receipt of an NIHR Senior Investigator Award. We thank Jill Boreham for help with access to data from the BDAT; Christine Knottenbelt and Marilyn Goulding for their help in accessing long-term follow-up data from TPT and Michelle Wilson for help in coding these data; Izzy Butcher for help with data from AAA; and Robert Lee for help with data from POPADAD. We thank Ziyah Mehta for help with analysis and production of graphs. References 1 Jemal A, Center MM, DeSantis C, Ward EM. Global patterns of cancer incidence and mortality rates and trends. Cancer Epidemiol Biomarkers Prev 2010; 19: 1893–907.
40
2
3 4
5
6 7
8 9
10
11
12
13
14
15 16
17
18
19 20
21
22
23
Ferlay J, Autier P, Boniol M, Heanue M, Colombet M, Boyle P. Estimates of the cancer incidence and mortality in Europe in 2006. Ann Oncol 2007; 18: 581–92. The Lancet. Preventable cancer in the USA. Lancet 2010; 375: 1665. Dannenberg AJ, DuBois RD, eds. Progress in experimental tumour research vol. 37: COX-2—a new target for cancer prevention and treatment. Basel, Switzerland: Karger, 2003. Thun MJ, Henley SJ, Patrono C. Nonsteroidal anti-inflammatory drugs as anticancer agents: mechanistic, pharmacologic, and clinical issues. J Natl Cancer Inst 2002; 94: 252–66. Elwood PC, Gallagher AM, Duthie GG, Mur LAJ, Morgan G. Aspirin, salicylates, and cancer. Lancet 2009; 373: 1301–09. Cuzick J, Otto F, Baron JA, et al. Aspirin and non-steroidal anti-inflammatory drugs for cancer prevention: an international consensus statement. Lancet Oncol 2009; 10: 501–07. Bosetti C, Gallus S, La Vecchia C. Aspirin and cancer risk: a summary review to 2007. Recent Results Cancer Res 2009; 181: 231–51. Farquhar C, Marjoribanks J, Lethaby A, Suckling JA, Lamberts Q. Long term hormone therapy for perimenopausal and postmenopausal women. Cochrane Database Syst Rev 2005; 3: CD004143. Lawlor DA, Smith GD, Bruckdorfer KR, Kundu D, Ebrahim S. Those confounded vitamins: what can we learn from the differences between observational versus randomised trial evidence? Lancet 2004; 363: 1724–27. Serebruany V, Floyd J, Chew D. Excess of solid cancers after prasugrel: the Food and Drug Administration outlook. Am J Ther 2010; published online July 10. DOI:10.1097/MJT.0b013e3181e9b675. Flossmann E, Rothwell PM, on behalf of the British Doctors Aspirin Trial and the UK-TIA Aspirin Trial. Effect of aspirin on long-term risk of colorectal cancer: consistent evidence from randomised and observational studies. Lancet 2007; 369: 1603–13. Rothwell PM, Wilson M, Elwin C-E, et al. Long-term effect of aspirin on colorectal cancer incidence and mortality: 20-year follow-up of five randomised trials. Lancet 2010; 376: 1741–50. Cole BF, Logan RF, Halabi S, et al. Aspirin for chemoprevention of colorectal adenomas: meta-analysis of the randomised trials. J Natl Cancer Inst 2009; 101: 256–66. Chan AT, Ogino S, Fuchs CS. Aspirin and risk of colorectal cancer in relation to expression of COX-2. N Engl J Med 2007; 356: 2131–42. Cook NR, Lee IM, Gaziano JM, et al. Low-dose aspirin in the primary prevention of cancer: the Women’s Health Study: a randomized controlled trial. JAMA 2005; 294: 47–55. The Medical Research Council’s General Practice Research Framework. Thrombosis prevention trial: randomised trial of low-intensity oral anticoagulation with warfarin and low-dose aspirin in the primary prevention of ischaemic heart disease in men at increased risk. Lancet 1998; 351: 233–41. Farrell B, Godwin J, Richards S, Warlow C. The United Kingdom transient ischaemic attack (UK-TIA) aspirin trial: final results. J Neurol Neurosurg Psychiatry 1991; 54: 1044–54. Peto R, Gray R, Collins R, et al. Randomised trial of prophylactic daily aspirin in British male doctors. BMJ 1988; 296: 313–16. ETDRS Investigators. Aspirin effects on mortality and morbidity in patients with diabetes mellitus. Early Treatment Diabetic Retinopathy Study report 14. JAMA 1992; 268: 1292–300. Ogawa H, Nakayama M, Morimoto T, et al. Japanese Primary Prevention of Atherosclerosis With Aspirin for Diabetes (JPAD) Trial Investigators. Low-dose aspirin for primary prevention of atherosclerotic events in patients with type 2 diabetes: a randomized controlled trial. JAMA 2008; 300: 2134–41. Juul-Möller S, Edvardsson N, Jahnmatz B, Rosén A, Sørensen S, Ömblus R, for the Swedish Angina Pectoris Aspirin Trial (SAPAT) Group. Double-blind trial of aspirin in primary prevention of myocardial infarction in patients with stable chronic angina pectoris. Lancet 1992; 340: 1421–25. Belch J, MacCuish A, Campbell I, et al, Prevention of Progression of Arterial Disease and Diabetes Study Group; Diabetes Registry Group; Royal College of Physicians Edinburgh. The prevention of progression of arterial disease and diabetes (POPADAD) trial: factorial randomised placebo controlled trial of aspirin and antioxidants in patients with diabetes and asymptomatic peripheral arterial disease. BMJ 2008; 337: a1840.
www.thelancet.com Vol 377 January 1, 2011
Articles
24
25
26 27
28
29
30
31
32
33
34
Fowkes FG, Price JF, Stewart MC, et al. Aspirin for Asymptomatic Atherosclerosis Trialists. Aspirin for prevention of cardiovascular events in a general population screened for a low ankle brachial index: a randomized controlled trial. JAMA 2010; 303: 841–48. Fasey N, Brennan PJ, Meade TW, MRC General Practice Research Framework. Medical Research Council. Thrombosis prevention trial: follow-up study of practical implications. Br J Gen Pract 2002; 52: 208–09. Brewster DH, Stockton D, Harvey J, Mackay M. Reliability of cancer registration data in Scotland, 1997. Eur J Cancer 2002; 38: 414–17. Hawkins MM, Swerdlow AJ. Completeness of cancer and death follow-up obtained through the National health Service Central Register for England and Wales. Br J Cancer 1992; 66: 408–13. Pollock AM, Vickers N. Reliability of data of the Thames cancer registry on 673 cases of colorectal cancer: effect of the registration process. Qual Health Care 1995; 4: 184–89. Robinson MH, Rodrigues VC, Hardcastle JD, Chamberlain JO, Mangham CM, Moss SM. Faecal occult blood screening for colorectal cancer at Nottingham: details of the verification process. J Med Screen 2000; 7: 97–98. Doria-Rose VP, Marcus PM. Death certificates provide an adequate source of cause of death information when evaluating lung cancer mortality: an example from the Mayo Lung Project. Lung Cancer 2009; 63: 295–300. Ederer F, Geisser M, Mongin S, Church T, Mandel J. Colorectal cancer deaths as determined by expert committee and from death certificate. The Minnesota Study. J Clin Epidemiol 1999; 52: 447–52. Sivak-Sears NR, Schwartzbaum JA, Miike R, Moghadassi M, Wrensch M. Case-control study of use of nonsteroidal antiinflammatory drugs and glioblastoma multiforme. Am J Epidemiol 2004; 159: 1131–39. Amin R, Kamitani H, Sultana H, et al. Aspirin and indomethacin exhibit antiproliferative effects and induce apoptosis in T98G human glioblastoma cells. Neurol Res 2003; 25: 370–76. Sibon I, Orgogozo JM. Antiplatelet drug discontinuation is a risk factor for ischemic stroke. Neurology 2004; 62: 1187–89.
www.thelancet.com Vol 377 January 1, 2011
35
36
37
38 39
40
41
42
43
44
Maulaz AB, Bezerra DC, Michel P, Bogousslavsky J. Effect of discontinuing aspirin therapy on the risk of brain ischemic stroke. Arch Neurol 2005; 62: 1217–20. Antithrombotic Trialists’ (ATT) Collaboration. Aspirin in the primary and secondary prevention of vascular disease: collaborative meta-analysis of individual participant data from randomised trials. Lancet 2009; 373: 1849–60. Greving JP, Buskens E, Koffijberg H, Algra A. Cost-effectiveness of aspirin treatment in the primary prevention of cardiovascular disease events in subgroups based on age, gender, and varying cardiovascular risk. Circulation 2008; 117: 2875–83. Morgan G. Aspirin for the primary prevention of vascular events? Public Health 2009; 123: 787–88. Annemans L, Wittrup-Jensen K, Bueno H. A review of international pharmacoeconomic models assessing the use of aspirin in primary prevention. J Med Econ 2010; 13: 418–27. Lai KC, Lam SK, Chu KM, et al. Lansoprazole for prevention of recurrences of ulcer complications from long-term low-dose aspirin use. N Engl J Med 2002; 346: 2033–38. Chan FK, Ching JY, Hung LC, et al. Clopidogrel versus aspirin and esomeprazole to prevent recurrent ulcer bleeding. N Engl J Med 2005; 352: 238–44. Lanas A, Fuentes J, Benito R, Serrano P, Bajador E, Sainz R. Helicobacter pylori increases the risk of upper gastrointestinal bleeding in patients taking low-dose aspirin. Aliment Pharmacol Ther 2002; 16: 779–86. McIlhatton MA, Tyler J, Burkholder S, Ruschoff J, Rigas B, Kopelovich L, Fishel R. Nitric oxide-donating aspirin derivatives suppress microsatellite instability in mismatch repair-deficient and hereditary nonpolyposis colorectal cancer cells. Cancer Res 2007; 67: 10966–75. Ruschoff J, Wallinger S, Dietmaier W, et al. Aspirin suppresses mutator phenotype associated with hereditary nonpolyposis colorectal cancer by genetic selection. Proc Natl Acad Sci USA 1998; 95: 11301–06.
41
Articles
Rituximab maintenance for 2 years in patients with high tumour burden follicular lymphoma responding to rituximab plus chemotherapy (PRIMA): a phase 3, randomised controlled trial Gilles Salles, John Francis Seymour, Fritz Offner, Armando López-Guillermo, David Belada, Luc Xerri, Pierre Feugier, Réda Bouabdallah, John Vincent Catalano, Pauline Brice, Dolores Caballero, Corinne Haioun, Lars Moller Pedersen, Alain Delmer, David Simpson, Sirpa Leppa, Pierre Soubeyran, Anton Hagenbeek, Olivier Casasnovas, Tanin Intragumtornchai, Christophe Fermé, Maria Gomes da Silva, Catherine Sebban, Andrew Lister, Jane A Estell, Gustavo Milone, Anne Sonet, Myriam Mendila, Bertrand Coiffier, Hervé Tilly
Summary Lancet 2010; 377: 42–51 Published Online December 21, 2010 DOI:10.1016/S01406736(10)62175-7 See Comment page 4 Hospices Civils de Lyon, Université Claude Bernard, UMR CNRS5239, Pierre-Bénite, France (Prof G Salles MD, Prof B Coiffier MD); Peter MacCallum Cancer Centre and University of Melbourne, Melbourne, VIC, Australia (Prof J F Seymour FRACP); Ghent University, Ghent, Belgium (Prof F Offner MD); Hospital Clinic, Barcelona, Spain (A López-Guillermo MD); Charles University in Prague, Faculty of Medicine, University Hospital Hradec Králové, Prague, Czech Republic (D Belada MD); Institut Paoli Calmettes, Marseille, France (Prof L Xerri MD, R Bouabdallah MD); CHU de Nancy, Université Henri Poincaré, Nancy, France (Prof P Feugier MD); Frankston Hospital, Frankston, VIC, Australia (J V Catalano FRACP); Hôpital Saint-Louis, AP-HP, Paris, France (P Brice MD); Universitario de Salamanca, Salamanca, Spain (D Caballero MD); Hôpital Henri Mondor, AP-HP, Créteil, France (Prof C Haioun MD); Odense University Hospital, Odense, Denmark (L M Pedersen MD); CHU de Reims, Reims, France (Prof A Delmer MD); North Shore Hospital, Auckland, New Zealand (D Simpson FRACP); Helsinki University Central Hospital, Helsinki, Finland (S Leppa MD); Institut Bergonié and Université Victor Segalen Bordeaux 2, Bordeaux, France (Prof P Soubeyran MD); Academic Medical Centre,
42
Background Patients with follicular lymphoma can have long survival times, but disease progression typically occurs 3–5 years after initial treatment. We assessed the potential benefit of 2 years of rituximab maintenance after first-line treatment in patients with follicular lymphoma receiving a rituximab plus chemotherapy regimen. Methods The randomised, open-label PRIMA study was undertaken in 223 centres in 25 countries. 1217 patients with previously untreated follicular lymphoma needing systemic therapy received one of three non-randomised immunochemotherapy induction regimens used in routine practice. 1019 patients achieving a complete or partial response were then randomly assigned to receive 2 years of rituximab maintenance therapy (375 mg/m² every 8 weeks) or observation. Treatment was assigned equally by centralised block randomisation, stratified by induction regimen, response, region, and centre. Neither the participants nor those giving the interventions, assessing outcomes, and analysing data were masked to group assignments. The primary endpoint was progression-free survival (PFS). Analysis was by intention to treat. This trial is registered with ClinicalTrials.gov, number NCT00140582. Findings 505 patients were assigned to rituximab maintenance and 513 to observation (one patient died during randomisation). With a median follow-up of 36 months (IQR 30–42), PFS was 74·9% (95% CI 70·9–78·9) in the rituximab maintenance group (130 patients progressed) and 57·6% (53·2–62·0) in the observation group (218 progressed; hazard ratio [HR] 0·55, 95% CI 0·44–0·68, p<0·0001). 2 years after randomisation, 361 patients (71·5%) in the rituximab maintenance group were in complete or unconfirmed complete response versus 268 (52·2%) in the observation group (p=0·0001). Overall survival did not differ significantly between groups (HR 0·87, 95% CI 0·51–1·47). Grade 3 and 4 adverse events were recorded in 121 patients (24%) in the rituximab maintenance group and 84 (17%) in the observation group (risk ratio 1·46, 95% CI 1·14–1·87; p=0·0026). Infections (grades 2–4) were the most common adverse event, occurring in 197 (39%) and 123 (24%) patients, respectively (risk ratio 1·62, 95% CI 1·35–1·96; p<0·0001). Interpretation 2 years of rituximab maintenance therapy after immunochemotherapy as first-line treatment for follicular lymphoma significantly improves PFS. Funding Groupe d’Etude des Lymphomes de l’Adulte (GELA) and F Hoffmann-La Roche.
Introduction Follicular lymphoma is the second most common lymphoma subtype and, despite substantial improvements in survival, disseminated disease is usually incurable.1,2 The disease characteristically responds well to first-line therapy but typically manifests repeated relapses with the need for recurrent therapeutic interventions, with disease-free intervals becoming progressively shorter.3,4 Although some patients can initially be managed with watchful waiting because they are asymptomatic with no adverse prognostic features, most need systemic cytotoxic-based treatment. In the past decade, several randomised studies5–8 have established that the combination of the anti-CD20
monoclonal antibody rituximab with various chemotherapy regimens can improve patients’ overall survival, and this combination is now regarded as the standard of care in first-line follicular lymphoma. Rituximab, in view of its efficacy, pharmacokinetic characteristics, and safety profile, has already been investigated as maintenance treatment in patients with follicular lymphoma.9–13 Previous studies have shown a significant clinical benefit of rituximab maintenance in patients with relapsed disease after chemotherapy with or without rituximab.11,12 The use of rituximab maintenance has also been studied after initial treatment with single-agent rituximab9,10 or chemotherapy alone;13 however, neither of these induction www.thelancet.com Vol 377 January 1, 2011
Articles
regimens is considered optimum as initial treatment for patients in need of therapy. Hence, the PRIMA (Primary RItuximab and MAintenance) study was designed to assess the potential benefit of 2 years of rituximab maintenance after first-line treatment in patients with follicular lymphoma receiving a rituximab plus chemotherapy regimen.
Methods Study design and patients This open-label, international, multicentre randomised study was undertaken between December, 2004, and April 2007, in 223 centres in 25 countries. The trial consisted of two phases, induction and maintenance. Patients were eligible for induction if they were older than 18 years and presented with untreated follicular lymphoma (grade 1, 2, or 3a) diagnosed by a lymph-node biopsy (done within 4 months of study registration). Eligibility required at least one criterion of high tumour burden—namely, bulky disease (one lymphoma lesion greater than 7 cm); three separate nodes of 3 cm or more; symptomatic splenic enlargement; organ compression by tumour, pleural, or peritoneal effusion; raised serum concentrations of either lactate dehydrogenase or β2-microglobulin; or the presence of B symptoms. Patients had to have a performance status of 2 or less on the Eastern Cooperative Oncology Group (ECOG) scale and adequate haematological function (unless due to lymphoma). Noneligibility criteria were a diagnosis of follicular lymphoma grade 3b or transformed into diffuse large B-cell lymphoma, CNS involvement, or a life expectancy of less than 6 months. Patients with a previous history of cancer (apart from adequately treated non-melanoma skin cancer or insitu cervical cancer), with poor renal or hepatic function (unless due to lymphoma), or a previous history of allergy to murine products were not eligible. Patients with a known HIV infection or an active hepatitis B or hepatitis C virus infection were also excluded, but pretreatment testing was not initially mandated. We excluded patients if they had had any major surgical procedure or used corticosteroids at doses greater than 20 mg per day within 1 month before study entry. The protocol was approved by local or national ethics committees according to the laws of each country, and the study was undertaken in accordance with the Declaration of Helsinki. Patients were required to provide written informed consent before registration.
Randomisation and masking Patients were randomly assigned in a 1:1 ratio to observation or rituximab maintenance (12 infusions of 375 mg/m² given intravenously, one every 8 weeks) starting 8 weeks after the last induction treatment. The duration and schedule of rituximab maintenance in this study was derived from previous studies using either eight infusions11 or 16 infusions13 over 2 years and from pharmacokinetic studies suggesting that 2-month intervals might be more www.thelancet.com Vol 377 January 1, 2011
suitable than 3 months to reach a trough in serum rituximab concentration of 25 μg/mL.14,15 Randomisation was stratified for induction regimen, response to induction treatment, geographical region, and centre, with a block size of four. Investigators enrolled the participants, and assignment to trial groups was done with a computerassisted randomisation allocation sequence (generated by a statistician) that took place centrally at Groupe d’Etude des Lymphomes de l’Adulte (GELA) central offices with a fax process, without the intervention of investigators. Neither the participants nor those giving the interventions, assessing outcomes, and analysing data were masked to group assignment.
Procedures Initial staging included physical examination; standard laboratory assessments; CT scans of the chest, abdomen, and pelvis; and bone marrow biopsy. Pathological specimens were centrally reviewed by a panel of expert pathologists in each country or in the GELA pathology centre. During the induction phase, patients were treated with one of three standard immunochemotherapy regimens, with each centre selecting the preferred regimen for all patients enrolled in that centre. The three chemotherapy regimens combined with rituximab were: CVP (cyclophosphamide 750 mg/m² given intravenously on day 1, vincristine 1·4 mg/m² [capped at 2 mg] given intravenously on day 1, and prednisone 40 mg/m² given orally on days 1–5, with each cycle repeated every 3 weeks for eight cycles), CHOP (cyclophosphamide 750 mg/m² given intravenously on day 1, vincristine 1·4 mg/m² [capped at 2 mg] given intravenously on day 1, doxorubicin 50 mg/m² given intravenously on day 1, and prednisone 100 mg given orally on days 1–5, with each cycle repeated every 3 weeks for six cycles), or FCM (fludarabine 25 mg/m² given intravenously on days 1–3, cyclophosphamide 200 mg/m² given orally on days 1–3, and mitoxantrone 6 mg/m² given intravenously on day 1, with each cycle repeated every 4 weeks for six cycles). Rituximab (375 mg/m² for each infusion) was administered at day 1 of each chemotherapy course, with two additional infusions administered in patients given CHOP (every 3 weeks after the last cycle) and FCM (2 weeks after the first and the fourth cycles) to provide an equivalent exposure to the antibody during induction. Response to induction16 was assessed 2–4 weeks after the last induction treatment course. Patients who obtained a complete response, an unconfirmed complete response, or a partial response were eligible for randomisation to the maintenance phase of the study. Additionally, patients were required to have received at least six cycles of rituximab plus CVP (R-CVP), four cycles of rituximab plus CHOP (R-CHOP), or four cycles of rituximab plus FCM (R-FCM) (each with at least six infusions of rituximab) without a delay of more than 2 weeks between each cycle. Any severe underlying
Amsterdam, Netherlands (Prof A Hagenbeek MD); CHU de Dijon, Dijon, France (O Casasnovas MD); Chulalongkorn University, Bangkok, Thailand (Prof T Intragumtornchai MD); Institut de Cancérologie Gustave Roussy, Villejuif, France (C Fermé MD); Portuguese Institute of Oncology, Lisbon, Portugal (M G da Silva MD); Centre Léon Bérard, Lyon, France (C Sebban MD); Centre for Medical Oncology, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London, UK (Prof A Lister MD); Concord Hospital, Concord, NSW, Australia (J A Estell FRACP); Fundaleu, Buenos Aires, Argentina (G Milone MD); UCL, Mont-Godinne, Yvoir, Belgium (A Sonet MD); F Hoffmann-La Roche, Basel, Switzerland (M Mendila MD); and Centre Henri Becquerel, Rouen, France (Prof H Tilly MD) Correspondence to: Prof Gilles Salles, Hospices Civils de Lyon, Université Claude Bernard, UMR CNRS5239, Centre Hospitalier Lyon-Sud, 69310 Pierre-Bénite, France
[email protected]
For the protocol for the PRIMA study see http://prima.gela.org
43
Articles
medical disorder that could impair participation or any major induction treatment-related toxic effect precluded eligibility.
Patients were assessed by clinical examination every 8 weeks and CT scans every 6 months during the 2-year maintenance phase of the study. An end-of-
1217 patients in induction phase 15 excluded (3 sites closed prematurely) 1202 patients registered
885 R-CHOP 4 withdrawn before treatment
44
44 patients treated
268 patients treated
15 withdrawn during treatment 2 major protocol violation 4 treatment failure 5 treatment toxicity 1 patient voluntary withdrawal 3 other
41 withdrawn during treatment 3 major protocol violation 12 treatment failure 5 treatment toxicity 1 patient voluntary withdrawal 1 death 19 other 791 completed treatment
Figure 1: Trial profile When the reasons for withdrawal categorised as other were investigated and centrally reviewed, of the 184 patients withdrawing from registration until randomisation, the main reasons leading to withdrawal were: major protocol violation (including eligibility criteria or inadequate induction treatment, n=61); complete or partial response not achieved after induction therapy (n=46); toxic effects or delays during induction treatment administration (n=44); underlying medical disease or death (n=18); patient decision (n=8); and investigator decision (n=7). R-CHOP=rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone. R-CVP=rituximab, cyclophosphamide, vincristine, and prednisone. R-FCM=rituximab, fludarabine, cyclophosphamide, and mitoxantrone. ITT=intention to treat.
1 withdrawn before treatment
4 withdrawn before treatment
881 patients treated 90 withdrawn during treatment 23 major protocol violation 15 treatment failure 17 treatment toxicity 1 patient voluntary withdrawal 5 death 29 other
45 R-FCM
272 R-CVP
227 completed treatment
29 completed treatment
5 not randomised 1 patient voluntary withdrawal 4 other
22 not randomised 1 major protocol violation 3 treatment toxicity 3 patient voluntary withdrawal 15 other 769 patients randomised
1 not randomised 1 treatment toxicity
28 patients randomised
222 patients randomised
1019 randomised to maintenance phase 1 excluded Patient died during randomisation 1018 ITT population 513 observation
505 rituximab
5 withdrawn before treatment
4 withdrawn before treatment
508 patients observed 166 withdrawn during observation 147 treatment failure 1 treatment toxicity 8 patient voluntary withdrawal 10 other 338 patients completed observation
501 patients treated
4 not assessed at end of observation
5 not assessed at end of observation
103 withdrawn during observation 69 treatment failure 10 treatment toxicity 10 patient voluntary withdrawal 1 death 13 other 393 patients completed treatment
www.thelancet.com Vol 377 January 1, 2011
Articles
treatment assessment also required a bone marrow biopsy if initially involved. Follow-up assessments included clinical evaluation every 3 months and CT scans every 6 months for an additional 3 years. Quality-of-life questionnaires for functional assessment of cancer therapy—general (FACT-G) score and European Organisation for Research and Treatment of Cancer (EORTC) QLQ-C30 scale were planned to be completed by patients at registration, at the end of induction, and then every 12 months.
Randomised patients Patients who received induction treatment (n=1193)
Age >60 years Age (years) Male sex Ann Arbor stage III/IV
423 (35%) 56 (22–87)
Observation (n=513)
Rituximab maintenance (n=505)
180 (35%)
176 (35%)
55 (22–84)
57 (26–79)
622 (52%)
263 (51%)
270 (53%) 459 (91%)
1075 (90%)
459 (89%)
Statistical analysis
ECOG performance status ≥1
434 (36%)
172 (34%)
181 (36%)
The study was designed to show a 45% increase in median progression-free survival (PFS) from the time of randomisation (6 months after the start of induction therapy) with a power of 80% and an overall two-sided type I error of 5%, with use of a two-sided log-rank test. The trial was intended in 2004 to register 640 patients and randomly assign 480 (assuming that 75% of the patients would be randomly assigned on the basis of response rate and eligibility criteria). Mature information then became available from earlier trials,11,17,18 indicating a 6-month delay before seeing a benefit of maintenance after immunochemotherapy, and patient accrual was more rapid than was expected. Therefore, two protocol amendments were implemented before the first data analysis, increasing the final sample size to 1200 patients registered and 900 randomly assigned, and allowing a more meaningful examination of the primary endpoint in subgroups. Two interim analyses were originally scheduled to be done after 50% (n=172) and 75% (n=258) of the total number of anticipated events (n=344). The protocol was subsequently amended to remove the first interim analysis because the results would have been regarded as clinically immature with short follow-up. The α-spending function with the O’Brien–Fleming boundary was applied for the interim analysis to maintain the overall two-sided type I error of 0·05. Subgroup analyses for PFS were planned according to age, sex, categories defined by the initial follicular lymphoma international prognostic index (FLIPI),19 induction regimen, and response to induction (complete or unconfirmed complete response, or partial response). The primary study endpoint was PFS from the time of randomisation to rituximab maintenance or no further treatment (observation). Secondary endpoints were event-free survival, time to next chemotherapy treatment, time to next antilymphoma treatment, overall survival, response rate at the end of maintenance, safety, toxic effects, and quality of life. Response and progression were defined with international standard criteria.16 Survival functions were estimated by the Kaplan-Meier method and compared by log-rank test stratified by induction regimen and induction treatment response. Cox regression analysis was done to adjust for the effect of known prognostic factors (age, sex, FLIPI category, induction treatment,
B symptoms present
388 (33%)
156 (30%)
160 (32%)
Bone marrow lymphoma involvement
654 (55%)
285 (56%)
275 (54%)
Lactate dehydrogenase >ULN*
403 (34%)
164 (32%)
173 (34%)
Haemoglobin <120 g/L
239 (20%)
96 (19%)
100 (20%)
β2 microglobulin ≥3 mg/L*
348 (32%)
132 (28%)
148 (32%)
www.thelancet.com Vol 377 January 1, 2011
FLIPI score† Low (0–1 risk factors)
254 (21%)
110 (21%)
106 (21%)
Intermediate (2 risk factors)
423 (36%)
187 (36%)
183 (36%)
High (3–5 risk factors)
514 (43%)
216 (42%)
215 (43%)
Initial local diagnosis of FL (other than grade 3B)
1188 (100%)
512 (100%)
504 (100%)
Central pathological review done
1115 (93%)
487 (95%)
467 (92%)
994 (84%)
433 (84%)
425 (84%)
Diagnosis of other lymphoma subtype‡
56 (5%)
28 (5%)
16 (3%)
Unclassifiable or not assessable for technical reasons
65 (6%)
26 (5%)
26 (5%)
R-CHOP
885 (74%)
386 (75%)
382 (76%)
R-CVP
272 (23 %)
113 (22%)
109 (22%)
R-FCM
45 (4%)
14 (3%)
14 (3%) 205 (41%)
Confirmed FL (other than grade 3B)
Induction immunochemotherapy regimen
Response to induction Complete response
..
195 (38%)
Unconfirmed complete response
..
165 (32%)
155 (31%)
Partial response
..
152 (30%)
139 (28%)
Other§
..
1 (≤1%)
6 (1%)
Data are number (%) or median (range). ECOG=Eastern Cooperative Oncology Group. ULN=upper limit of normal. FLIPI=follicular lymphoma international prognostic index. FL=follicular lymphoma. R-CHOP=rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone. R-CVP=rituximab, cyclophosphamide, vincristine, and prednisone. R-FCM=rituximab, fludarabine, cyclophosphamide, and mitoxantrone. *Lactate dehydrogenase and β2-microglobulin serum concentrations available for only 1188 and 1101 patients, respectively. †FLIPI scores were collected at registration. The risk score includes five factors: age (>60 years), Ann Arbor stage (III or IV), haemoglobin (≤120 g/L), serum lactate dehydrogenase (>ULN), and number of nodal areas involved (five or more).17 Data are available for 1191 patients. ‡Other lymphoma subtypes enrolled included FL grade 3B (n=12), combined diffuse large B-cell lymphoma and FL (n=24), diffuse large B-cell lymphoma (n=10), mantle cell lymphoma (n=6), small lymphocytic lymphoma (n=2), Hodgkin’s lymphoma (n=1), and angioimmunoblastic T-cell lymphoma (n=1). §After data cleaning, one patient in the observation group and four in the rituximab maintenance group were assessed as having stable disease at randomisation, and two other patients in the rituximab maintenance group missed standard assessment procedures.
Table 1: Baseline demographics and clinical characteristics of patients at enrolment (before immunochemotherapy induction treatment), and at randomisation, according to study group
and response to induction) in the assessment of maintenance effect. Response rates and frequency of adverse events were compared with the χ² test. To establish whether the study treatment groups differed in quality of life assessed at the end of treatment, independent of any differences in quality of life assessed at the end of induction, we used ANCOVA. FACT-G total 45
Articles
scores and EORTC QLQ-C30 global health scores were collected during rituximab maintenance or observation, with repeated measurements over time (censored at the date of disease progression to avoid potential biases related to toxic effects of second-line treatment), and analysed with an unstructured covariance model. All efficacy analyses were done in the intention-to-treat randomised population. On Jan 14, 2009, 267 events were recorded, leading to the protocol-specified interim analysis undertaken by the data safety monitoring committee that declared the primary endpoint of PFS had been met and that the study should be fully analysed. The study was then terminated; at that time the median follow-up after randomisation was 25 months (IQR 19–30) with 231 patients still to complete the maintenance or observation period. This report describes the analysis done after an additional year of follow-up, with a cutoff date of Jan 15, 2010, when all randomised patients had reached the time for end-of-treatment evaluation or were withdrawn. The study is registered with ClinicalTrials.gov, number NCT00140582.
Role of the funding source The GELA acted as the sponsor of the study and developed and undertook the study protocol in collaboration with several other lymphoma study
groups, F Hoffmann-La Roche, and Biogen IDEC. All case report forms were sent to the GELA central operation office (GELA-RC), and double data entry was undertaken for verification purposes. Queries and onsite monitoring data were used for final validation, and GELA-RC had full control of the database. An independent data safety monitoring committee examined the safety data every year, advised on protocol amendments, and undertook the prescheduled interim analysis of the efficacy results. Final statistical analyses were done independently by GELA-RC and F HoffmannLa Roche. The corresponding author was responsible for data analysis, data interpretation, and writing of the report; had full access to all the data in the study; and had the final responsibility for the decision to submit for publication.
Results Figure 1 shows the trial profile. On-site monitoring showed that three centres did not adhere to good clinical practice and were closed (excluding 15 patients); nine other registered patients withdrew before receiving any treatment. Therefore, 1193 patients had complete data at study enrolment (table table 1). Of those, 1115 (93%) had central pathology review done, and follicular lymphoma (except grade 3B) was confirmed in 994 cases (89%). We noted no differences in demographics or disease characteristics at enrolment between patients who
A
B
Event-free rate
1·0
Rituximab maintenance Observation
0·8 0·6 0·4 0·2 HR 0·55 (95% CI 0·44–0·68); p<0·0001
0
Number at risk Rituximab 505 Observation 513
HR 0·60 (95% CI 0·47–0·76); p<0·0001
Time (months) 472 469
445 415
423 367
404 334
307 247
207 161
Time (months) 84 70
17 16
0 0
505 513
C
483 487
455 452
441 417
414 380
312 286
209 170
91 71
17 18
0 0
30 36 42 Time (months)
48
54
22 26
1 0
D
Event-free rate
1·0 0·8 0·6 0·4 0·2 HR 0·62 (95% CI 0·47–0·81); p=0·0004
HR 0·87 (95% CI 0·51–1·47); p=0·60
0 0 Number at risk Rituximab 505 Observation 513
6
12
18
24
484 492
459 460
444 425
428 393
30 36 42 Time (months) 325 302
220 188
93 75
48
54
19 20
0 0
60
0
6
12
18
24
505 513
499 507
492 501
483 492
474 472
365 381
246 243
108 97
60
Figure 2: Kaplan-Meier estimates of progression-free survival (A), time to next antilymphoma treatment (B), time to next chemotherapy (C), and overall survival (D) from randomisation with rituximab maintenance versus observation HR=hazard ratio.
46
www.thelancet.com Vol 377 January 1, 2011
Articles
received R-CHOP (n=885), R-CVP (n=272), or R-FCM (n=45) induction (data not shown). 146 patients withdrew during induction treatment, a further 28 completed induction but were not randomly assigned (figure 1), and one patient died before notification of the randomisation. Therefore, from randomisation, the intention-to-treat population consisted of 1018 patients (505 in the rituximab maintenance group and 513 in the observation group; table 1). With a median follow-up of 36 months in both groups (IQR 30–42), 130 of 505 patients in the rituximab maintenance group and 218 of 513 in the observation group had documented disease progression, and five and three patients, respectively, had died before disease progression. 3-year PFS was 74·9% (95% CI 70·9–78·9) in the rituximab maintenance group and 57·6% (53·2–62·0) in the observation group (stratified log rank, p<0·0001; figure 2A). The risk of progression was significantly reduced for the rituximab maintenance group (hazard ratio [HR] 0·55, 95% CI 0·44–0·68). The median time to progression was not reached in the rituximab maintenance group and was estimated to be 48·3 months (95% CI 38·0 to not reached) in the observation group. Pre-planned analyses of patient subgroups categorised by age, sex, FLIPI score category, induction chemotherapy, and response to induction showed that the effect of rituximab maintenance was consistent across these different subgroups, although with borderline results for patients having received R-CVP (number for those having received R-FCM are too small to conclude) (figure 3). In a Cox regression multivariate analysis adjusted by prognostic factors, a longer PFS was significantly associated with randomisation to the rituximab maintenance group (HR 0·55, 95% CI 0·44–0·68; p<0·0001), an age of 60 years or older (0·68, 0·54–0·86; p=0·0013), female sex (0·76, 0·62–0·94; p=0·013), lower FLIPI score categories (overall p<0·0001), and R-CHOP or R-FCM as induction treatment (0·39, 0·17–0·89; p=0·0029). Overall, 102 of 505 patients in the rituximab maintenance group and 167 of 513 patients in the observation group had begun a new antilymphoma treatment, which was a chemotherapy regimen in 80 patients in the rituximab maintenance group and 129 in the observation group. We recorded significant reductions in the risk of starting a new antilymphoma treatment or death (figure 2B) or starting a new chemotherapy or death (figure 2C) in the rituximab maintenance group. Event-free survival was also significantly improved in the rituximab maintenance group (stratified HR 0·59, 95% CI 0·48–0·72). With only 26 deaths recorded in the rituximab maintenance group and 30 in the observation group with present follow-up, we noted no significant difference in overall survival (figure 2D). At the end of the maintenance phase of the study, 361 of 505 patients (71·5%; 95% CI 67·3–75·4) in the rituximab maintenance group were in complete or www.thelancet.com Vol 377 January 1, 2011
unconfirmed complete response compared with 268 of 513 patients (52·2%, 47·8–56·6) in the observation group (estimated difference 18·0%, 12·3–23·6; p=0·0001). More patients who were in partial response at the time of randomisation converted to complete or unconfirmed complete response after 2 years in the rituximab maintenance group (72/139 [52%]) than did those in the observation group (45/152 [30%]; estimated difference 22·2%, 95% CI 11·2–33·3, p=0·0001). HR (95% CI)
N
All
0·55 (0·44–0·68)
1018
<60
0·49 (0·37–0·65)
624
≥60
0·67 (0·47–0·94)
394
Women
0·63 (0·45–0·87)
485
Men
0·48 (0·36–0·64)
533
≤1
0·39 (0·21–0·72)
216
2
0·44 (0·30–0·64)
370
≥3
0·68 (0·51–0·92)
431
R-CHOP
0·51 (0·39–0·65)
768
R-CVP
0·68 (0·45–1·02)
222
R-FCM
0·54 (0·13–2·24)
28
Age (years)
Sex
FLIPI index
Induction chemotherapy
Response to induction CR/CRu
0·57 (0·44–0·74)
720
PR
0·48 (0·32–0·72)
291
1
0 Favours maintenance
2 Favours observation
3
Figure 3: Risk of progression with rituximab maintenance versus observation, according to prespecified subgroups HR=hazard ratio. FLIPI=follicular lymphoma international prognostic index. R-CHOP=rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone. R-CVP=rituximab, cyclophosphamide, vincristine, and prednisone. R-FCM=rituximab, fludarabine, cyclophosphamide, and mitoxantrone. CR=complete response. CRu=unconfirmed complete response. PR=partial response. Observation (n=508)
Rituximab maintenance (n=501)
Grade 3/4
Leading to treatment discontinuation
Grade 3/4
Leading to treatment discontinuation 19 (4%)†
All adverse events
84 (17%)
8 (2%)
121 (24%)
Neoplasia
17 (3%)
6 (1%)
20 (4%)
5 (1%)
0
18 (4%)
0
Neutropenia
5 (1%)
Febrile neutropenia
2 (<1%)
0
Infections
5 (1%)
0
22 (4%)
4 (1%)
13 (3%)
0
10 (2%)
0
5 (1%)
0
11 (2%)
CNS disorders Cardiac disorders Pregnancy
NA
2 (<1%)
1 (<1%)
NA
1 (<1%)
1 (<1%) 3 (1%)
Data are number (%). NA=not applicable. *Safety during maintenance was assessed for patients who undertook at least one visit (rituximab treatment or observation) after randomisation. All adverse events, defined as any adverse change from the patient’s baseline condition, whether considered related to treatment or not, were collected and graded according to the Common Terminology Criteria for Adverse Events 3.0 grading system.18 All grade 3 and 4 events plus grade 2 infections were recorded in detail during maintenance or observation and 6 months thereafter. †Other events leading to treatment discontinuation were pyrexia, fulminant hepatitis, hypersensitivity, post-procedural fistula, and lung disorder (one case each).
Table 2: Grade 3 and 4 adverse events* experienced by 2% or more of patients and adverse events leading to treatment discontinuation, after randomisation to rituximab maintenance or observation
47
Articles
Baseline (randomisation)
After 1 year
End of maintenance phase
Maintenance
Observation
Maintenance
Observation
Maintenance
Observation
131
125
118
89
111
61
IgA n g/L
1·33 (0·63)
1·57 (3·55)
1·26 (0·58)
1·40 (0·88)
1·25 (0·50)
1·55 (0·91)
IgG n g/L
131
125
7·87 (2·26)
118
7·76 (2·13)
89
7·73 (2·12)
100
8·22 (2·22)
60
7·48 (2·13)
8·31 (2·32)
IgM n g/L
127
121
0·64 (0·52)
113
0·59 (0·55)
89
0·55 (0·39)
110
0·55 (0·31)
61
0·51 (0·39)
0·58 (0·25)
Immunoglobulin concentrations are given as mean (SD). Serum concentrations of immunoglobulins were assessed only in a subset of patients in one participating country.
Table 3: Serum concentrations of immunoglobulins at randomisation, after 1 year of maintenance, and at the end of the maintenance phase
A 120
Rituximab maintenance Observation
Mean global health status score
100 80 60 40 20
Induction
Maintenance
Follow-up
0 Number of patients Rituximab maintenance Observation
394 374
343 341
217 166
224 158
93 56
B 120
Mean total score
100 80 60 40 20
Induction
Maintenance
Follow-up
0 Number of patients Rituximab maintenance Observation
Baseline
After induction
1 year
2 years
3 years
384 375
334 331
213 168
219 160
96 54
Figure 4: Quality of life during 2 years of treatment with rituximab maintenance versus observation alone (A) European Organisation for Research and Treatment of Cancer quality-of-life questionnaire C30 scale. (B) Functional assessment of cancer therapy—general scale. Quality-of-life assessments were censored at time of progression.
48
Of the 1009 patients assessed for safety, adverse events were reported in 281 of 501 (56%) patients in the rituximab maintenance group and 189 of 508 (37%) in the observation group (risk ratio 1·51, 95% CI 1·32–1·73; p<0·0001). The most common adverse events reported were grade 2–4 infections in 197 of 501 (39%) patients and 123 of 508 (24%) patients, respectively (risk ratio 1·62, 1·35–1·96; p<0·0001). The five most common infections reported in the rituximab and observation groups were bronchitis (54 and 28 cases, respectively), upper respiratory tract infections (28 and 11 cases), sinusitis (21 and eight cases), urinary tract infections (14 and nine cases), nasopharyngitis (11 and 14 cases), and urinary tract infections (14 and nine cases), whereas the cumulative number of herpes viruses-related infections were 19 and 12, respectively. Grade 3 or 4 adverse events20 (table table 2) occurred in 121 of 501 (24%) patients in the rituximab group and 84 of 508 (17%) patients in the observation group (risk ratio 1·46, 1·14–1·87; p=0·0026). 19 (4%) and eight (2%) events, respectively, resulted in treatment discontinuation (risk ratio 2·41, 1·06–5·45; p=0·029). Only one death (fulminant hepatitis B in the absence of viral suppressive therapy in the rituximab maintenance group) was reported to be possibly related to treatment toxic effects; other causes of deaths before lymphoma recurrence were attributed to other malignant diseases (four cases; one in rituximab maintenance group and three in observation group), or pulmonary haemorrhage, accident, or unknown (sudden death) (one case each in rituximab maintenance group). Two other patients (one from each group) developed progressive multifocal leukoencephalopathy after lymphoma relapse and subsequent treatments, which included investigational agents for both patients. At the end of 2 years of rituximab maintenance or observation, median serum concentrations of immunoglobulin isotypes did not differ significantly between the rituximab maintenance and observation groups (table 3). The mean adjusted FACT-G total scores at the end of treatment were 86·6 (95% CI 85·0–88·3) in the rituximab maintenance group and 87·2 (85·3–89·1) in the observation group, suggesting no association of these scores with treatment group (ANCOVA adjusted for scores at the end of induction, p=0·68). Consistently, the EORTC QLQ-C30 global health status mean scores were 75·5 (72·8–78·2) and 75·2 (72·0–78·4), respectively (p=0·89; figure 4). The analysis of repeated measures was consistent with ANCOVA, and did not show any statistical difference on either scale (data not shown).
Discussion Results of the PRIMA study show that 2 years of rituximab maintenance therapy significantly prolongs PFS, delays the time to next antilymphoma treatment and next chemotherapy, and improves the quality of response in patients with previously untreated follicular www.thelancet.com Vol 377 January 1, 2011
Articles
lymphoma that is responsive to first-line rituximab plus chemotherapy. Rituximab maintenance was well tolerated, with a limited number of adverse events resulting in treatment discontinuation, and there were no unexpected safety findings. In line with other studies,21 we recorded a significantly increased incidence of infectious events, mostly of mild to moderate severity, despite no significant decrease in serum immunoglobulin concentrations. Physicians and patients should be aware of this risk to optimise the management of these patients. Further follow-up of immunoglobulin concentrations will also be done in the trial. Nevertheless, despite the higher frequency of adverse events in the rituximab maintenance group than in the observation group, few patients withdrew from the study because of toxic effects. Furthermore, the burden associated with repeated infusions over 2 years did not seem to impair patient quality of life, which was similar in both study groups. However, since only a subset of patients completed the quality-of-life questionnaires, these data should be interpreted cautiously, because we cannot exclude a reporting bias favouring patients who did not have treatment-related adverse events. The significant reduction in rate of lymphoma progression in the rituximab maintenance group was consistent between patients with different demographics, disease characteristics, and prognostic factors, within the limitations of the study eligibility criteria. Together with the results of the multivariate Cox regression analysis, these data suggest that all analysed categories of patients eligible for first-line immunochemotherapy benefited from rituximab maintenance. Notably, we recorded a reduction in the risk of lymphoma recurrence irrespective of the intensity of the first-line induction therapy and the response achieved. Although we noted no gain in overall survival, the reduction in the risk of disease progression after responding to induction is likely to be preferred by patients with follicular lymphoma. This preference should be balanced with the constraints and costs associated with 2-year rituximab maintenance. However, in view of the substantial improvement in patient survival during the past decade2 and the indication that rituximab maintenance might also result in overall survival benefits in relapsed follicular lymphoma,11,21,22 we should not exclude such a possible benefit of this intervention in the long term. Different mechanisms of action have been proposed for the therapeutic activity of rituximab.23 When used as a single agent in a prolonged maintenance scheme, the immune-mediated activity of the antibody could possibly be more potent than other mechanisms. Data suggest that anti-CD20 antibodies might induce a T-cell-specific response against lymphoma cells.24,25 Other studies investigating the mechanisms of follicular lymphoma development have suggested that lymphoma precursor cells are able to survive for years both in www.thelancet.com Vol 377 January 1, 2011
Panel: Research in context Systematic review We searched Medline from January, 1995, to November, 2010, for full papers reporting randomised clinical trials and meta-analyses with the terms “rituximab maintenance” and “lymphoma”. We identified five randomised clinical trials in patients with follicular lymphoma,9–13 with two updates22,33 and one meta-analysis.21 Interpretation Together, these studies provide substantial evidence that rituximab maintenance improves the outcome of patients with follicular lymphoma in term of progression-free and overall survival. The use of rituximab maintenance is associated with an increased risk of infections. Our results are consistent with findings from other studies, but provide evidence that this intervention improves progression-free survival and response rate in patients with follicular lymphoma responding to a combination of chemotherapy plus rituximab administered as first-line treatment.
healthy individuals26 and in patients,27,28 where they can lead to disease recurrence. Rituximab maintenance might eventually exert a long-term control on these cells. Follicular lymphoma is an indolent disease, with prolonged survival even in cases that are ultimately fatal. The efficacy of salvage therapies used after initial treatment failure could also preclude the demonstration of an overall survival benefit associated with first-line therapy.22,29–31 With the present 3-year follow-up, less than 5% of patients in either group had died, with no significant difference between the two groups, and salvage therapy results are still immature. However, a higher proportion of patients achieved a complete response at the end of rituximab maintenance treatment than in the observation group. Attainment of a complete response is associated with improved longterm survival in patients with follicular lymphoma.32 Since longer follow-up will be needed to show any possible effect of rituximab maintenance on overall survival, we will continue to follow up these patients. The results of other studies assessing radioimmunotherapy consolidation with iodine-131 labelled tositumomab (regis-tered with ClinicalTrials.gov, numbers NCT00006721 and NCT00770224) or rituximab maintenance for 4 or 5 years (NCT00877214 and NCT00227695) might also provide some insights about how to improve the outcome of patients with follicular lymphoma. In summary, the data from this study suggest that rituximab maintenance in patients with high tumour burden follicular lymphoma, who respond to rituximab plus chemotherapy induction, improves PFS and should now be considered as first-line treatment for these patients (panel). 49
Articles
Contributors GS, JFS, FO, AL-G, LX, PF, PB, CH, AH, AL, AS, MM, BC, and HT designed the study. GS, JFS, FO, AL-G, DB, LX, LMP, SL, TI, MGdS, AH, AL, and MM were responsible for the conduct of the study centrally or at the country-specific level. GS, JFS, FO, AL-G, LX, PF, PB, CH, LMP, SL, TI, MGdS, RB, JVC, DC, AD, DS, PS, OC, CF, CS, JAE, GM, AH, AL, AS, MM, BC, and HT contributed research data to the study. GS, JFS, FO, AL-G, LX, PF, PB, CH, AH, AL, AS, BC, and HT contributed to data analysis and interpretation. GS drafted the report, which all co-authors critically revised for significant scientific content. Investigators participating in the study France G Salles, P Feugier, R Bouabdallah, C Gisselbrecht, C Haioun, A Delmer, H Tilly, O Casasnovas, C Ferme, P Soubeyran, C Sebban, B Christian, R Delarue, D Guyotat, H Maisonneuve, O Fitoussi, J Gabarre, T Lamy, F Morschhauser, J F Rossi, D Decaudin, P Colombat, V Delwail, M Janvier, C Recher, C Salanoubat, Z Marjanovic, M Blanc, C Foussard, J-L Harrousseau, E Jourdan, F Maloisel, L Al Jassem, T De Revel, A Devidas, J C Eisenmann, E Fleck, G Lepeu, C Martin, B Corront, P Moreau, A Thyss, B Anglaret, B Salles, M Alexis, K Bouabdallah, S Castaigne, F Dreyfus, P Fenaux, C Fruchart, M Macro, F Bauduer, D Bordessoule, M Fabbro, A Le Pourhiet, S Sadot Le Bouvier, P Solal Celigny, X Vallantin, C Kulekci, S Lefort, L Mosser, J F Ramee, N Morineau, B Audhuy, F Boue, M Flesch, H Gonzalez, J Gutnecht, F Marechal, A Belhabri, W Abarah, S Cailleres, N Denizon, O Fain, J-M Karsenti, P Morel, J-N Munck, H Cure, O Tournilhac, M Wetterwald. Australia J Catalano, J Estell, N Wickham, P Marlton, J Seymour, M Walsh, P Bardy, U Hahn, M Hertzberg, D Ma, I Prosser, C Tiley, R Filshie, C Arthur, K Fay, P Campbell, G Kannourakis, J Bashford, R Blum, R Herrmann, I Irving, M Leahy, I Lewis, R Lowenthal, J McKendrick, A Spencer, C Underhill, T Brighton, G Cull, B Augustson. Y-L Kwan. Belgium F Offner, A Bosly, P Zachee, M Maerevoet, T Connerotte, E Van Den Neste, A Van Hoof, K Van Eygen, B De Prijck, S Van Steenweghen, D Bron, A Kentos, P Pierre, H Demuynck, M Andre, O Hamdam, V Mathieux, P Mineur, V Verschaeve. Spain D Caballero, E Montserrat, A López Guillermo, J J Bargay, S Gardella, J A Marquez, J A Soler, J Briones, C Estany, J Besalduch, J Capote, L Escoda, E Monzo, E Perez Ceballos, J Ribera, A Bailen, I Espagnol, F Losal, J A Garcia Marco, A Lopez, J M Macia Virgili. Denmark L Moller Pedersen, P B Hansen, O Vestergaard Gadeberg, M Hansen, I B Sillesen, A Bukh, H Hasselbalch, N Toffner-Clausen, L Enggaard, F D’Amore, S Ingeberg, M Petersen, A J Vangsted. Czech Republic A Oborilova, D Belada, M Trneny, T Papajik, H Siffnerova, M Brejcha, M Jankovska, M Matuska, J Prausova. New Zealand D Simpson, L Berkahn, P Ganly, H Blacklock. Finland S Leppa, K Vasala, T Lehtinen, S Jyrkkio. Netherlands A Hagenbeek, J W Baars, M Chamuleau, M R Schipperus, D H Biesma, O De Weerdt, R E Brouwer, F Heyning, J M M Raemaekers, P Sonneveld, J K Doorduijn, R Van der Griend, M B Van’t Veer, E Lugtenburg. Thailand I Tanin, C Suporn, S Noppadol. Portugal M Gomes da Silva. UK R Pettengell, A Lister, K Ardeshna, D Culligan, D Cunningham, P Johnson, H McCarthy, D Moir, D White, J Radford. Argentina G Milone, J Milone, G Kusminsky, D Riveros. Brazil A Zanichelli, C Chiattone, D Chamone, J Vaz, D Tabak. Colombia M Urrego, A Zambrano, M Duarte. India G Babu, A Ranade, A Vaid, D C Doval. Peru L Casanova, F Hurtado de Mendoza. Israel O Shpilberg, D Attias, I Levi. Serbia D Boskovic, S Popovic, D Jovanovic. Venezuela M A Torres. China Z Yongqiang, S Zhi-Xiang, G Zhong-Zhen, S Yuan Kai. Croatia J Branimir. Turkey B Ferhanoglu, M Cetin, M Ozcan, B Undar. Uruguay E Bodega, M Nesse. Conflicts of interest GS, JFS, ALG, DB, PF, PS, CH, SL, AH, OC, AL, BC, AD, and HT declared consultancy, honoraria, and/or advisory board membership from Roche and/or Genentech. PF, CH, AD, SL, PS, OC, AL, and BC received grant support from Roche. GS, JFS, DB, CH, AD, SL, PS, MGdS, and BC received travel support from Roche. GS, JFS, ALG, SL, and AL have received support for preparation of educational materials and/or lectures from Roche. PF has received payment for participation in data review meetings and has stock options with Roche. MM is a
50
paid employee of Roche and has stock options with Roche. GS is also an advisory board member for Celgene, Janssen-Cilag, Genzyme, and Calistoga; and has received travel support from Pfizer and support for educational materials from Janssen-Cilag and Celgene. ALG has received support for educational materials from Celgene. HT is an advisory board member for Celgene and Seattle Genetics; has received grant support from Celgene and Amgen; and has received support for preparation of educational materials from Celgene, Amgen, Janssen-Cilag, and Pfizer. SL declared consultancy and honoraria for Bayer and has received travel support from GlaxoSmithKline. FO, LX, JVC, RB, PB, DC, LMP, DS, TI, CF, CS, JE, GM, and AS declare that they have no conflicts of interest. Acknowledgments This study was funded by Groupe d’Etude des Lymphomes de l’Adulte (GELA; Paris, France) and F Hoffmann-La Roche (Basel, Switzerland). We thank the pathologists undertaking central case review, particularly N Brousse, D Canioni, F Charlotte, C Chassagne-Clément, P Dartigues, B Fabiani, L Deleval, E Campos, and D DeJong; the statisticians C Bergé, M Fournier, J Maurer, and their teams; the entire GELA-RC team, including D Germain as project manager, all monitors and clinical research associates, and the data management and pharmacovigilance teams; the teams from the Australasian Leukaemia and Lymphoma Group, the Fundación Farreras Valentí, the Czech Lymphoma Study Group, the Hemato-Oncologie Volwassenen Nederland (HOVON), and the UK Haematology Trials Group for their contribution in organising the study in their respective countries; the Roche teams for their support and contribution to the study; the Data Safety Monitoring Committee members J O Armitage, D Hasenclever, and M Ghielmini; and R Marcus for his original input into the study design. Editing assistance for the manuscript was provided by John Carron, an employee of an independent medical writing agency funded by F Hoffmann-La Roche. References 1 Swenson WT, Wooldridge JE, Lynch CF, Forman-Hoffman VL, Chrischilles E, Link BK. Improved survival of follicular lymphoma patients in the United States. J Clin Oncol 2005; 23: 5019–26. 2 Pulte D, Gondos A, Brenner H. Ongoing improvement in outcomes for patients diagnosed as having Non-Hodgkin lymphoma from the 1990s to the early 21st century. Arch Intern Med 2008; 168: 469–76. 3 Gallagher CJ, Gregory WM, Jones AE, et al. Follicular lymphoma: prognostic factors for response and survival. J Clin Oncol 1986; 4: 1470–80. 4 Johnson PW, Rohatiner AZ, Whelan JS, et al. Patterns of survival in patients with recurrent follicular lymphoma: a 20-year study from a single center. J Clin Oncol 1995; 13: 140–47. 5 Marcus R, Imrie K, Belch A, et al. CVP chemotherapy plus rituximab compared with CVP as first-line treatment for advanced follicular lymphoma. Blood 2005; 105: 1417–23. 6 Hiddemann W, Kneba M, Dreyling M, et al. Frontline therapy with rituximab added to the combination of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) significantly improves the outcome for patients with advanced-stage follicular lymphoma compared with therapy with CHOP alone: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group. Blood 2005; 106: 3725–32. 7 Herold M, Haas A, Srock S, et al. Rituximab added to first-line mitoxantrone, chlorambucil, and prednisolone chemotherapy followed by interferon maintenance prolongs survival in patients with advanced follicular lymphoma: an East German Study Group Hematology and Oncology Study. J Clin Oncol 2007; 25: 1986–92. 8 Salles G, Mounier N, de Guibert S, et al. Rituximab combined with chemotherapy and interferon in follicular lymphoma patients: results of the GELA-GOELAMS FL2000 study. Blood 2008; 112: 4824–31. 9 Hainsworth JD, Litchy S, Burris HA 3rd, et al. Rituximab as first-line and maintenance therapy for patients with indolent non-Hodgkin’s lymphoma. J Clin Oncol 2002; 20: 4261–67. 10 Ghielmini M, Schmitz SF, Cogliatti SB, et al. Prolonged treatment with rituximab in patients with follicular lymphoma significantly increases event-free survival and response duration compared with the standard weekly x 4 schedule. Blood 2004; 103: 4416–23.
www.thelancet.com Vol 377 January 1, 2011
Articles
11
12
13
14
15
16
17
18
19 20
21
van Oers MH, Klasa R, Marcus RE, et al. Rituximab maintenance improves clinical outcome of relapsed/resistant follicular non-Hodgkin lymphoma in patients both with and without rituximab during induction: results of a prospective randomized phase 3 intergroup trial. Blood 2006; 108: 3295–301. Forstpointner R, Unterhalt M, Dreyling M, et al. Maintenance therapy with rituximab leads to a significant prolongation of response duration after salvage therapy with a combination of rituximab, fludarabine, cyclophosphamide, and mitoxantrone (R-FCM) in patients with recurring and refractory follicular and mantle cell lymphomas: Results of a prospective randomized study of the German Low Grade Lymphoma Study Group (GLSG). Blood 2006; 108: 4003–08. Hochster H, Weller E, Gascoyne RD, et al. Maintenance rituximab after cyclophosphamide, vincristine, and prednisone prolongs progression-free survival in advanced indolent lymphoma: results of the randomized phase III ECOG1496 Study. J Clin Oncol 2009; 27: 1607–14. Gordan LN, Grow WB, Pusateri A, Douglas V, Mendenhall NP, Lynch JW. Phase II trial of individualized rituximab dosing for patients with CD20-positive lymphoproliferative disorders. J Clin Oncol 2005; 23: 1096–102. Kahl BS, Williams ME, Hong F, Gascoyne R, Horning SJ. Preliminary pharmacokinetic (PK) analysis of Eastern Cooperative Oncology Group Protocol E4402: Rituximab extended schedule or re-treatment trial (RESORT). Blood 2007; 110: 3420 (abstr). Cheson BD, Horning SJ, Coiffier B, et al. Report of an international workshop to standardize response criteria for non-Hodgkin’s lymphomas. NCI Sponsored International Working Group. J Clin Oncol 1999; 17: 1244–53. Marcus R, Imrie K, Solal-Céligny P, et al. Phase III study of R-CVP compared with cyclophosphamide, vincristine, and prednisone alone in patients with previously untreated advanced follicular lymphoma. J Clin Oncol 2008; 26: 4579–86. Buske C, Hoster E, Dreyling M, et al. Rituximab in combination with CHOP in patients with follicular lymphoma: Analysis of treatment outcome of 552 patients treated in a randomized trial of the German Low Grade Lymphoma Study Group (GLSG) after a follow up of 58 months. Blood 2008; 112: 2599 (abstr). Solal-Céligny P, Roy P, Colombat P, et al. Follicular lymphoma international prognostic index. Blood 2004; 104: 1258–65. Common Terminology Criteria for Adverse Events v3.0 (CTCAE). August, 2006. http://ctep.cancer.gov/protocoldevelopment/ electronic_applications/docs/ctcaev3.pdf (accessed Nov 1, 2010). Vidal L, Gafter-Gvili A, Leibovici L, et al. Rituximab maintenance for the treatment of patients with follicular lymphoma: systematic review and meta-analysis of randomized trials. J Natl Cancer Inst 2009; 101: 248–55.
www.thelancet.com Vol 377 January 1, 2011
22
23
24
25
26
27
28
29
30
31
32
33
van Oers MH, Van Glabbeke M, Giurgea L, et al. Rituximab maintenance treatment of relapsed/resistant follicular non-Hodgkin’s lymphoma: long-term outcome of the EORTC 20981 phase III randomized intergroup study. J Clin Oncol 2010; 28: 2853–58. Cartron G, Watier H, Golay J, Solal-Céligny P. From the bench to the bedside: ways to improve rituximab efficacy. Blood 2004; 104: 2635–42. Hilchey SP, Hyrien O, Mosmann TR, et al. Rituximab immunotherapy results in the induction of a lymphoma idiotype-specific T-cell response in patients with follicular lymphoma: support for a «vaccinal effect» of rituximab. Blood 2009; 113: 3809–12. Abès R, Gélizé E, Fridman WH, Teillaud JL. Long-lasting antitumor protection by anti-CD20 antibody through cellular immune response. Blood 2010; 116: 926–34. Roulland S, Navarro JM, Grenot P, et al. Follicular lymphoma-like B cells in healthy individuals: a novel intermediate step in early lymphomagenesis. J Exp Med 2006; 203: 2425–31. Carlotti E, Wrench D, Matthews J, et al. Transformation of follicular lymphoma to diffuse large B-cell lymphoma may occur by divergent evolution from a common progenitor cell or by direct evolution from the follicular lymphoma clone. Blood 2009; 113: 3553–57. Ruminy P, Jardin F, Picquenot JM, et al. S(mu) mutation patterns suggest different progression pathways in follicular lymphoma: early direct or late from FL progenitor cells. Blood 2008; 112: 1951–59. Friedberg JW, Cohen P, Chen L, et al. Bendamustine in patients with rituximab-refractory indolent and transformed non-Hodgkin’s lymphoma: results from a phase II multicenter, single-agent study. J Clin Oncol 2008; 26: 204–10. Kahl BS, Bartlett NL, Leonard JP, et al. Bendamustine is effective therapy in patients with rituximab-refractory, indolent B-cell non-Hodgkin lymphoma: results from a Multicenter Study. Cancer 2010; 116: 106–14. Sebban C, Brice P, Delarue R, et al. Impact of rituximab and/or high-dose therapy with autotransplant at time of relapse in patients with follicular lymphoma: a GELA study. J Clin Oncol 2008; 26: 3614–20. Bachy E, Brice P, Delarue R, et al. Long-term follow-up of patients with newly diagnosed follicular lymphoma in the prerituximab era: effect of response quality on survival-A study from the Groupe d’Etude des Lymphomes de l’Adulte. J Clin Oncol 2010; 28: 822–29. Martinelli G, Schmitz SF, Utiger U, et al. Long-term follow-up of patients with follicular lymphoma receiving single-agent rituximab at two different schedules in trial SAKK 35/98. J Clin Oncol 2010; 28: 4480–84.
51
Articles
Effect of single-dose anthelmintic treatment during pregnancy on an infant’s response to immunisation and on susceptibility to infectious diseases in infancy: a randomised, double-blind, placebo-controlled trial Emily L Webb, Patrice A Mawa, Juliet Ndibazza, Dennison Kizito, Alice Namatovu, Jacqueline Kyosiimire-Lugemwa, Bridget Nanteza, Margaret Nampijja, Lawrence Muhangi, Patrick W Woodburn, Hellen Akurut, Harriet Mpairwe, Miriam Akello, Nancy Lyadda, Joseph Bukusuba, Macklyn Kihembo, Moses Kizza, Robert Kizindo, Juliet Nabulime, Christine Ameke, Proscovia B Namujju, Robert Tweyongyere, Moses Muwanga, James A G Whitworth, Alison M Elliott
Summary Lancet 2011; 377: 52–62 Published Online December 21, 2010 DOI:10.1016/S01406736(10)61457-2 See Comment page 6 MRC/UVRI Uganda Research Unit on AIDS, Entebbe, Uganda (P A Mawa MSc, J Ndibazza MSc, D Kizito BBLT, J Kyosiimire-Lugemwa BBLT, B Nanteza MSc, M Nampijja MSc, L Muhangi MA(Econ), P W Woodburn MSc, H Akurut BSc, H Mpairwe MSc, J Bukusuba PGD ID, M Kihembo BBLT, M Kizza BA, R Kizindo, J Nabulime MSc, C Ameke, P B Namujju MSc, R Tweyongyere PhD, Prof J A G Whitworth MD, Prof A M Elliot MD); London School of Hygiene and Tropical Medicine, London, UK (E L Webb PhD, Prof J A G Whitworth, Prof A M Elliot); Entebbe Hospital, Entebbe, Uganda (M Akello RN, N Lyadda EN, M Muwanga MBChB); and Faculty of Veterinary Medicine, Makerere University, Kampala, Uganda (A Namatovu BVM) Correspondence to: Dr Emily L Webb, London School of Hygiene and Tropical Medicine, Department of Infectious Disease Epidemiology, Keppel Street, London WC1E 7HT, UK.
[email protected]
52
Background Helminth infections affect the human immune response. We investigated whether prenatal exposure to and treatment of maternal helminth infections affects development of an infant’s immune response to immunisations and unrelated infections. Methods In this randomised, double-blind, placebo-controlled trial, we enrolled 2507 women in the second or third trimester of pregnancy who were planning to deliver in Entebbe General Hospital, Entebbe, Uganda. With a computergenerated random number sequence in blocks of 100, we assigned patients to 440 mg albendazole and 40 mg/kg praziquantel (n=628), 440 mg albendazole and a praziquantel-matching placebo (n=625), 40 mg/kg praziquantel and an albendazole-matching placebo (n=626), or an albendazole-matching placebo and praziquantel-matching placebo (n=628). All participants and hospital staff were masked to allocation. Primary outcomes were immune response at age 1 year to BCG, tetanus, and measles immunisation; incidence of infectious diseases during infancy; and vertical HIV transmission. Analysis was by intention-to-treat. This trial is registered, number ISRCTN32849447. Findings Data were available at delivery for 2356 women, with 2345 livebirths; 2115 (90%) of liveborn infants remained in follow-up at 1 year of age. Neither albendazole nor praziquantel treatments affected infant response to BCG, tetanus, or measles immunisation. However, in infants of mothers with hookworm infection, albendazole treatment reduced interleukin-5 (geometric mean ratio 0·50, 95% CI 0·30–0·81, interaction p=0·02) and interleukin-13 (0·52, 0·34–0·82, 0·0005) response to tetanus toxoid. The rate per 100 person-years of malaria was 40·9 (95% CI 38·3–43·7), of diarrhoea was 134·1 (129·2–139·2), and of pneumonia was 22·3 (20·4–24·4). We noted no effect on infectious disease incidence for albendazole treatment (malaria [hazard ratio 0·95, 95% CI 0·79–1.14], diarrhoea [1·06, 0·96–1·16], pneumonia [1·11, 0·90–1·38]) or praziquantel treatment (malaria [1·00, 0·84–1·20], diarrhoea [1·07, 0·98–1·18], pneumonia [1·00, 0·80–1·24]). In HIV-exposed infants, 39 (18%) were infected at 6 weeks; vertical transmission was not associated with albendazole (odds ratio 0·70, 95% CI 0·35–1·42) or praziquantel (0·60, 0·29–1·23) treatment. Interpretation These results do not accord with the recently advocated policy of routine antenatal anthelmintic treatment, and the value of such a policy may need to be reviewed. Funding Wellcome Trust.
Introduction Worldwide, infectious diseases account for more than 50% of deaths of children younger than 5 years; pneumonia, diarrhoeal disease, and malaria are the three most common causes.1 More than half of these deaths occur in sub-Saharan Africa, where roughly 2·5 million children younger than 5 years are estimated to die every year from one of these three diseases. Immunisation is a key strategy to combat infectious diseases, but immunisation programmes in developing countries vary in effectiveness.2 For example, increases in measles immunisation coverage have led to substantial falls in the incidence of measles;3 by contrast, despite high
immunisation coverage, the prevalence of tuberculosis is high in sub-Saharan Africa, and is one of the leading causes of deaths in adults.4 BCG is the only available vaccine against tuberculosis and its effectiveness is lowest in countries closest to the equator.5,6 Soiltransmitted helminth infections and schistosomiasis are also prevalent in developing countries,7 and their geographical distribution has extensive overlap with areas in which rates of infectious diseases are highest and the effectiveness of BCG immunisation is lowest.8,9 Such overlap in disease distributions has led to the suggestion that chronic helminth infections could affect the epidemiological patterns of other diseases,10 through www.thelancet.com Vol 377 January 1, 2011
Articles
impairment of immune responses to immunisations and unrelated infections. This hypothesis was supported by the finding that T-helper-1 (Th1) cell responses (characterised by interferon-γ production, induced by viral and bacterial antigens, and required for protection against mycobacteria and other intracellular pathogens) and T-helper-2 (Th2) cell responses (characterised by production of interleukins 4, 5, and 13 and induced by allergens such as helminth antigens) are mutually inhibitory.11–13 However, helminth infections induce immunoregulation by various other mechanisms, such as interleukin-10 production, and affect responses to non-helminth antigens.14 Thus helminth infections could inhibit protective Th1 responses to unrelated organisms, such as viruses, bacteria, and vaccines, by inducing a Th1 to Th2 switch and through immunoregulatory mechanisms. Several studies of animals and human beings lend support to this hypothesis.14 Although helminth infections are generally thought to have detrimental effects on host responses to infectious pathogens, a beneficial effect against other diseases is also possible through suppression of pathological inflammatory responses; some researchers have suggested an inverse association between helminth infections and incidence of severe malaria.15 Helminth infections are rare in infancy, when immunisations are given and many infectious diseaserelated deaths occur. However, evidence exists that prenatal exposure to maternal helminth infection could have important effects on an infant’s immune response. Prenatal sensitisation to Wuchereria bancrofti, a filarial worm, is associated with increased susceptibility to W bancrofti infection in childhood,16 and prenatal sensitisation to filarial or schistosome antigens with reduced Th1 and increased Th2 cytokine responses to mycobacterial antigens after BCG immunisation at birth.2,17 Such findings suggest that prenatal exposure to maternal helminth infection modulates an infant’s immune response to vaccination and infectious pathogens, and that anthelmintic treatment during pregnancy can prevent these effects. We designed this trial of anthelmintic treatment during pregnancy to address this hypothesis. We also examined effects of anthelmintic treatment during pregnancy on infant mortality, and on anaemia and growth at age 1 year, to assess the overall risks and benefits of anthelmintic treatment during pregnancy.
Methods Study design and patients The study is described in detail elsewhere.18 Briefly, the study area was Entebbe Municipality and Katabi subcounty, beside Lake Victoria, Uganda. The area is occupied by urban, rural, and fishing communities. Helminth infection is highly prevalent in the area,19 and malaria, diarrhoea, and pneumonia are common in www.thelancet.com Vol 377 January 1, 2011
young children.20 The study population consisted of pregnant women who presented at the governmentfunded antenatal clinic at Entebbe General Hospital between April 9, 2003, and Nov 24, 2005, where roughly 70% of pregnant women from the study area received antenatal care.21 Women were included if resident in the study area, planning to deliver in the hospital, willing to know their HIV status, and in the second or third trimester of pregnancy. They were excluded if they had possible helminth-induced pathological changes (haemoglobin <80 g/L, clinically apparent severe liver disease, or diarrhoea with blood in stool), a history of an adverse reaction to anthelmintics, already been enrolled in the trial during an earlier pregnancy, or if the pregnancy was deemed abnormal by a midwife. All participants gave written, informed consent. Ethics approval was given by the Uganda Virus Research Institute, Uganda National Council for Science and Technology, and London School of Hygiene and Tropical Medicine.
Randomisation and masking We used a two-by-two factorial design to randomly assign patients in a 1:1:1:1 ratio to receive simultaneously either single-dose albendazole (440 mg) and single-dose praziquantel (40 mg/kg), albendazole and a praziquantelmatching placebo, an albendazole-matching placebo and praziquantel, or an albendazole-matching placebo and a praziquantel-matching placebo (albendazole and matching placebo, Glaxosmithkline, Brentford, UK; praziquantel and matching placebo, Medochemie Ltd, Limassal, Cyprus). The randomisation code was generated by the trial statistician with a computergenerated random number sequence, with block size 100. Treatments were packed in sealed envelopes and labelled with an allocation number by colleagues at the Medical Research Council Unit in Entebbe who did not otherwise contribute to the trial. Treatments were allocated in numerical order by trained interviewer-counsellors who observed the patients taking the treatment correctly on enrolment to the study. Treatment allocation was masked from all participants and staff during the study.
Procedures Demographic and clinical details, and blood samples were obtained at screening; stool samples were obtained before enrolment. After enrolment, women continued to receive standard antenatal care, including haematinics, tetanus immunisation, and intermittent presumptive treatment for malaria twice after their first trimester of pregnancy; women with HIV were offered intrapartum and neonatal single-dose nevirapine for prevention of mother-to-child (vertical) HIV transmission.22 Stool samples were obtained after delivery to assess effectiveness of anthelmintic treatment; thereafter, all mothers received praziquantel and albendazole. Infants received BCG and polio immunisation at birth; 53
Articles
See Online for webappendix
54
diphtheria, pertussis, tetanus, Haemophilus influenzae, hepatitis B, and polio immunisation at 6, 10, and 14 weeks; and measles immunisation at 9 months. Children also attended the study clinic when unwell; doctors diagnosed, treated, and recorded their illnesses. Community fieldworkers visited each participant’s home twice a month, measured the child’s temperature, and recorded symptoms reported by the child’s carer. At age 12 months, blood and stool samples were obtained from the children and growth outcomes were measured. Children who were unwell at their 12-month visit were given appropriate treatment and asked to return to complete the visit procedures when well. The primary outcomes were immune response at age 1 year to BCG, tetanus, and measles immunisation; incidence of malaria, diarrhoea, pneumonia, measles, and tuberculosis during infancy as diagnosed by doctors at the study clinic; and vertical HIV transmission.18 Planned secondary outcomes were growth and anaemia at age 1 year. Community-reported data for illness events were included as a secondary outcome for comparison with doctor-diagnosed illness events from clinic visits. We also considered two additional unplanned secondary outcomes: infant mortality and asymptomatic malaria (presence of malaria parasitaemia) at 1 year of age. Cytokine responses at 1 year of age to crude culture filtrate proteins of Mycobacterium tuberculosis (cCFP) were measured as an indicator of response to BCG immunisation. Cytokine responses at age 1 year to tetanus toxoid were measured as an indicator of response to tetanus immunisation. We examined stimulated interferon-γ (type 1), interleukin-5 (type 2), interleukin-13 (type 2), and interleukin-10 (regulatory) responses in a whole-blood assay, as previously described.23 Total serum IgG, IgG4, and IgE responses to tetanus toxoid were measured by ELISA (webappendix p 1). Total serum measles-specific IgG was measured by ELISA (Dade Behring/Siemens, Eschborn, Germany) according to the manufacturer’s protocol. Immunological assays were done after all samples had been obtained, in a randomised sequence (by use of a computer-generated random number sequence), to avoid confounding of secular trends with variations in assay performance. For the primary outcome, doctor-diagnosed illness events, clinical malaria was fever (temperature ≥37·5°C) with parasitaemia; diarrhoea was an infant’s carer’s definition, with stool frequency recorded;24 pneumonia was cough with difficulty in breathing, and age-specific fast breathing;25 measles was defined by standard clinical criteria and confirmed by measurement of specific antibody;26 and children with suspected tuberculosis were investigated as clinically indicated.27 For the secondary outcome, community-reported illnesses, febrile illness was defined as measured by fieldworkers (temperature ≥37·5°C) or as reported by the child’s carer; diarrhoea as reported by the carer, with stool frequency recorded; presumptive pneumonia was cough with difficulty in
breathing, or age-specific fast breathing as measured by fieldworkers. Stool samples were examined for helminth ova with the Kato-Katz method28 and by charcoal culture for Strongyloides sterocoralis infection;29 two Kato-Katz slides were prepared from each sample and examined for hookworm ova within 30 mins of preparation, or examined the next day for other species. Hookworm and Schistosoma mansoni infections were classified into low, medium, and high intensities according to WHO guidelines.30 Blood samples were examined by a modified Knott’s method for Mansonella perstans31 and by thick film for malaria parasites. Haemoglobin was estimated by Coulter analyser (Beckman Coulter, Nyon, Switzerland). Quality control for Kato-Katz analyses was provided by the Vector Control Programme of the Ministry of Health, Uganda, and for haematology and malaria parasitology through the UK National External Quality Assessment Schemes. Mothers’ HIV serology was done by rapid test algorithm.30 Blood was obtained from cord and at 6 weeks of age from infants of mothers with HIV for assessment of vertical HIV transmission. Plasma and whole blood cell pellet were separated by centrifugation and stored at –80°C until assays were done. For detection of HIV-1 proviral DNA in infants at 6 weeks, DNA was extracted from stored whole blood cell pellets and amplified by nested PCR of three conserved viral regions, tat, gp41, and nef (webappendix p 1). For both cord and 6-week samples, plasma HIV load was measured with Bayer Versant branched DNA assay version 3.0 (Bayer, Leverkusen, Germany) or Roche Amplicor HIV-1 RNA Monitor test version 1.5 (Roche, NJ, USA). Infants were regarded as being HIV positive if the 6-week sample had a positive DNA PCR for any of the viral regions and a viral load of 1000 copies per mL or more; for four infants, only viral load data were available, so they were used to establish HIV status. Viral load and DNA-PCR results were concordant apart from one infant (viral load 6699 copies per mL, PCR negative) who was seronegative by rapid test algorithm at age 18 months and was classified as HIV negative. In infants with HIV infection, transmission was regarded as likely to have been intrauterine if the viral load in cord blood was 1000 copies per mL or more.
Statistical analysis Analysis was done after all children were older than 15 months. Data for samples and measurements obtained at routine, 1-year visits were included if the child attended within 2 months after their first birthday. Data for illness events and mortality were censored strictly at 1 year. Results for younger twins were excluded from all analyses. On the basis of our preliminary study,32 the planned cohort size of 2500 was expected to accrue 1860 personyears of follow-up in infancy and 1594 infants were www.thelancet.com Vol 377 January 1, 2011
Articles
11 783 assessed for eligibility
2515 randomised
628 allocated to praziquantel and albendazole
628 allocated to praziquantel and albendazole-matching placebo
0 double enrolment
628 baseline group
581 analysed for illness episodes
630 allocated to the two placebo only
4 double enrolment
626 baseline group
Baseline
625 baseline group
0 haemoglobin <80 g/L 35 lost to follow-up before delivery 4 miscarriages
2 double enrolment
628 baseline group
0 haemoglobin <80 g/L 34 lost to follow-up before delivery 6 miscarriages
587 deliveries 596 births 13 stillbirths 583 livebirths
6 younger twins excluded from analyses 58 died or lost to follow-up before 1 year 98 still in follow-up but did not attend 1 year visit
629 allocated to albendazole and praziquantel-matching placebo
2 double enrolment
2 haemoglobin <80 g/L 31 lost to follow-up before delivery 4 miscarriages
591 deliveries 599 births 12 stillbirths 587 livebirths
Allocation
6243 not resident in Entebbe/Katabi 1186 unwilling to receive HIV result 874 unwilling to participate 646 incomplete screening procedures 115 already had baby in study 83 pregnancy not normal 37 not planning to deliver in hospital 17 haemoglobin <80 g/L 15 diarrhoea with blood in stool 5 adverse reaction to anthelmintics 0 clinically apparent severe liver disease 47 other or no reason stated
Birth
9 younger twins excluded from analyses 56 died or lost to follow-up before 1 year 100 still in follow-up but did not attend 1 year visit
585 deliveries 594 births 11 stillbirths 583 livebirths
0 haemoglobin <80 g/L 31 lost to follow-up before delivery 4 miscarriages
593 deliveries 600 births 8 stillbirths 592 livebirths
8 younger twins excluded from analyses 47 died or lost to follow-up before 1 year 99 still in follow-up but did not attend 1 year visit
574 analysed for illness episodes
7 younger twins excluded from analyses 69 died or lost to follow-up before 1 year 96 still in follow-up but did not attend 1 year visit
575 analysed for illness episodes
585 analysed for illness episodes
432 analysed at 1 year
422 analysed at 1 year
Analysis at 1 year 427 analysed at 1 year
420 analysed at 1 year
Figure: Trial profile
expected to be seen at age 1 year. For either maternal treatment, assuming no interaction between treatments, this number would give 80% power to detect rate ratios of 0·82 for malaria, 0·91 for diarrhoea, and 0·76 for pneumonia, with p values of less than 0·05, assuming frequency of disease in the placebo groups to be 50 per 100 person-years for malaria,33 190 per 100 person-years for diarrhoea,34 and 25 per 100 person-years for pneumonia.35 Incidence of both tuberculosis and measles was expected to be low,18 therefore only very large differences in incidence would be detected. Samples from 1594 infants assessed at 1 year would www.thelancet.com Vol 377 January 1, 2011
detect differences in infant cytokine responses of 0·11 log10 between intervention groups.18,32 The patients that were included in analysis for each of the primary outcomes differed because data for each outcome was obtained at different times. For immune response at age 1 year, analysis included all children who provided a blood sample at 1 year and who had received full BCG (for cCFP analysis) or tetanus (for tetanus toxoid analysis) immunisation at Entebbe Hospital; second-born twins were excluded. For incidence of infectious diseases during infancy, analysis included all liveborn children, excluding second-born twins. For vertical HIV 55
Articles
For more on igrowup macros see http://www.who.int/ childgrowth/software/en/
transmission, analysis included all children whose mothers were not receiving highly-active antiretroviral therapy, and from whom blood samples at age 6 weeks were available; second-born twins were excluded. Cytokine and antibody responses showed skewed distributions, with disproportionate numbers of zero values. Results were transformed to log10(concentration+1) and analysed by linear regression with bootstrapping to estimate bias-corrected accelerated confidence intervals.36 Regression coefficients were back-transformed to give geometric mean ratios. Interactions were examined with Wald tests. For doctor-diagnosed disease incidence, time at risk began at birth and was censored at loss to follow-up, death, or age 1 year. All children with known date of birth were included in the analysis until censoring, irrespective of whether they had made a clinic visit for illness. For each disease, we calculated incidence rates for all events. Disease episodes within 14 days of an initial presentation with the same disease were regarded as part of the same episode and excluded from the analysis; time at risk was adjusted accordingly, excluding these 14-day periods from the total person-time denominator. Hazard ratios (HRs) for effects of treatment were calculated with Cox regression, with robust SEs to allow for within-child clustering. For community-reported illness data, a generalised-estimatingequation approach with exchangeable correlation structure was used to model effects of treatment on repeated binary outcomes. Odds ratios (ORs) for the effects of treatment on vertical HIV transmission were calculated with logistic regression. The prevalence of asymptomatic malaria at 1 year was compared between treatment groups with logistic regression. Infant mortality per 1000 livebirths was estimated from Kaplan-Meier survival probabilities to age 1 year, and effects of maternal anthelmintic treatment were assessed by Cox regression. Weight-for-age, height-for-age, and weight-for-height Z scores at 1 year were derived from WHO growth standard reference scales, with igrowup macros. We examined effects of maternal treatment on Z scores and on haemoglobin at 1 year by linear regression. We did two prespecified subgroup analyses, examining effects of albendazole treatment in children of mothers with a hookworm infection, and effects of praziquantel treatment in children of mothers with schistosomiasis. Differences between subgroups were examined by fitting interaction terms in regression models. All p values were two-sided with no adjustment made for multiple comparisons. Analyses were done with Stata 10.1.
Role of the funding source The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. AME had full access to all the data in the study and had final responsibility for the decision to submit for publication. 56
Results The figure shows the trial profile. 2507 women were randomly assigned to treatment groups, 1309 (52%) of whom were in the second trimester of pregnancy and 1195 (48%) of whom were in the third trimester of pregnancy. The median length of time between receiving the study intervention and delivery was 87 days (IQR 54–118). Data were available for 2356 women (94%) at delivery and there were 2345 livebirths; results for 30 liveborn younger twins or triplets were excluded, which left data for follow-up of 2315 liveborn infants. Of these infants, 2092 (90%) remained in follow-up at 1 year of age, of whom 1701 (81%) attended the 1-year visit. Baseline characteristics of mothers were similar between the four groups (table 1).37 Mothers of children who did not contribute to follow-up were, on average, younger, more likely to be pregnant for the first time, have malaria or HIV infection, and have enrolled earlier during pregnancy than mothers of children who contributed to follow up (data not shown). Mothers of children who provided samples at 1 year of age were of higher socioeconomic status and less likely to reside in remote parts of the study area than were mothers of children who did not provide samples at age 1 year (data not shown). At enrolment, 1693 (68%) of women were infected with at least one helminth species. Individual prevalences of each species were: hookworm 45%, M perstans 21%, S mansoni 18%, S stercoralis 12%, Trichuris trichiura 9%, and Ascaris lumbricoides 2%. Infections were generally mild, with 942 (85%) of hookworm infections and 297 (65%) of S mansoni infections classified as light. Subgroups of 648 mothers who received albendazolematching placebo and 658 mothers who received praziquantel-matching placebo had three stool samples examined before being treated 6 weeks after delivery; in these women, the sensitivity of one stool sample compared with three stool samples was 89% for hookworm infection and 66% for schistosomiasis. 1957 mothers (78%) received at least one documented dose of tetanus immunisation during pregnancy; 1447 (74%) received all documented tetanus immunisations before the anthelmintic treatment intervention (median 4 days [IQR 3–7] before intervention). We noted no evidence of interaction between maternal albendazole and praziquantel treatments for any outcome (all interaction p values>0·1; webappendix pp 3–4); therefore the effects of each treatment were assessed independently. Results for cytokine responses at age 1 year were available for 1542 infants, of whom 1506 had received BCG immunisation at birth at Entebbe Hospital and 1015 had received all three doses of tetanus immunisation at Entebbe Hospital. The proportion of infants for whom positive responses to cCFP and tetanus toxoid were detected varied by cytokine; for cCFP, 1356 (90%) infants had positive responses to interferon-γ, 449 (30%) had positive responses to interleukin 5, 1005 (67%) had positive responses to interleukin 13, and 1316 (87%) www.thelancet.com Vol 377 January 1, 2011
Articles
had positive responses to interleukin-10. For tetanus toxoid, 655 (65%) infants had positive responses to interferon-γ, 499 (49%) had positive responses to interleukin 5, 766 (75%) had positive responses to interleukin 13, and 498 (49%) had positive responses to interleukin 10. Albendazole treatment of mothers was associated with a small reduction in unstimulated interleukin-10 production by their children (geometric mean ratio 0·83, 95% CI 0·70–0·98). We recorded no other overall effects of maternal treatment on unstimulated cytokine production. Results for effects on responses to stimulation were similar, irrespective of whether background cytokine production was subtracted (data not shown). We detected no effects of either albendazole or praziquantel treatments on the overall antigen-specific response after immunisation (table 2). However, in infants of mothers with a hookworm infection, maternal albendazole treatment was associated with a small nonsignificant reduction in the interferon-γ response to cCFP and with reductions in their response to tetanus toxoid for the type 2 cytokines interleukin 5 and interleukin 13 (table 3). We recorded no significant effect of praziquantel treatment in infants of mothers with schistosomiasis (table 3). We recorded no overall effects for either albendazole or praziquantel treatments on antibody concentrations for tetanus or measles (table 2), nor did we detect differential effects for either albendazole or praziquantel treatment according to susceptible worm species (table 3). Of the 2315 liveborn children, 2083 (90%) made at least one illness-related clinic visit during infancy. The number of illness visits by children in all four treatment groups were similar (webappendix p 4). During the study period, the rate of malaria was 41 episodes per 100 personyears (95% CI 38–44), of diarrhoea 134 episodes per 100 person-years (129–139), and of pneumonia 22 episodes per 100 person-years (20–24). Measles and tuberculosis were rare, with two episodes of each, so we could not assess the effect of maternal anthelmintic treatment on their frequency. We noted no evidence for effects of either albendazole or praziquantel treatments on doctor-diagnosed malaria, diarrhoea, or pneumonia incidence (table 4). Subgroup analyses by maternal hookworm and schistosomiasis infection status showed no evidence of a differential effect of either treatment according to susceptible worm species (table 5). 299 women tested positive for HIV at enrolment; of these, 16 were lost to follow-up before delivery, five had miscarriages, and nine had stillbirths. Five women were on highly-active antiretroviral therapy; their children, and seven second-born twins were excluded from the analysis, leaving 264 infants suitable for inclusion. 6-week blood samples were available from 211 infants (80%), of whom 39 (18%) were diagnosed with HIV infection. Of infants with HIV infection, 31 had cord samples available, www.thelancet.com Vol 377 January 1, 2011
Albendazole+ praziquantel (n=628)
Praziquantel only (n=626)
Albendazole only (n=625)
Placebo (n=628)
Age (years) <20
145 (23%)
165 (26%)
162 (26%)
158 (25%)
20–24
223 (36%)
230 (37%)
238 (38%)
255 (41%)
25–29
159 (25%)
143 (23%)
139 (22%)
122 (19%)
30–34
66 (11%)
62 (10%)
55 (9%)
70 (11%)
≥35
35 (6%)
26 (4%)
31 (5%)
23 (4%)
Education (4 missing values) None
25 (4%)
23 (4%)
17 (3%)
32 (5%)
Primary
311 (50%)
318 (51%)
318 (51%)
316 (50%)
Secondary
235 (37%)
229 (37%)
244 (39%)
226 (36%)
56 (9%)
55 (9%)
45 (7%)
53 (8%)
Tertiary Tribe (1 missing value) Baganda
303 (48%)
314 (50%)
313 (50%)
301 (48%)
Banyankole
64 (10%)
47 (8%)
56 (9%)
67 (11%)
Batoro
19 (3%)
27 (4%)
24 (4%)
32 (5%)
Basoga
36 (6%)
23 (4%)
27 (4%)
20 (3%) 37 (6%)
Luo
32 (5%)
41 (7%)
30 (5%)
Banyarwanda
38 (6%)
36 (6%)
31 (5%)
37 (6%)
136 (22%)
138 (22%)
144 (23%)
133 (21%) 39 (6%)
Others
Household socioeconomic status (49 missing values)* 1 (low)
33 (5%)
40 (7%)
35 (6%)
2
54 (9%)
57 (9%)
54 (9%)
52 (8%)
3
190 (31%)
199 (33%)
182 (30%)
194 (32%)
4
174 (28%)
174 (28%)
179 (29%)
183 (30%)
5
130 (21%)
109 (18%)
131 (21%)
115 (19%)
37 (6%)
32 (5%)
33 (5%)
32 (5%)
6 (high) Gravidity 1
178 (28%)
172 (27%)
181 (29%)
164 (26%)
2–4
339 (54%)
354 (57%)
350 (56%)
369 (59%)
≥5
111 (18%)
100 (16%)
94 (15%)
95 (15%)
Hookworm (11 missing values)
270 (43%)
301 (48%)
262 (42%)
277 (44%)
Schistosoma mansoni (11 missing values)
117 (19%)
104 (17%)
123 (20%)
114 (18%)
Mansonella perstans (8 missing values)
135 (22%)
136 (22%)
117 (19%)
143 (23%)
79 (13%)
61 (10%)
71 (11%)
88 (14%)
59 (10%)
63 (10%)
85 (14%)
Helminth infections
HIV status Positive
Malaria parasitaemia (48 missing values) Positive
61 (10%)
Data have been previously reported by Ndibazza and colleagues.37 *Household socioeconomic status was scored on the basis of building materials of the home, number of rooms, and items owned.
Table 1: Baseline characteristics
of which 20 (65%) showed an HIV viral load of 1000 copies per mL or more, which is consistent with intrauterine HIV transmission. We recorded no evidence for any effect of either albendazole (OR 0·70, 95% CI 0·35–1·42; p=0·33) or praziquantel (0·60, 0·29–1·23; p=0·17) treatment on vertical HIV transmission, although statistical power was restricted because the study was only powered to detect a large effect for this outcome. No 57
Articles
Geometric mean*
Geometric mean ratio (95% CI)†
Albendazole Placebo
Geometric mean* Praziquantel
Geometric mean ratio (95% CI)†
Placebo
Cytokine responses cCFP‡ Interferon-γ (pg/mL)
322
Interleukin 5 (pg/mL)
3·5
313 3·8
1·03 (0·83–1·27)
320
0·91 (0·74–1·11)
314
3·6
1·02 (0·82–1·26)
3·7
0·95 (0·78–1·17)
Interleukin 13 (pg/mL)
17
18
0·95 (0·77–1·18)
18
17
1·05 (0·84–1·30)
Interleukin 10 (pg/mL)
90
102
0·88 (0·73–1·06)
98
95
1·03 (0·85–1·24)
Interferon-γ (pg/mL)
31
36
0·85 (0·61–1·20)
34
33
1·03 (0·75–1·44)
Interleukin 5 (pg/mL)
10
13
0·76 (0·56–1·04)
11
11
0·93 (0·68–1·27)
Interleukin 13 (pg/mL)
40
41
0·97 (0·73–1·27)
37
44
0·84 (0·63–1·12)
Tetanus toxoid§
Interleukin 10 (pg/mL)
4·7
5·9
0·79 (0·61–1·02)
5·0
5·6
0·89 (0·69–1·16)
Antibody concentrations Tetanus toxoid¶ Total IgG (mIU/mL) IgG4 (ng/mL) IgE (ng/mL)
160
251
0·64 (0·39–1·05)
227
176
1·32 (0·78–2·12)
50
69
0·72 (0·37–1·37)
48
71
0·68 (0·36–1·29)
1389
1507
0·92 (0·76–1·11)
1460
1430
1·02 (0·85–1·24)
336
348
0·96 (0·85–1·10)
332
353
0·94 (0·83–1·07)
Measles|| Total IgG (mIU/mL)
cCFP=crude culture filtrate proteins of Mycobacterium tuberculosis. *Geometric mean of response concentration + 1. † Bias-corrected accelerated CIs computed by bootstrapping. ‡cCFP cytokine responses were available for 1506 infants given BCG at Entebbe hospital. §Tetanus toxoid cytokine results were available for 1015 infants who received all three doses of tetanus toxoid; one missing value for interferon-γ response. ¶ Tetanus toxoid antibody results were available for 1058 (IgG), 1057 (IgG4), and 979 (IgE) infants who received all three doses of tetanus toxoid. || Measles antibody results were available for 1233 infants who received measles immunisation.
Table 2: Effects of maternal anthelmintic treatment in pregnancy on infant response to BCG, tetanus, and measles immunisation
Albendazole Effect of treatment in women with hookworm infection (GMR, 95% CI)
Praziquantel Effect of treatment in women without hookworm infection (GMR, 95% CI)
p value for interaction
Effect of treatment in women with schistosomiasis (GMR, 95% CI)
Effect of treatment in women without schistosomiasis (GMR, 95% CI)
p value for interaction
cCFP* Interferon-γ
0·73 (0·52–1·02)
1·31 (1·00–1·74)
0·009
0·88 (0·50–1·56)
1·05 (0·83–1·33)
0·56
Interleukin 5
1·06 (0·78–1·44)
0·82 (0·62–1·08)
0·22
1·20 (0·75–1·87)
0·91 (0·73–1·15)
0·30
Interleukin 13
0·96 (0·69–1·32)
0·95 (0·70–1·26)
0·97
1·07 (0·64–1·80)
1·04 (0·82–1·32)
0·92
Interleukin 10
0·93 (0·70–1·26)
0·85 (0·66–1·08)
0·62
1·14 (0·68–1·92)
1·00 (0·82–1·22)
0·65
Interferon–γ
0·61 (0·36–1·05)
1·08 (0·71–1·66)
0·10
0·71 (0·32–1·60)
1·11 (0·78–1·60)
0·32
Interleukin 5
0·50 (0·30–0·81)
1·02 (0·68–1·52)
0·02
0·60 (0·30–1·28)
1·02 (0·72–1·43)
0·19
Interleukin 13
0·52 (0·34–0·82)
1·45 (1·02–2·10)
0·0005
0·56 (0·28–1·09)
0·92 (0·68–1·25)
0·18
Interleukin 10
0·74 (0·48–1·13)
0·83 (0·59–1·16)
0·68
0·94 (0·52–1·71)
0·89 (0·66–1·18)
0·86
Total IgG
0·52 (0·24–1·16)
0·74 (0·39–1·42)
0·51
0·72 (0·22–2·40)
1·43 (0·83–2·49)
0·30
IgG4
0·71 (0·25–1·96)
0·74 (0·31–1·71)
0·95
0·51 (0·10–2·56)
0·71 (0·36–1·45)
0·72
IgE
0·93 (0·69–1·25)
0·92 (0·71–1·16)
0·95
0·70 (0·45–1·17)
1·10 (0·90–1·36)
0·09
1·02 (0·83–1·26)
0·93 (0·79–1·09)
0·51
0·92 (0·69–1·24)
0·94 (0·82–1·08)
0·88
Tetanus toxoid†
Tetanus toxoid‡
Measles§ Total IgG
cCFP=crude culture filtrate proteins of Mycobacterium tuberculosis. GMR=geometric mean ratio. *cCFP cytokine responses were available for 1506 infants given BCG at Entebbe Hospital; four missing values in this analysis for infants of women with no stool result for Kato-Katz analysis. †Tetanus toxoid cytokine results were available for 1015 infants who received all three doses of tetanus toxoid; two missing values in this analysis for infants of women with no stool result for Kato-Katz analysis; one missing value for interferon-γ response to tetanus toxoid. ‡Tetanus toxoid antibody results were available for 1058 (IgG), 1057 (IgG4), and 979 (IgE) infants who received all three doses of tetanus toxoid; one missing value in this analysis for an infant of a woman with no stool result for Kato-Katz analysis. § Measles antibody results were available for 1233 infants who received measles immunisation; three missing values in this analysis for infants of women with no stool result for Kato-Katz analysis.
Table 3: Effect of maternal anthelmintic treatment on response to BCG, tetanus, and measles immunisation, by maternal helminth infection status
58
www.thelancet.com Vol 377 January 1, 2011
Articles
Malaria Diarrhoea Pneumonia
Albendazole
Albendazole-matching placebo
Events Rate per (person-years) 100 personyears (95% CI)
Events (personyears)
431 (1081) 1438 (1043) 256 (1087)
39·9 (36·3–43·8) 137·9 (131·0–145·3) 23·5 (20·8–26·6)
446 (1062) 1339 (1029) 226 (1070)
HR (95% CI)*
p value* Praziquantel Events (person– years)
Rate per 100 personyears (95% CI) 42·0 (38·3–46·1)
0·95 (0·79–1·14)
0·56
440 (1073)
130·2 (123·4–137·3)
1·06 (0·96–1·16)
0·24
1437 (1036)
21·1 (18·5–24·1)
1·11 (0·90–1·38)
0·33
241 (1081)
Praziquantel-matching placebo Rate per 100 person– years (95% CI) 41·0 (37·3–45·0) 138·7 (131·7–146·1) 22·3 (19·7–25·3)
Events (person– years) 437 (1070) 1340 (1035) 241 (1077)
HR (95% CI)†
p value†
40·9 (37·2–44·9)
1·00 (0·84–1·20)
0·97
129·4 (122·7–136·6)
1·07 (0·98–1·18)
0·15
22·4 (19·7–25·4)
1·00 (0·80–1·24)
0·97
Rate per 100 person– years (95% CI)
HR=hazard ratio. *Between albendazole and albendazole-matching placebo. †Between praziquantel and praziquantel-matching placebo.
Table 4: Effects of maternal anthelmintic treatment in pregnancy on incidence of malaria, diarrhoea, and pneumonia in infancy
Albendazole
Praziquantel
Effect of treatment in women with hookworm infection HR (95% CI)
Effect of treatment in women without hookworm infection HR (95% CI)
p value for interaction
Effect of treatment in women with schistosomiasis HR (95% CI)
Effect of treatment in women without schistosomiasis HR (95% CI)
p value for interaction
Malaria*
1·01 (0·78–1·31)
0·89 (0·69–1·15)
0·48
0·94 (0·62–1·41)
1·03 (0·84–1·26)
0·70
Diarrhoea*
1·10 (0·95–1·27)
1·02 (0·90–1·16)
0·44
1·23 (0·99–1·54)
1·05 (0·94–1·16)
0·18
Pneumonia*
1·08 (0·77–1·53)
1·12 (0·85–1·48)
0·88
0·94 (0·61–1·45)
1·01 (0·79–1·30)
0·79
HR=hazard ratio. *Excludes nine women with no stool result for Kato-Katz analysis.
Table 5: Effect of maternal anthelmintic treatment in pregnancy on disease incidence, by maternal helminth status
evidence was noted for any differential effect of treatment according to susceptible worm species; ORs for the effect of albendazole treatment on HIV transmission were 0·77 (0·25–2·40) in mothers with hookworm infection and 0·66 (0·27–1·60) in those without hookworm infection (interaction p=0·83). We recorded similar results for praziquantel (OR 0·75, 0·18–3·14, for mothers with schistosomiasis; 0·56, 0·24–1·29 for those without schistosomiasis; interaction p=0·73). For the secondary outcome of community-reported illness data, the overall recorded prevalences of each illness collated every 14 days throughout infancy was 13% for febrile illness, 12% for diarrhoea, and 1% for presumptive pneumonia. These data were consistent with doctor-diagnosed outcomes and showed no effects of treatment with albendazole or praziquantel on febrile illnesses (OR 1·07, 95% CI 0·97–1·18, p=0·18; and 1·03, 0·93–1·13, 0·58), diarrhoea (1·08, 0·98–1·20, 0·13; and 0·93, 0·84–1·03, 0·18), or presumptive pneumonia (0·92, 0·65–1·31, 0·66; and 0·91, 0·64–1·30, 0·62). The prevalence of asymptomatic malaria parasitaemia at 1 year was similar between treatment groups (webappendix p 5). 81 infants in the cohort died, 44 of whom were neonates, giving a neonatal mortality rate of 19·0 per 1000 livebirths (95% CI 14·2–25·5) and an infant mortality rate of 35·7 per 1000 livebirths (28·8–44·2). Maternal anthelmintic treatment had no pronounced effect on mortality (webappendix p 5). Height-for-age and weightfor-age Z scores were low compared with the WHO www.thelancet.com Vol 377 January 1, 2011
standards; however, the weight-for-height distribution was similar. Mean haemoglobin score at age 1 year was 102 g/L (SD 14 g/L). We noted no evidence for any difference in mean growth indices or mean haemoglobin between children of albendazole-treated and placebotreated mothers, or between children of praziquanteltreated and placebo-treated mothers (webappendix p 5). Numbers of serious adverse events are reported elsewhere19 and were distributed evenly between treatment groups.
Discussion In this randomised, placebo-controlled trial, we have shown that maternal anthelmintic treatment during pregnancy can have a small effect on an infant’s response to tetanus immunisation, but has no effects, either beneficial or detrimental, on the occurrence of infectious diseases during infancy, infant mortality, or growth and anaemia outcomes at 1 year of age. One dose of albendazole was effective for the treatment of hookworm and A lumbricoides infections, and praziquantel was effective for the treatment of schistosomiasis. At delivery, the prevalence of hookworm infections had decreased to 5% in the albendazole group, and remained at 45% in the placebo group; the prevalence of S mansoni infection had decreased to 5% in the praziquantel group, and remained at 21% in the placebo group.37 Thus we would have expected to see any noticeable effects of the removal of these maternal helminth infections. A possible source of imprecision is that the Kato-Katz method has suboptimum 59
Articles
sensitivity for diagnosis of intestinal helminth infections when one stool sample is used.38,39 In this study, the sensitivity for detecting both hookworm infection and schistosomiasis was greater when three stool samples were taken than when only one was taken. As expected, one dose of albendazole was not effective for treatment of S stercoralis, T trichiura, or M perstans.37 Albendazole treatment in pregnancy was associated with reduced interleukin-5 and interleukin-13 responses to tetanus toxoid in infants of mothers with hookworm infection. These results should be interpreted with caution because of the large number of statistical tests done, but they accord with earlier findings40,41 on effects of helminths on type-2 responses to tetanus immunisation in adults; by contrast with these previous studies,40,41 we recorded no reciprocal effect on interferon-γ response. Such responses will probably not change the effectiveness of tetanus immunisation, because this immunisation depends on the production of toxin-neutralising IgG antibodies, which was not affected by maternal treatment. However, the patterns recorded lend support to the hypothesis that maternal helminths promote a type-2 bias in an infant’s response to unrelated antigens and that anthelmintic treatment modifies this effect. These findings suggest that the effect of anthelmintic treatment can be transmitted in utero. Neither intervention showed any overall effect on response to BCG immunisation in infants. Our preliminary study32 had suggested an unexpected positive association between maternal hookworm infection and interferon-γ responses to cCFP in infants. In the present, larger study, interferon-γ responses were higher in infants of mothers with hookworm infection, and albendazole treatment was associated with a weak reduction in interferon-γ responses in infants of women with hookworm infection. These findings contrast with our initial hypothesis that maternal hookworm infection would inhibit the Th1 response and that albendazole treatment would reverse this effect. The absence of effect of praziquantel treatment was also unexpected, in view of Malhotra and colleagues’ finding17 that infants who are sensitised to schistosome or filarial antigens in utero had reduced interferon-γ and increased interleukin-5 responses to neonatal BCG immunisation. However, all infants in Malhotra and colleagues’ study were exposed to maternal helminth infections, and comparisons were made between those not sensitised or sensitised to helminth antigens.17 The idea that single-dose praziquantel treatment completely removes the likelihood of fetal sensitisation to schistosome antigens might be too simplistic, because worm and egg antigens can take some time to be cleared from maternal tissues. Moreover, the killing of adult schistosomes results in release of worm antigen into the circulation and might have complex effects on the fetus, dependent on placental transfer of antigens or antibodies and the stage of development of the fetal immune system.41,42 60
The recorded effects on an infant’s response to tetanus immunisation contrasted with the absence of effects seen for measles and BCG. A possible explanation is that tetanus immunisation given to women during pregnancy could have led to the priming of an infant’s antitetanus response in utero,42 and the profile of the primed response could have been affected by concomitant exposure to a maternal helminth infection, which depended on whether or not an infant’s mother received treatment with albendazole. In keeping with the broad absence of effect of maternal anthelmintic treatment on an infant’s response to vaccination, we recorded no effect of either treatment for the co-primary outcome of infectious disease incidence during infancy. A possible source of imprecision is that data for doctor-diagnosed disease incidence were obtained passively, and therefore rates are likely to have been underestimated. However, 90% of infants attended the clinic for illness at least once, and underestimation would be non-differential between treatment groups; therefore this imprecision is unlikely to have affected our findings. Moreover, results from communityreported illness events (which are less likely to be underreported), and malaria parasitaemia at 1 year of age (which is independent of reporting biases) were both consistent with the recorded absence of treatment effect. We also detected no evidence that anthelmintic treatment affected vertical transmission of HIV, although power to detect an effect was restricted. The only other study43 to investigate the relation between anthelmintic treatment during pregnancy and post-neonatal outcomes, undertaken in Nepal, showed that treatment with albendazole during pregnancy was associated with a pronounced reduction in infant mortality at 6 months. We recorded no such effect, although the number of deaths in the cohort was small so the possibility of a reduction cannot be ruled out. The Nepal study was not randomised and recorded a 41% reduction in 6-month mortality associated with two doses of albendazole but no effect on infant mortality for one dose. The researchers speculated that the greater benefit of two doses might result from prevention of reinfection between treatment and delivery, but this theory would not explain the contrast with our findings, because only 5% of women who received albendazole in our study had hookworm infection at delivery.19 Our analyses needed many statistical tests, especially for the measurement of cytokine responses to immunisation, because we aimed to assess overall response profiles rather than concentrations of only one cytokine. Consistency between all cytokine responses was expected, and can be interpreted as compelling evidence of an effect, compared with isolated change in an individual response. Therefore we did not increase confidence limits or use methods that assume independence of tests to formally adjust p values. In the event, very few outcomes showed any evidence of association at a p value less than 0·05 so www.thelancet.com Vol 377 January 1, 2011
Articles
absence of adjustment for multiplicity is unlikely to have affected the interpretation of our results. A strength of this study was the randomised intervention design; effects of helminth infections reported in observational studies could have been affected by confounding factors. Our findings do not show that exposure to maternal helminth infections in utero has no important effect on the development of the fetal immune system. Some observational studies16,17 do suggest such an effect. Rather, our findings suggest that one effective anthelmintic intervention given in the second or third trimester of pregnancy is insufficient to alter any effect of maternal worms on vaccine and infectious disease outcomes in infancy. This study examined only effects of maternal anthelmintic treatment on vaccine responses and disease incidence in newborn children. Further studies are needed to assess the effects of anthelmintic treatment on these outcomes in the individual who receives the treatment. Our results are generalisable to areas with high prevalence but low intensity of helminth infection in young adults, which is a common pattern in areas that are endemic for helminth infection. Findings might differ in populations with higher infection intensities. These results suggest that, in settings such as Entebbe Municipality and Katabi subcounty, single-dose anthelmintic treatment during pregnancy has no benefit for an infant’s response to immunisation, or for their health and development. We have previously reported a similar absence of effect on maternal and perinatal outcomes.37 These results contrast with the expected benefits of routine antenatal anthelmintic treatment recently advocated,44 and the value of such a policy may need to be reviewed. The study cohort is being followed up to establish whether these conclusions are still valid up to when children are aged 5 years, and to relate effects on the cytokine response to BCG immunisation to incidence of M tuberculosis infection and disease.18 Contributors AME had the idea for, designed, and led the study. PAM and PBN contributed to sample processing; PAM did all the cytokine assays. DK led the work on measles antibodies, and AN led the work on tetanus antibodies, under the supervision of AME and RT. JNd led the clinical component of the study. JK-L led the virological investigations with a PCR assay developed by BN. MN, HM, MA, NL, MKiz, RK, JNa, and CA contributed to recruitment and follow-up of participants and to clinical care. LM, PWW, HA, and JNd contributed to data management and analysis. MKih and JB did parasitological investigations. ELW did the statistical analysis. ELW and AME drafted the report, with contributions to interpretation of the results and preparation of the manuscript from JNd and HM. JNd, PBN, MM, and JAGW contributed to the design and implementation of the study. ELW, PAM, JNd, DK, AN, and JK-L contributed equally to the work. All authors reviewed the final paper. Conflicts of interest JAGW is now a member of staff with the Wellcome Trust, the funders of the study. His role in the initial design and implementation of the study preceded his appointment at the Wellcome Trust. He has had no role in the study since his appointment. All other authors declare that they have no conflicts of interest.
www.thelancet.com Vol 377 January 1, 2011
Acknowledgments The study was funded by Wellcome Trust grant numbers 064693 and 079110; Medical Research Council (MRC) staff were supported by MRC programme grant E743; ELW was supported by the UK MRC; and HM was supported in part by a Wellcome Trust PhD studentship, grant number 074791. Mycobacterial antigens were provided through the National Institutes of Health contract NOI-AI-25147. We thank all staff and participants of the Entebbe Mother and Baby Study, the midwives of the Entebbe Hospital Maternity Department, the community field team in Entebbe and Katabi, and the staff of the Clinical Diagnostic Services Laboratory at the MRC/UVRI Uganda Research Unit on AIDS. We thank Richard Hayes for helpful comments on the manuscript. References 1 Bryce J, Boschi-Pinto C, Shibuya K, Black RE, WHO Child Health Epidemiology Reference Group. WHO estimates of the causes of death in children. Lancet 2005; 365: 1147–52. 2 Labeaud AD, Malhotra I, King MJ, King CL, King CH. Do antenatal parasite infections devalue childhood vaccination? PLoS Negl Trop Dis 2009; 3: e442. 3 WHO. Global reductions in measles mortality 2000–2008 and the risk of measles resurgence. Wkly Epidemiol Rec 2009; 84: 509–16. 4 Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC. Consensus statement. Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country—WHO Global Surveillance and Monitoring Project. JAMA 1999; 282: 677–86. 5 Fine PE. Variation in protection by BCG: implications of and for heterologous immunity. Lancet 1995; 346: 1339–45. 6 Colditz GA, Berkey CS, Mosteller F, et al. The efficacy of bacillus Calmette-Guerin vaccination of newborns and infants in the prevention of tuberculosis: meta-analyses of the published literature. Pediatrics 1995; 96: 29–35. 7 Hotez PJ, Fenwick A, Savioli L, Molyneux DH. Rescuing the bottom billion through control of neglected tropical diseases. Lancet 2009; 373: 1570–75. 8 Brooker S, Akhwale W, Pullan R, et al. Epidemiology of plasmodium-helminth co-infection in Africa: populations at risk, potential impact on anemia, and prospects for combining control. Am J Trop Med Hyg 2007; 77: 88–98. 9 Brooker S, Clements AC, Bundy DA. Global epidemiology, ecology and control of soil-transmitted helminth infections. Adv Parasitol 2006; 62: 221–61. 10 Bundy D, Sher A, Michael E. Good worms or bad worms: do worm infections affect the epidemiological patterns of other diseases? Parasitol Today 2000; 16: 273–74. 11 Abbas AK, Murphy KM, Sher A. Functional diversity of helper T lymphocytes. Nature 1996; 383: 787–93. 12 Mosmann TR, Cherwinski H, Bond MW, Giedlin MA, Coffman RL. Two types of murine helper T cell clone: I—definition according to profiles of lymphokine activities and secreted proteins. J Immunol 1986; 136: 2348–57. 13 Bentwich Z, Kalinkovich A, Weisman Z, Borkow G, Beyers N, Beyers AD. Can eradication of helminthic infections change the face of AIDS and tuberculosis? Immunol Today 1999; 20: 485–87. 14 Maizels RM, Pearce EJ, Artis D, Yazdanbakhsh M, Wynn TA. Regulation of pathogenesis and immunity in helminth infections. J Exp Med 2009; 206: 2059–66. 15 Hartgers FC, Yazdanbakhsh M. Co-infection of helminths and malaria: modulation of the immune responses to malaria. Parasite Immunol 2006; 28: 497–506. 16 Malhotra I, Ouma JH, Wamachi A, et al. Influence of maternal filariasis on childhood infection and immunity to Wuchereria bancrofti in Kenya. Infect Immun 2003; 71: 5231–37. 17 Malhotra I, Mungai P, Wamachi A, et al. Helminth- and Bacillus Calmette-Guerin-induced immunity in children sensitized in utero to filariasis and schistosomiasis. J Immunol 1999; 162: 6843–48. 18 Elliott AM, Kizza M, Quigley MA, et al. The impact of helminths on the response to immunization and on the incidence of infection and disease in childhood in Uganda: design of a randomized, double-blind, placebo-controlled, factorial trial of deworming interventions delivered in pregnancy and early childhood [ISRCTN32849447]. Clin Trials 2007; 4: 42–57.
61
Articles
19
20
21
22
23
24
25
26
27 28
29
30
31
32
62
Woodburn PW, Muhangi L, Hillier S, et al. Risk factors for helminth, malaria, and HIV infection in pregnancy in Entebbe, Uganda. PLoS Negl Trop Dis 2009; 3: e473. Uganda Bureau of Statistics (UBOS) and Macro International Inc. Uganda Demographic and Health Survey 2006. Maryland: UBOS and Macro International Inc, 2007. Tann CJ, Kizza M, Morison L, et al. Use of antenatal services and delivery care in Entebbe, Uganda: a community survey. BMC Pregnancy Childbirth 2007; 7: 23. Guay LA, Musoke P, Fleming T, et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet 1999; 354: 795–802. Elliott AM, Hurst TJ, Balyeku MN, et al. The immune response to Mycobacterium tuberculosis in HIV-infected and uninfected adults in Uganda: application of a whole blood cytokine assay in an epidemiological study. Int J Tuberc Lung Dis 1999; 3: 239–47. Cousens S, Feachem R, Kirkwood B, Mertens T, Smith P. Case-control studies of childhood diarrhoea: I minimising bias. Geneva: World Health Organization, 1988; WHO/CDD/EDP/88.2. WHO Programme for the Control of Acute Respiratory Infections. Acute respiratory infections in children: case management in small hospitals in developing countries. Geneva: World Health Organisation, 1990; WHO/ARI/90.5. Morley D. Severe measles in Africa. Paediatric priorites in the developing world. In: Morley D, ed. Paediatric priorities in the developing world. London: Butterworths, 1973: 207–30. Khan EA, Starke JR. Diagnosis of tuberculosis in children: increased need for better methods. Emerg Infect Dis 1995; 1: 115–23. Katz N, Chaves A, Pellegrino J. A simple device for quantitative stool thick-smear technique in Schistosomiasis mansoni. Rev Inst Med Trop Sao Paulo 1972; 14: 397–400. Friend J. Helminths. In: Collee J, Frase A, Marmion B, Simmons A, eds. Mackie and McCartney Practical Medical Microbiology. Edinburgh: Churchill Livingstone, 1996: 757–78 Muhangi L, Woodburn P, Omara M, et al. Associations between mild-to-moderate anaemia in pregnancy and helminth, malaria and HIV infection in Entebbe, Uganda. Trans R Soc Trop Med Hyg 2007; 101: 899–907. Melrose WD, Turner PF, Pisters P, Turner B. An improved Knott’s concentration test for the detection of microfilariae. Trans R Soc Trop Med Hyg 2000; 94: 176. Elliott AM, Namujju PB, Mawa PA, et al. A randomised controlled trial of the effects of albendazole in pregnancy on maternal responses to mycobacterial antigens and infant responses to Bacille Calmette-Guerin (BCG) immunisation [ISRCTN32849447]. BMC Infect Dis 2005; 5: 115.
33
34
35
36
37
38
39
40
41
42
43
44
Snow RW, Craig M, Deichmann U, Marsh K. Estimating mortality, morbidity and disability due to malaria among Africa’s nonpregnant population. Bull World Health Organ 1999; 77: 624–40. Ross DA, Dollimore N, Smith PG, et al. Vitamin A supplementation in northern Ghana: effects on clinic attendances, hospital admissions, and child mortality. Lancet 1993; 342: 7–12. Greenwood B. Epidemiology of acute lower respiratory tract infections, especially those due to Haemophilus influenzae type b, in The Gambia, west Africa. J Infect Dis 1992; 165 (suppl 1): S26–28. McGuinness D, Bennett S, Riley E. Statistical analysis of highly skewed immune response data. J Immunol Methods 1997; 201: 99–114. Ndibazza J, Muhangi L, Akishule D, et al. Effects of deworming during pregnancy on maternal and perinatal outcomes in Entebbe, Uganda: a randomized controlled trial. Clin Infect Dis 2010; 50: 531–40. Knopp S, Rinaldi L, Khamis IS, et al. A single FLOTAC is more sensitive than triplicate Kato-Katz for the diagnosis of low-intensity soil-transmitted helminth infections. Trans R Soc Trop Med Hyg 2009; 103: 347–54. Utzinger J, Booth M, N’Goran EK, Muller I, Tanner M, Lengeler C. Relative contribution of day-to-day and intra-specimen variation in faecal egg counts of Schistosoma mansoni before and after treatment with praziquantel. Parasitology 2001; 122: 537–44. Nookala S, Srinivasan S, Kaliraj P, Narayanan RB, Nutman TB. Impairment of tetanus-specific cellular and humoral responses following tetanus vaccination in human lymphatic filariasis. Infect Immun 2004; 72: 2598–604. Sabin EA, Araujo MI, Carvalho EM, Pearce EJ. Impairment of tetanus toxoid-specific Th1-like immune responses in humans infected with Schistosoma mansoni. J Infect Dis 1996; 173: 269–72. Gill TJ 3rd, Repetti CF, Metlay LA, et al. Transplacental immunization of the human fetus to tetanus by immunization of the mother. J Clin Invest 1983; 72: 987–96. Christian P, Khatry SK, West KP Jr. Antenatal anthelmintic treatment, birthweight, and infant survival in rural Nepal. Lancet 2004; 364: 981–83. Allen HE, Crompton DW, de Silva N, LoVerde PT, Olds GR. New policies for using anthelmintics in high risk groups. Trends Parasitol 2002; 18: 381–82.
www.thelancet.com Vol 377 January 1, 2011
Articles
Clostridium difficile infection in Europe: a hospital-based survey Martijn P Bauer, Daan W Notermans, Birgit H B van Benthem, Jon S Brazier, Mark H Wilcox, Maja Rupnik, Dominique L Monnet, Jaap T van Dissel, Ed J Kuijper, for the ECDIS Study Group*
Summary Background Little is known about the extent of Clostridium difficile infection in Europe. Our aim was to obtain a more complete overview of C difficile infection in Europe and build capacity for diagnosis and surveillance. Methods We set up a network of 106 laboratories in 34 European countries. In November, 2008, one to six hospitals per country, relative to population size, tested stool samples of patients with suspected C difficile infection or diarrhoea that developed 3 or more days after hospital admission. A case was defined when, subsequently, toxins were identified in stool samples. Detailed clinical data and stool isolates were collected for the first ten cases per hospital. After 3 months, clinical data were followed up. Findings The incidence of C difficile infection varied across hospitals (weighted mean 4·1 per 10 000 patient-days per hospital, range 0·0–36·3). Detailed information was obtained for 509 patients. For 389 of these patients, isolates were available for characterisation. 65 different PCR ribotypes were identified, of which 014/020 (61 patients [16%]), 001 (37 [9%]), and 078 (31 [8%]) were the most prevalent. The prevalence of PCR-ribotype 027 was 5%. Most patients had a previously identified risk profile of old age, comorbidity, and recent antibiotic use. At follow up, 101 (22%) of 455 patients had died, and C difficile infection played a part in 40 (40%) of deaths. After adjustment for potential confounders, an age of 65 years or older (adjusted odds ratio 3·26, 95% CI 1·08–9·78; p=0·026), and infection by PCR-ribotypes 018 (6·19, 1·28–29·81; p=0·023) and 056 (13·01; 1·14–148·26; p=0·039) were significantly associated with complicated disease outcome. Interpretation PCR ribotypes other than 027 are prevalent in European hospitals. The data emphasise the importance of multicountry surveillance to detect and control C difficile infection in Europe. Funding European Centre for Disease Prevention and Control.
Introduction Clostridium difficile infection is prevalent in health-care facilities throughout the developed world, but also presents as large outbreaks. Less often, it is acquired in the community from an unknown source. It characteristically occurs in elderly patients with comorbidity in whom the intestinal flora has been disrupted by previous use of antibiotics.1,2 Since early 2003, increasing rates of C difficile infection have been reported in Canada and the USA, with a larger proportion of severe and recurrent cases occuring in these countries than previously reported. The raised incidence and virulence of such infection have partly been explained by the spread of fluoroquinoloneresistant strains belonging to the PCR-ribotype 027.3–5 In addition to the usual toxins A and B, these fluoroquinoloneresistant strains produce a binary toxin, with a hitherto uncertain pathogenic significance.1–6 In Europe, PCRribotype 027 was first reported in 2005 in England and shortly thereafter in the Netherlands.7,8 Subsequently, epidemics of C difficile infection caused by PCRribotype 027 have been recognised in hospitals in many European countries.9 The attention given to this infection, diagnostic procedures in hospitals, presence and methodology of national surveillance, and availability of typing vary widely www.thelancet.com Vol 377 January 1, 2011
across Europe, which hampers comparisons between countries.9,10 We did this study to obtain a more complete overview of the situation in Europe and build capacity for diagnosis and surveillance of C difficile infection both nationally and Europe-wide.
Methods Study design and patients With support from the European Centre for Disease Prevention and Control, we appointed national coordinators for 34 European countries (including 27 member states, three candidate states, and four European-Free-Trade-Association countries) who selected hospitals in each country, relative to the country’s population size. No randomisation was used for this selection. The aim was to include one hospital for countries with fewer than two million inhabitants, three for those with between two and 20 million inhabitants, and five for those with more than 20 million inhabitants, with a balance between academic and non-academic institutions. A study protocol noting all procedures was distributed to national coordinators and coordinators in all hospitals. Hospitals and laboratories completed a webbased questionnaire (webappendix pp 1–5) with epidemiological data, including numbers of patient-days,
Lancet 2011; 377: 63–73 Published Online November 16, 2010 DOI:10.1016/S01406736(10)61266-4 See Comment page 8 *Members listed at end of paper Centre for Infectious Disease Control Netherlands, National Institute for Public Health and the Environment, Bilthoven, Netherlands (M P Bauer MD, D W Notermans PhD, B H B van Benthem PhD); Centre for Infectious Diseases, Leiden University Medical Centre, Leiden, Netherlands (M P Bauer, E J Kuijper PhD, Prof J T van Dissel PhD); Anaerobe Reference Laboratory, National Public Health Service for Wales, Microbiology Cardiff University Hospital of Wales, Cardiff, UK (J S Brazier PhD); Department of Microbiology, Old Medical School, Leeds General Infirmary and University of Leeds, Leeds, UK (Prof M H Wilcox MD); Institute of Public Health Maribor, Centre for Microbiology, Maribor, Slovenia (Prof M Rupnik PhD); and Scientific Advice Unit, European Centre for Disease Prevention and Control, Stockholm, Sweden (D L Monnet PhD) Correspondence to: Dr Martijn P Bauer, Leiden University Medical Centre, Infectious Diseases, Albinusdreef 2, Leiden, 2300 RC, Netherlands
[email protected]
See Online for webappendix
63
Articles
admissions, and stool samples tested for C difficile infection in November, 2008, and technical data such as assays and culture methods used.
Procedures Hospitals were asked to test for C difficile infection in outpatients and inpatients suspected of having the infection by their treating physician and all inpatients who developed diarrhoea 3 days or more after admission.
Number of toxin-positive cases/number of patients tested
Number of Number of participating patients tested per hospitals* 10 000 patientdays
Weighted mean health-careassociated C difficile infection incidence rate per hospital (minimum to maximum range)†
Per 10 000 patient-days
Clinical grounds on which to suspect recurrence were left to the attending physicians’ judgment, who could use the definition of C difficile infection according to the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) treatment guidance for C difficile infection.11 Only patients aged 2 years or older were included in the study. Patients with suspected C difficile infection and diarrhoea, whose stool samples were positive for toxin A, B, or both (EIA, cytotoxicity test, or PCR) or revealed
Percentage of healthcare-associated C difficile infection cases in health-care-associated and communityassociated C difficile infections
Toxin tests used (number of Number of hospitals) complicated cases/number of cases with available data (%)
Per 10 000 admissions
Austria
53/330 (16%)
52
3
7·5 (4·3–10·9)
36 (20–46)
92%
4/26 (15%)
A+B (2); A+B and Cu (1)
Belgium
16/283 (6%)
55
3
2·8 (0·0–6·2)
19 (0–39)
91%
0/11 (0%)
A+B (1); Cy and A+B (1); A (1)
Bulgaria
2/9 (22%)
3
Croatia
22/197 (11%)
41
Cyprus
3
0·6 (0·0–2·1)
3 (0–10)
100%
1/1 (100%)
A+B (3)
3 (2)
0·7 (0·5–2·1)
6 (4–20)
18%
1/14 (7%)
A+B (2)
1/28 (4%)
34
1
1·2
5
100%
0/1 (0%)
A+B (1)
Czech Republic
10/152 (7%)
17
3
1·1 (0·0–1·3)
7 (0–9)
100%
2/7 (29%)
A+B (3)
Denmark
28/330 (8%)
74
3
5·5 (4·4–9·6)
18 (10–25)
88%
1/19 (5%)
A+B (1); Cu (2)
Finland
52/351 (15%)
141
3
19·1 (8·7–28·5)
80 (30–132)
91%
2/22 (9%)
A+B and Cu (1); Cu (1); A&B (1)
France
37/626 (6%)
42
5 (4)
2·1 (1·0–3·1)
15 (6–27)
84%
4/34 (12%)
A+B (2); Cu (1); Cy (1)
Germany
93/602 (15%)
72
6 (5)
7·4 (2·9–16·4)
60 (25–276)
91%
2/24 (8%)
A+B (3); Cu (1); Cy (1)
Greece
21/288 (9%)
60
3
3·7 (1·3–4·9)
29 (9–44)
84%
0/17 (0%)
A+B (3)
Hungary
22/333 (7%)
38
3
2·0 (0·4–3·9)
9 (1–23)
68%
1/25 (4%)
A+B (3)
··
1
100%
0/6 (0%)
··
Iceland
6/0
··
··
Ireland
38/493 (8%)
94
3
7·3 (6·5–7·9)
63 (39–92)
100%
5/21 (24%)
A+B (3)
Italy
57/533 (11%)
39
5
3·6 (0·4–5·8)
22 (2–61)
85%
5/18 (28%)
AB (2), GluD and A+B (1); Cy (1)
Latvia
13/64 (20%)
10
3
1·9 (0·0–2·8)
13 (0–20)
91%
0/13 (0%)
A (2); A+B (1)
Luxembourg
0/28 (0%)
49
1
0·0
NA
0
A+B A+B (2); Cy (1)
0
Netherlands
18/309 (6%)
69
3
4·0 (2·3–8·5)
23 (13–43)
100%
1/15 (9%)
Norway
37/241 (15%)
50
3
7·6 (0·4–16·5)
56 (3–229)
100%
1/16 (6%)
A+B (3)
Poland
102/263 (39%)
45
3
12·5 (3·8–36·3)
76 (29–189)
79%
1/11 (9%)
A+B (2); Cu (1)
14/158 (9%)
45
3 (2)
2·6 (1·9–8·2)
13 (13–14)
3
5 (1)
0·3
Portugal Romania
1/11 (9)
2
86%
0/10 (0%)
A+B (3)
100%
0/1 (0%)
A+B (2) A (1); Cu (1)
Slovakia
10/91 (11%)
16
3 (2)
1·4 (0·0–2·1)
11 (0–15)
71%
0/5 (0%)
Slovenia
24/123 (20%)
17
3 (2)
2·8 (1·5–3·2)
19 (10–23)
67%
1/10 (10%)
A+B (2)
Spain
46/485 (9%)
45
5
4·3 (0·0–16·7)
30 (0–47)
100%
5/28 (18%)
A+B (2); Cu (1); A+B and Cy and Cu (1); A+B and Cu (1)
Sweden
69/430 (16%)
74
3
9·8 (6·3–15·7)
50 (28–71)
86%
2/30 (7%)
A+B (2); Cy (1)
Switzerland
16/150 (11%)
45
3
4·8 (0·0–7·5)
50 (0–84)
100%
0/12 (0%)
A+B (2); Cu (1)
Turkey
4
5
0·0 (0·0–0·6)
20%
0/4 (0%)
A+B (3); A (1)
United Kingdom
164/1695 (10%)
4/105 (4%)
115
6
10·6 (6·7–30·3)
50 (44–135)
92%
5/40 (13%)
A+B (3); Cy (3)
Total
NA
NA
4·1 (0·0–36·3)
23 (0–276)
NA
97 (87)
0 (0–4)
44/442 (10%)
NA
A+B=enzyme immunoassay for C difficile toxin A and B. A=enzyme immunoassay for C difficile toxin A only. Cu=toxigenic culture. Cy=cytotoxicity test. GluD=enzyme immunoassay for C difficile-specific glutamate dehydrogenase. NA=not applicable. ··=data not available. *Number of hospitals on which incidence data are based is shown in parentheses. The remaining hospitals did not provide denominator data. †Weight factor for weighted-mean incidence per 10 000 patient-days=number of patient-days; weight factor for weighted-mean incidence per 10 000 admissions=number of admissions. The UK and Germany were each granted one extra hospital. In Poland, three hospitals rather than five were recruited. No hospitals were recruited in Lithuania, and one was recruited in Malta. From Estonia, Liechtenstein, and the Former Yugoslav Republic of Macedonia no data or isolates were received.
Table 1: Summary of Clostridium difficile infection in countries and hospitals
64
www.thelancet.com Vol 377 January 1, 2011
Articles
the presence of toxin-producing C difficile were defined as having C difficile infection. A web-based questionnaire (webappendix pp 5–9) was used to gather additional information about demography, clinical data, and risk factors associated with the infection in the first patients to be diagnosed, with a maximum of ten patients included per participating hospital. If patients had episodes of C difficile infection in the previous 8 weeks, they were reported as having recurrent disease at inclusion. Stool samples from the first ten patients were cultured for C difficile according to local protocols, and the isolates were sent to a central laboratory (Leiden University Medical Centre, Leiden, Netherlands) for further characterisation. 3 months after diagnosis, follow-up clinical data were obatined as part of the web-based questionnaire, including overall mortality, mortality attributable to C difficile infection, colectomy, intensive-care-unit (ICU) admission, and recurrences during follow-up. Clinical grounds on which to suspect recurrence were left to the attending physicians’ judgment, who could use the definition of recurrence according to the ESCMID treatment guidance for C difficile infection.11 All patients suspected of recurrence, who had toxin-positive-stool samples, were reported as having recurrence. No attempt was made to differentiate between relapses and reinfections. Identification of C difficile was confirmed by an inhouse PCR test for the glutamate dehydrogenase gene specific to C difficile.12 Isolates were further characterised by PCR ribotyping.13 Since PCR-ribotypes 014 and 020 are nearly identical and differ only by one band on a specific agarose-gel electrophoresis, the types were reported together as ribotype 014/020. The presence of toxin A, toxin B, and binary toxin genes were investigated with standardised PCRs.14,15 Isolates that were difficult to type were sent to the Anaerobe Reference Laboratory in Cardiff, UK, for further characterisation by the Cardiff PCR-ribotyping library, which currently consists of more than 300 ribotypes.16 These isolates, and isolates of PCR ribotypes for which the toxinotype was unknown, were sent to the Institute of Public Health in Maribor, Slovenia, for toxinotyping.17 No attempt was made to identify more than one causative ribotype, because infection by C difficile resulting from more than one ribotype is thought to be rare. We adhered to the epidemiological recommendations as defined by the ad hoc C difficile surveillance working group.1,18 Briefly, C difficile infection is divided into healthcare-associated cases (ie, occurring in a hospital or nursing home after 48 h of admission or within 4 weeks after discharge from such a facility), community-associated cases (ie, occurring in the community, provided that the patient had not been admitted to a health-care facility in the previous 12 weeks), and an indeterminate group for infections occurring between 4 and 12 weeks after discharge from a health-care facility. Furthermore, complicated disease was defined as C difficile infection www.thelancet.com Vol 377 January 1, 2011
that contributed to or caused ICU admission or death, or led to colectomy. Severe comorbidity was defined as having a chronic-health points score over 0, as defined by the Acute Physiology, Age, Chronic Health Evaluation (APACHE) II score.19 Quinolones were classified as old quinolones (nalidixic acid, norfloxacin, ofloxacin, ciprofloxacin) and new quinolones (levofloxacin, moxifloxacin, gatifloxacin).
Statistical analysis For all hospitals, incidence rates of health-care-associated C difficile infection were obtained by dividing the number of health-care-associated occurrences in November, 2008, (extrapolated by multiplication of the proportion of healthcare-associated infection in the questionnaires with all n/ N (%) Epidemiological characteristics Female
287/509 (56%)
Age ≥65 years*
319/509 (63%)
Epidemiological association Health-care associated
408/506 (80%)
Community associated
70/506 (14%)
Indeterminate association
28/506 (6%)
Explicit request to test for infection
441/507 (87%)
Use of an antibiotic not directed at C difficile infection Any antibiotic not directed at C difficile infection
366/463 (79%)
Aminopenicillin
28/463 (6%)
Aminopenicillin–β-lactamase inhibitor combination
86/463 (19%)
Antipseudomonal penicillin–β-lactamase inhibitor combination
38/463 (8%)
Second-generation cephalosporin
60/463 (13%)
Ceftazidime
78/463 (17%)
Any cephalosporin
155/463 (34%)
Carbapenem
41/463 (9%)
Aminoglycoside
27/463 (6%)
Old quinolone
80/463 (17%)
New quinolone
29/463 (6%)
Any quinolone
104/463 (23%)
Intravenous glycopeptide
33/463 (7%)
Lincosamide
28/463 (6%)
Macrolide
27/463 (6%)
Co-trimoxazole
25/463 (5%)
Use of any antibiotic not directed at C difficile infection during previous 3 months
426/463 (92%)
Comorbidity Severe comorbidity (APACHE II CHP >0)
204/468 (44%)
Liver cirrhosis (APACHE II)
21/488 (4%)
Heart disease (APACHE II)
47/484 (10%)
Pulmonary disease (APACHE II)
54/480 (11%)
Chronic dialysis (APACHE II) Immunocompromised status (APACHE II)
30/496 (6%) 106/488 (22%)
Treatment for inflammatory bowel disease
21/492 (4%)
Episodes of infection in previous 8 weeks
68/431 (16%) (Continues on next page)
65
Articles
n/ N (%) (Continued from previous page) Disease characteristics Outpatient
56/509 (11%)
Duration of diarrhoea <1 week 1 to 3 weeks >3 weeks Diarrhoea mixed with blood at any time in previous week Fever (temperature >38·5°C) Ileus at any time in previous week Last leucocyte count in previous week ≥15×10⁹ per L† Serum creatinine rise >50% from baseline before onset of symptoms
334/461 (73%) 92/461 (20%) 35/461 (8%) 48/416 (12%) 167/446 (37%) 20/509 (4%) 122/428 (29%) 31/395 (8%)
Sigmoidoscopy or colonoscopy‡ Pseudomembranes Ulceration
7/29 (24%) 13/29§ (45%)
Imaging‡ Colonic wall thickening on CT Pericolonic fat stranding on CT Bowel distension on plain abdominal radiograph or CT
26/63 (41%) 7/63 (11%) 27/117 (23%)
Microbiological characteristics Most frequent PCR-ribotypes of toxigenic isolates 014/020
61/389 (16%)
001
37/389 (10%)
078
31/389 (8%)
018
23/389 (6%)
106
20/389 (5%)
027
19/389 (5%)
002
18/389 (5%)
012
17/389 (4%)
017
14/389 (4%)
015
13/389 (3%)
126
12/389 (3%)
023
10/389 (3%)
046
8/389 (2%)
003
7/389 (2%)
011
6/389 (2%)
053
6/389 (2%)
056
6/389 (2%)
Presence of either or both binary toxin genes in toxigenic isolates
90/389 (23%)
Toxin A negative, toxin B positive strains in toxigenic isolates
13/389 (3%)
All time periods mentioned are related to the time of collection of the stool sample. Only antibiotics that were administered to more than 5% of patients are given. APACHE II=acute physiology, age, chronic health evaluation version two. CHP=chronic health points. N=total number of patients for whom information was available. *Median 71 (IQR 56—81). †Leucocyte count distribution×10⁹ per L (11; 11–15). ‡Data apply to current episode of C difficile infection. If several procedures were done during an episode, only the first was considered. §Two patients were treated for inflammatory bowel disease.
Table 2: Characteristics of patients with Clostridium difficile infection for whom questionnaires were completed
66
cases recorded in November, 2008) by the number of patient-days in November, 2008. Health-care-associated C difficile infection incidence rates were also calculated with the total number of admissions as the denominator. Weighted mean incidence rates per hospital were calculated for each country from the incidence rates of all hospitals in that country, using the number of patientdays and the number of admissions per hospital as a weighting factor. The associations of patient and pathogen characteristics with two outcome measures (complicated infections or recurrence within the 3-month follow up) were investigated. Since patients were nested within hospitals and might be exposed to common characteristics of their hospitals that could be important determinants of outcome, we could not assume independence of observations. Therefore, we chose a two-level multilevelregression method, which takes into account within-group (hospital) and between-group relations, and allows for integration of hospital and patient variables. Since the outcome (complicated infection or recurrence) was binary, we used the logistic form of the multilevel-regression model. An odds ratio with a 95% CI was calculated for all associations between a patient or pathogen characteristic and an outcome—ie, complicated infection or recurrence. Cases in which non-toxigenic strains were cultured were classified as culture negative, since these strains were not thought to be the cause of symptoms. Many of the associations reported in the analysis could be subject to confounding. For example, an association between the acquisition of C difficile infection in a health-care facility (as opposed to the community) and a complicated outcome might be confounded by age. To adjust the odds ratios for such potential confounders, we did a multivariate analysis for a selection of variables, again using a twolevel logistic-regression model. As potential confounders, we selected variables for which a role as a confounder was biologically plausible and that were correlated to outcome with an alpha level less than 0·2, since significanceselection strategies to select for possible confounders do best at this level.20 We tested whether confounders were highly collinear (variance inflation factor >10), in which case only one of them would be introduced as a covariate in multivariate analysis. Generally, statistical significance was declared for p values less than 0·05. Data were analysed with Stata 10.1.
Role of funding source The study was funded by the European Centre for Disease Prevention and Control (ECDC) through a specific service contract (ECD.894) to the Centre for Infectious Disease Control Netherlands, National Institute for Public Health and the Environment, Bilthoven, Netherlands. The decision to submit for publication was taken by the study coordinator in the Netherlands. ECDC provided support on the study design, suggested national coordinators, and provided comments on the analysis and the final report. www.thelancet.com Vol 377 January 1, 2011
Articles
Results In total, 97 hospitals provided patients or epidemiological data, or both. Because some hospitals were unable to supply denominator data, we could not calculate incidences for all hospitals (table 1). Most hospitals were large, as judged by the number of patient-days and admissions (median number of admissions per month 2645; IQR 1808– 4257); 62 hospitals (67%) were academic hospitals. The estimated incidence of health-careassociated infection varied widely between hospitals. We calculated the proportion of health-care-associated C difficile infection by the sum of health-care-associated and community-associated infections (table 1). We tested associations between high-incidence hospitals (>10 per 10 000 patient-days) and antibiotics used by the patients in the month preceding inclusion. Use of aminopenicillins (odds ratio [OR] 2·70, 95% CI 1·17–6·22), first-generation cephalosporins (6·98, 1·83–26·62), or second-generation cephalosporins (2·40, 1·28–4·50) was significantly associated with high-incidence hospitals. 395 isolates from 73 hospitals in 26 countries were available for detailed characterisation. 65 different PCR ribotypes were identified (figure), including six new PCR ribotypes: 228, 229, 230, 231, 232, and 234. The most common PCR ribotypes were 014 and 020 (found in 19 countries), 001 (in 13 countries), and 078 (in 18 countries); PCR-ribotype 027 ranked sixth (in six countries; table 2). Some commonly encountered PCR ribotypes were identified in a few countries and their distribution suggested regional spread (figure). Among these were PCR-ribotype 106, which was reported in the UK (13 isolates), Ireland (five), and Spain (two), and PCRribotype 018, which was recorded in Italy (19), Spain (two), Austria (one), and Slovenia (one). 12 different toxinotypes were identified. Of these, toxinotype 0 was most prevalent, representing 248 (65%) of 383 isolates; toxinotype III was identified predominantly in PCR-ribotype 027 strains (19 isolates) and only in five isolates belonging to rare PCR ribotypes (075, 099, 176, and 208); toxinotype IV predominantly in PCR-ribotype 023; and toxinotype V in PCR-ribotypes 078 (30 isolates) and 126 (12); toxinotype XII fully correlated with PCR-ribotype 056. 13 (3%) isolates were C difficile-toxin-A negative and C difficile-toxin-B positive. 11 of these isolates belonged to PCR-ribotype 017 and one each to the newly identified PCR-ribotypes 232 and 234. Six (2%) isolates were non-toxigenic and were not included in further analyses. Most cases were health-care associated or community associated, leaving 6% of indeterminate association (table 2). Most patients fitted the previously established risk profile, with almost two-thirds aged 65 years or more, about two-fifths having severe comorbidity, and almost all having received antibiotics during the 3 months before their infection, most commonly cephalosporins, quinolones, and aminopenicillin–β-lactamase-inhibitor combinations (table 2). 68 (16%) of 431 patients had recurrent C difficile at inclusion. www.thelancet.com Vol 377 January 1, 2011
Data after 3-months’ follow-up were obtained for about 90% of patients (table 3). An exact number cannot be provided, since follow-up was incomplete for some patients and therefore the number of patients with followup data differs for each variable. Of the 101 patients who had died, 40 (40%) of 101 deaths were judged to be related to C difficile infection. All seven patients who died from C difficile infection as a main cause were aged 75 years or older and their infection was health-care associated. Six of them had severe comorbidity (four had pulmonary disease, three were immunocompromised, and two had heart disease). Two of these patients had a recurrent episode of infection at presentation. Two had leucocyte counts of 30×10⁹ per L or greater and two of 4×10⁹ per L or less. The strains causing these infections belonged to PCR-ribotypes 015, 018, 027 (two patients), and 056. No isolate could be obtained for two patients. An age of 65 years or older, severe pulmonary comorbidity, previous use of a new quinolone, and infection by PCR-ribotypes 027, 015, and 018 were significant risk factors for complicated infections in univariate analysis (table 4). Patients with this comorbidity were distributed evenly among all hospitals. No disease characteristic—such as duration of diarrhoea, presence of fever, or leucocyte count—was significantly associated with complicated infection nor was the presence of binary Other 001 (9·3%) 002 (4·6%) 012 (4·3%) 014/020 (15·7%) 015 (3·3%) 017 (3·5%) 018 (5·8%) 023 (2·5%) 027 (4·8%) 078 (7·8%) 106 (5·1%) 126 (3·0%)
3
14
18
19
28
16
43 15
22
12
14 6
26
1 21
12
26 6
11
27
1
1
29
2
19
3
1
Figure: Geographical distribution of Clostridium difficile PCR ribotypes in European countries with more than five typable isolates, November, 2008 Pie charts show proportion of most common PCR-ribotypes per country. The number in the centre of pie charts is the number of typed isolates in the country.
67
Articles
n/N (%) Initial episode treated with Oral metronidazole
341/477 (71%)
Intravenous metronidazole
50/472 (11%)
Oral vancomycin
89/483 (18%)
Intracolonic vancomycin ICU admissions CDI contributive CDI primary cause Colectomy for CDI Death CDI contributive CDI primary cause
1/473 (0·2%) 31/459 (7%) 6/459 (1%) 1/459 (0·2%) 3/460 (0·7%) 101/455 (22%) 33/455 (7%) 7/455 (2%)
Complicated CDI
44/442 (10%)
Recurrent CDI*
86/484 (18%)
Both complicated and recurrent CDI
10/440 (2%)
Of 491 (96%) of 509 patients, complete or partial follow-up information was available. n=characteristics of patients with Clostridium difficile for whom questionnaires were completed. N=total number studied. ICU=intensive care unit. CDI=C difficile infection. *Number of recurrences during follow-up in those patients who had recurrences: median 1; 1–3.
Table 3: Treatment and outcome (3-month follow up) characteristics of patients with Clostridium difficile infection
toxin. After correction for potential confounders, an age of 65 years or older and infection by PCR-ribotypes 018 and 056 were significantly associated with complicated infection. These PCR ribotypes were binary-toxin negative and belonged to toxinotype 0 (type 018) and XII (type 056). The seven complicated cases caused by PCR-ribotype 018 occurred in four different hospitals in two countries, and the two complicated cases caused by PCR-ribotype 056 occurred in two hospitals in two countries. An age of 65 years or older, previous use of ceftazidime, and recent episodes of C difficile infection were significantly associated with recurrences during followup in univariate analysis (table 5). After correction for potential confounders, previous use of ceftazidime and recent episodes of infection were significantly associated with recurrence. Since differences between patients with follow-up information and those without were possible, the characteristics of patients with available follow-up information about C difficile infection complications (n=442) were compared with patients for whom this information was not available (n=67). Patients without this information were more likely to be outpatients at the time of presentation (OR 1·97, 95% CI 0·98–3·97), to have community-associated infection (2·59, 1·39–4·84), and be infected by PCR-ribotype 018 (3·24, 1·20–8·73) or PCR-ribotype 106 (3·96, 1·44–10·95); they were less likely to be aged 65 years or older (0·61, CI 0·36–1·02) and to have severe comorbidity (0·56, 0·31–1·01), especially pulmonary disease (0·26, 0·06–1·10). A separate analysis in which non-complicated C difficile infection was assumed for patients with missing information resulted 68
in closely similar values for the association of PCRribotype 018 with complicated infection (5·65; 1·63–19·57). Because death or colectomy could have precluded a patient from having a recurrence, a separate analysis was done for risk factors for recurrence in only those patients who did not die or undergo a colectomy. Results of the univariate analysis mirrored the analysis for the whole group, except that previous use of intravenous glycopeptides and chronic dialysis were significantly associated with recurrence (3·28, 1·12–13·78 and 2·87, 1·02–8·14, respectively). Different cutoff values for the continuous variables age and leucocyte count, as assessed by receiver operator characteristics, did not lead to improved performance in the prediction of complicated C difficile infection.
Discussion We have shown that the incidence of C difficile infection and the distribution of causative PCR ribotypes differed greatly between hospitals in Europe; overall and attributable mortality were strikingly high. The strengths of this panEuropean study are the large number of participating countries and hospitals, and a study design with a fixed 3-month follow-up. The high follow-up rate and the fact that patients with missing follow-up were younger, were more likely to be outpatients, and had less comorbidity than patients with follow-up, minimised the risk that cases of complicated infection were missed. If all patients with missing follow-up information had had an uncomplicated course, this factor would not have affected predictors for complicated infection. This study has some limitations. First, selection of the hospitals in each country was left to the national coordinators, and the number of hospitals per country was small. Therefore, results derived from this sample of hospitals might not be representative of each country. Furthermore, some hospitals might have been selected because of outbreaks of C difficile infection, thus introducing bias. Second, there might have been differences in physician awareness of infection between hospitals and countries. We note that the frequency of testing for infection varied up to 47 times between countries (as expressed by number of patients tested per 10 000 patient-days; table 1). Additionally, because there is no consensus on optimum testing for C difficile infection, diagnostic (and culture) methods were not uniform. Third, detailed information for cases of infection was obtained only for the first ten patients enrolled in each hospital, which might have introduced bias if risk factors varied across hospitals. Furthermore, this low number might have led to under-representation of PCR ribotypes that caused outbreaks of infection in some hospitals. Results from endoscopy or CT might be biased since these examinations tend to be triggered by a more severe course of disease. The proportion of patients with severe comorbidity might be overestimated because one of www.thelancet.com Vol 377 January 1, 2011
Articles
Univariate analysis OR
Multivariate analysis
95% CI
p
OR
95% CI
p
Epidemiological characteristics Age ≥65 years
4·84
1·78–13·13
0·002
3·26*
1·08–9·78
0·035
Health-care-associated vs community-associated and indeterminate infection
3·23
0·92–11·40
0·068
4·86*
0·59–40·04
0·141
Severe comorbidity (APACHE II CHP >0)
1·17
0·57–2·40
0·666
··
··
Liver cirrhosis (APACHE II)
0·53
0·06–4·56
0·562
··
··
··
Heart disease (APACHE II)
1·71
0·62–4·76
0·302
··
··
··
Pulmonary disease (APACHE II)
2·66
1·11–6·37
0·028
Chronic dialysis (APACHE II)
0·29
0·04–2·35
0·248
··
··
··
Immunocompromised status (APACHE II)
0·92
0·39–2·17
0·850
··
··
··
··
··
··
Treatment for inflammatory bowel disease†
··
··
··
1·38*
0·48–4·02
··
0.543
Use of an antibiotic not directed at C difficile infection during previous month Aminopenicillin
2·69
0·69–10·51
0·156
2·39*
0·43–13·33
0·320
Aminopenicillin–β-lactamase inhibitor combination
1·81
0·80–4·06
0·153
1·18*
0·43–3·23
0·741
Antipseudomonal penicillin–β-lactamase inhibitor combination‡ Second-generation cephalosporin
·· 0·53
·· 0·14–1·97
··
··
··
··
0·343
··
··
··
Ceftazidime
1·34
0·52–3·46
0·546
··
··
··
Any cephalosporin
0·92
0·42–2·02
0·831
··
··
··
Carbapenem
1·29
0·42–4·00
0·657
··
··
··
Aminoglycoside
1·65
0·45–6·05
0·453
··
··
··
Old quinolone
1·41
0·57–3·53
0·459
New quinolone
3·45
1·07–11·06
0·038
··
Any quinolone
2·29
1·03–5·09
0·043
··
··
··
Intravenous glycopeptide
1·95
0·61–6·20
0·257
··
··
··
2·57*
Lincosamide
0·32
0·04–2· 79
0·303
2·69
0·80–9·00
0·108
Co-trimoxazole
0·33
0·04–2·83
0·321
··
··
··
0·77
0·27–2·19
0·621
··
··
··
4·60*
··
·· 0·163
Macrolide Episodes of infection in 8 weeks before current episode
··
·· 0·68–9·72
0·72–29·37
·· 0·107
C difficile infection characteristics Duration of diarrhoea >1 week
0·55
0·23–1·32
0·182
··
··
··
Diarrhoea mixed with blood
1·06
0·33–3·42
0·928
··
··
··
Fever (temperature >38·5ºC)
1·28
0·59–2·76
0·533
··
··
··
Ileus
2·84
0·73–11·08
0·132
··
··
··
Leucocyte count ≥15×109 per L
1·50
0·67–3·35
0·324
··
··
..
Serum creatinine rise >50%
2·33
0·63–8·63
0·205
··
··
··
Bowel distension
2·06
0·38–11·25
0·405
··
··
··
PCR-ribotype 027§
4·72
1·34–16·56
0·016
PCR-ribotype 078§
1·08
0·29–4·10
0·909
PCR-ribotype 014-020§
0·43
0·12–1·50
0·184
PCR-ribotype 015§
3·77
1·01–14·08
0·048
4·56¶
0·98–21·20
0·053
PCR-ribotype 018§
9·22
2·24–38·09
0·002
6·19¶
1·28–29·81
0·023
Microbiological characteristics 2·56¶ ·· 0·60¶
PCR-ribotype 023§
1·00
0·11–9·11
0·999
··
PCR-ribotype 056§
10·96
0·96–126
0·054
13·01¶
0·64–10·25 ·· 0·17 – 2·16
·· 1·14–148·26
Presence of either or both binary toxin genes
1·09
0·46–2·54
0·847
··
··
Toxin A negative, toxin B positive strains vs all other strains
0·69
0·08–6·08
0·739
··
··
Toxinotype III (including IIIb and IIIc) vs all other toxinotypes
3·18
0·96–10·56
0·059
1·81¶
0·48–6·75
0·184 ·· 0·433
·· 0·039 ·· ·· 0·378
OR=odds ratio. APACHE II=acute physiology, age, chronic health evaluation version II. CHP=chronic health points. ··=data not available. *Adjusted for other variables: age ≥65 years, health-care association, pulmonary disease, previous use of aminopenicillin, previous use of aminopenicillin with β-lactamase inhibitor, previous use of a new quinolone, previous use of macrolide, PCR-ribotype 027, PCR-ribotype 014/020, and PCR-ribotype 056. †No complicated Clostridium difficile infection occurred in 16 patients treated for inflammatory bowel disease versus 44 cases of complicated C difficile infection in 419 patients without inflammatory bowel disease. ‡No cases of complicated C difficile infection occurred in 34 patients who received an antipseudomonal penicillin–β-lactamase inhibitor combination versus 43 cases of complicated C difficile infection in 381 patients who did not receive drug combination. §Versus all other ribotypes. ¶Adjusted for other variables: age ≥65 years, health-care association, pulmonary disease, previous use of aminopenicillin, previous use of aminopenicillin with β-lactamase inhibitor, previous use of a new quinolone, previous use of macrolide.
Table 4: Determinants of complicated Clostridium difficile infection
www.thelancet.com Vol 377 January 1, 2011
69
Articles
increase of PCR-ribotype 078 in Europe accords with findings for the Netherlands26 and reports of PCRribotype 078 in piglets with diarrhoea in the Netherlands and Spain.27,28 Interestingly, human and animal isolates of PCR-ribotype 078 are genetically highly related, supporting the hypothesis that no interspecies barrier exists for C difficile infection due to PCR-ribotype 078.26 Research suggests that food products might play a part in interspecies transmission.29,30 In one study, patients infected with PCRribotype 078 were younger than those infected with PCR-ribotype 027, but had a similar attributable mortality.27 We could not show an association between PCRribotype 078 and complicated infection; however, patients with infection as a result of this ribotype (n=31) were more likely to have a rise in serum creatinine than were patients with other ribotypes (n=362, OR 3·20, 95% CI 1·08–9·49), and had a slightly higher mean leucocyte count. Although we emphasise that C difficile infection incidence rates of participating hospitals were not representative of national incidence rates, many hospitals with high rates of C difficile infection were from countries in northern and central Europe. Most of these countries are thought to have low antibiotic consumption per head,
five items was sufficient to declare severe comorbidity, whereas if one item was scored missing, absence of severe comorbidity could not be declared. Barbut and colleagues21 reported a mean incidence of nosocomial C difficile infection in 23 European hospitals of 2·45 per 10 000 patient-days (minimum to maximum range; 0·1–7·1), which is lower than the overall figure of 4·1 per 10 000 patient-days in our study. However, that study differed from ours in methodology. Reports from Denmark, Finland, Germany, Spain, and the UK22–25 support the impression of an increase in incidence of C difficile infection in Europe. PCR ribotypes identified by Barbut and colleagues21 differed strikingly from those we identified. In their study, among isolates from 38 hospitals in 14 countries, PCR-ribotypes 001 and 014 were the most prevalent, followed by 027 and 020. Epidemic PCRribotype 027 was less prevalent in our study. By contrast, the prevalence of PCR-ribotypes 078 and 018 was increased. The high prevalence of PCR-ribotype 018 in our study is accounted for by its high prevalence in three Italian hospitals. Barbut and colleagues21 reported that PCRribotype 078 was dominant only in Greece, whereas in our study it was the third most prevalent PCR ribotype. This
Univariate analysis OR
Multivariate analysis
95% CI
p
OR
95% CI
p
Epidemiological characteristics Age ≥65 years
1·91
1·08–3·37
0·026
1·86*
0·88–3·92
0·104
Health-care-associated versus community-associated and indeterminate infection
1·77
0·83–3·78
0·139
1·93*
0·59–6·35
0·278
Severe comorbidity (APACHE II CHP >0)
1·35
0·79–2·31
0·273
··
··
Liver cirrhosis (APACHE II)
0·50
0·11–2·33
··
··
··
Heart disease (APACHE II)
1·16
0·50–2·68
··
··
Pulmonary disease (APACHE II)
0·51
0·20–1·32
0·165
0·62*
0·20 – 1·95
0·417
Chronic dialysis (APACHE II)
2·04
0·79–5·26
0·139
2·23*
0·59 – 8·37
0·235
Immunocompromised status (APACHE II)
1·22
0·66–2·24
0·531
··
··
··
··
··
··
··
Treatment for inflammatory bowel disease†
··
··
··
.·
Use of an antibiotic not directed at C difficile infection during previous month Aminopenicillin
1·04
0·35–3·13
0·941
··
··
··
Aminopenicillin–β-lactamase inhibitor combination
1·17
0·60–2·28
0·643
··
··
··
Antipseudomonal penicillin–β-lactamase inhibitor combination
1·78
0·76–4·20
0·186
2·32*
Second-generation cephalosporin
0·62
0·26–1·43
0·261
··
Ceftazidime
2·25
1·17–4·29
0·015
2·48*
0·79 – 6·82 ·· 1·06 – 5·81
0·125 ·· 0·036
Any cephalosporin
1·11
0·63–1·94
0·721
··
··
Carbapenem
0·81
0·31–2·11
0·661
··
··
·· ··
Aminoglycoside
1·60
0·59–4·28
0·354
··
··
··
Older quinolone
1·22
0·63–2·39
0·555
··
··
··
Newer quinolone
1·60
0·57–4·26
0·368
··
··
··
Any quinolone
1·35
0·73–2·47
0·335
··
··
··
Intravenous glycopeptide
1·73
0·71–4·20
0·228
··
··
··
Lincosamide
1·78
0·64–4·96
0·271
··
··
··
Macrolide
1·03
0·35–3·02
0·952
··
··
··
Co-trimoxazole
0·15
0·02–1·18
0·071
··
··
2·15
1·10–4·22
0·025
2·26*
Episodes of C difficile infection in 8 weeks before current episode
1·03–4·96
·· 0·041
(Continues on next page)
70
www.thelancet.com Vol 377 January 1, 2011
Articles
Univariate analysis OR
Multivariate analysis
95% CI
p
OR
95% CI
p
(Continued from previous page) C difficile infection characteristics Duration of diarrhoea >1 week
1·01
0·56–1·83
0·965
··
··
··
Diarrhoea mixed with blood
0·49
0·18–1·36
0·171
··
··
··
Fever (temperature >38·5ºC)
1·17
0·71–2·06
0·572
··
··
··
Ileus
0·24
0·03–1·92
0·177
··
··
··
Leukocyte count ≥15×109 per L
0·99
0·53–1·85
0·973
··
··
··
Serum creatinine rise >50%
0·90
0·30–2·69
0·850
··
··
··
··
··
··
Pseudomembranes‡
··
··
··
Ulceration
1·12
0·06–21·17
0·941
··
··
··
Colonic wall thickening
2·24
0·50–10·01
0·290
··
··
··
Pericolonic fat stranding
3·12
0·47–20·55
0·237
··
··
··
Bowel distension
0·60
0·16–2·24
0·445
··
··
··
PCR-ribotype 027§
2·06
0·66–6·43
0·211
··
··
··
PCR-ribotype 078§
1·62
0·67–3·90
0·286
··
··
··
PCR-ribotype 014/020§
0·86
0·39–1·89
0·700
··
··
··
PCR-ribotype 015§
1·72
0·47–6·30
0·411
··
··
PCR-ribotype 018§
2·39
0·70–8·16
0·165
0·50¶
0·07–3·71
0·495
PCR-ribotype 023§
2·89
0·72–11·61
0·135
1·76¶
0·33–9·29
0·508
PCR-ribotype 056§
1·75
0·27–11·47
0·557
··
··
··
Presence of either or both binary toxin genes
1·63
0·89–2·97
0·113
··
··
··
Toxin A negative, toxin B positive strains vs all other strains
0·69
0·14–3·46
0·654
··
··
··
Toxinotype III (including IIIb and IIIc) vs all other toxinotypes
1·38
0·48–3·94
0·551
··
··
··
Microbiological characteristics
··
OR=odds ratio. APACHE II=acute physiology, age, chronic health evaluation version II. CHP=chronic health points. ··=data not available. *Adjusted for other: age ≥65 years, health-care association, pulmonary disease, chronic dialysis, previous use of antipseudomonal penicillin with β-lactamase inhibitor, previous use of ceftazidime, episodes of C difficile infection 8 weeks before current episode, PCR-ribotype 018, PCR-ribotype 023, and presence of either or both binary toxin genes. †No recurrences in 19 patients with inflammatory bowel disease versus 83 recurrences in 419 patients without inflammatory bowel disease. ‡No recurrences in seven patients with pseudombranes versus two recurrences in 21 patients without pseudomembranes. §Versus all other ribotypes. ¶Adjusted for other variables: age ≥65 years, health-care association, pulmonary disease, chronic dialysis, previous use of antipseudomonal penicillin with β-lactamase inhibitor, previous use of ceftazidime and episodes of C difficile infection 8 weeks before current episode.
Table 5: Determinants of recurrence of Clostridium difficile infection during follow-up
even during the winter-respiratory-infection season.31 Heightened awareness of C difficile infection, as shown by the number of patients tested per 10 000 patient-days, might partly account for these differences in infectionincidence rates. Differences in the severity of illness of patients in hospital or those prescribed antibiotics might be other explanations. Patients admitted to high-incidence hospitals were more likely to have received aminopenicillins and first-generation and second-generation cephalosporins than were patients admitted to low-incidence hospitals. Most risk factors for complicated or recurrent infection were consistent with those reported in previous studies. Old age,32–34 previous hospital or nursing-home admission,33 ileus,33,34 and infection by PCR-ribotype 02735 have been associated with complicated C difficile infection. The use of certain antibiotics, especially fluoroquinolones, has been associated with infection by PCR-ribotype 027, and through this association with complicated or recurrent disease.35,36 We did not find an association between the use of fluoroquinolones and complicated or recurrent disease, www.thelancet.com Vol 377 January 1, 2011
possibly because of the small number of infections resulting from PCR-ribotype 027 in our study. Alternatively, some confounding effects in earlier studies—notably data for antimicrobial prescribing in outbreak settings that might overestimate C difficile infection risk associated with specific antibiotics—were not as likely in our study. An association of PCR-ribotypes 018 and 056 with complicated infection has not been reported before. However, the number of complicated infections for which these associations were based was small. Old age32,37 and a long cumulative duration of previous episodes of C difficile infection38 have been identified as predictors of recurrent infection. We could not confirm leucocytosis33,34,37,39 as a strong predictor of complicated infection, possibly because we included leucocyte counts only from the week before the patients’ inclusion, whereas in most studies the maximum leucocyte count during the course of the illness was examined. These findings underscore the importance of local surveillance to detect and control endemic and epidemic C difficile infection. 71
Articles
Contributors The study was designed by DWN, BHBB, MHW, and EJK, with support of DLM, on behalf of ECDC, and members of European Study group of Clostridum difficile, on behalf of European Society for Clinical Microbiology and Infectious Diseases. JSB and MR were responsible for PCR ribotyping and toxinotyping of strains, respectively. MPB did the study as principle coordinator, using support of DWN as principal investigator and EJK as microbiological coordinator. DLM helped in selecting national coordinators. BHBB and JTvD supervised clinical data collection and data analysis. MPB analysed the data and wrote the first draft of the article. All authors contributed substantially to the submitted version. ECDIS study group *=national coordinator. †=local coordinator. Austria—F Allerberger. (AGES–Institut für medizinische Mikrobiologie und Hygiene, Wien). G Hartman† (Hospital Feldkirch, Feldkirch). M Hell† (University Hospital Salzburg, Salzburg). A Wechsler-Fördös† (Hospital Rudolfstiftung, Wien). Belgium: M Delmee*† (University of Louvain, Brussels). B Gordts† (St Jan Hospital, Brugge). K Laffineur† (St Luc Hospital, Namur) Bulgaria: K Ivanova* (National Ref Lab for Anaerobes, Sofia). M Marina† (National Center of Infectious & Parasitic Diseases for the Queen Giovanna University Hospital, Sofia). E Dzhigosheva† (Saint Anna-University Multi-specialization Hospital for Activ Treatment, Sofia) G Filefski† (1st Multi-specialization Hospital for Activ Treatment, Sofia). Croatia: B Matica* (Institute of Public Health Andija Stampar, Zagreb). D Golubic† (County Hospital Cakovec, Cakovec). V Punda-Polic† (Split University Hospital, Split). A Tambic Andrasevic† (University Hospital for Infectious Diseases Zagreb, Zagreb). Cyprus: P Maikanti-Charalampous* † (Nicosia General Hospital, Nicosia). D Bagazouni (Nicosia General Hospital, Nicosia). Czech Republic: O Nyč* (Hospital FN Motol Prague, Prague). L Mejzlíkova† (Hospital U svate Anny FN Brno, Brno) L Geigerova† (Hospital FN Plzen Prague, Plzen– Lichotin). Denmark: KEP Olsen* (Statens Serum Institut, Copenhagen). I Brock† (Hvidovre Hospital, Hvidovre). H Holt† (Odense Universitetshospital, Odense). M Kragh Thomsen† (Regionshospitalet Viborg, Viborg). Estonia: M Jyrna-Ellam*† (North-Estonian Regional Hospital and North Estonia Medical Centre, Tallinn). Finland: A Virolainen-Julkunen* (KTL [National Public Health Institute], Helsinki). J Mikkola† (Central Hospital of Kanta-Häme, Hämeenlinna). E Mattila† (Helsinki University Central Hospital [Meilahti], Helsinki). M Broas† (Lapland Central Hospital [Infection Control Unit], Rovaniemi). Former Yugoslav Republic of Macedonia: M Petrovska* (Institute of Microbiology and Parasitology, Skopje). France: F Barbut* (Hôpital Saint-Antoine, Paris). D Descamps† (Centre hospitalier de Béthune, Béthune). J P Canone† (Centre hospitalier de Lens, Lens). V Lalande† (CHU Saint-Antoine, Paris) L Lemée† (CHU de Rouen-Charles Nicolle, Rouen). L Cavalie† (CHU de Toulouse, Toulouse). Germany: N Kleinkauf* (Robert KochInstitut, Berlin). A Kola† (Paulinenhaus Krankenanstalt e.V, Berlin). R Knüfermann† (Paulinenhaus Krankenanstalt e.V, Berlin). M Kist† (Universitätsklinikum Freiburg, Freiburg). I Fenner† (Albertinenkrankenhaus, Hamburg). A Rodloff† (Universitätsklilnikum Leipzig, Leipzig). C von Eichel-Streiber† (Institut für Medizinische Mikrobiologie und Hygiene, Mainz). S Borgmann† (Krankenhaus Tirschenreuth, Tirschenreuth). R Michael† (Krankenhaus Tirschenreuth, Tirschenreuth). Greece: E Malamou-Lada* (General Hospital of Athens). M Orfanidou† (G. Gennimatas General Hospital of Athens, Athens). A Avlami† (Laiko General Hospital, Athens). M Kanellopoulou† (Sismanoglion General Hospital, Athens). Hungary: E Nagy* (Faculty of Medicine, University of Szeged, Szeged). M Konkoly-Thege† (St István and St László Municipal Hospital, Budapest). G Terhes† (University Hospital of Szeged, Szeged). S Lenke† (St György Hospital, Székesfehérvár). Iceland: H Hardarson*† (Landspitali University Hospital, Reykjavik). H Hardardottir (Landspitali University Hospital, Reykjavik) Ireland: F Fitzpatrick* (Health Protection Surveillance Centre, Dublin) M Skally* (Health Protection Surveillance Centre, Dublin). E Smyth† (Beaumont Hospital, Dublin). L Fenelon† (St Vincents University Hospital, Dublin). K Schaffer† (St Vincents University Hospital, Dublin). M Cormican† (University College Hospital, Galway) Italy: P Mastrantonio* (Istituto Superiore di Sanità [National Institute of Health], Rome). A Raglio† (Ospedali Riuniti di Bergamo, Bergamo). P Nicoletti† (Azienda Ospedaliera Careggi, Firenze). G Dettori† (Azienda Ospedaliera Universitaria, Parma). P Mazzella†, (Policlinico Universitario, Università Cattolica del Sacro Cuore, Roma). T Zaccaria† (Azienda
72
Ospedaliera S.Giovanni Battista di Torino, Torino). Latvia: A O Balode* (Central laboratory, Paul Stradins Clinical University Hospital, Riga). K Aksenoka† (Paul Stradins Clinical University Hospital, Riga). G Zvingelis† (Riga 1st City Hospital, Riga). M Liepins† (Riga City Hospital Gailezers, Riga). Liechtenstein: C Lee* (Labormed. Zentrum Dr. Risch, Schaan). Luxembourg: J Mossong* (Laboratoire National de Santé, Luxembourg). J Even (Laboratoire National de Santé, Luxembourg) A Hakim† (Clinique Sainte Thérèse, Luxembourg). Malta: P Caruana* (Mater Dei Hospital, Msida). Netherlands: E J Kuijper*† (Leiden University Medical Centre, Leiden). S B Debast† (Meander Medical Centre, Amersfoort). P Bloembergen† (Isala klinieken, Zwolle). C Harmanus (Leiden University Medical Centre, Leiden). I Sanders (Leiden University Medical Centre, Leiden) Norway: A Ingrebretsen*† (Rikshospitalet University Hospital, Oslo). BG Iversen* (Norwegian Institute of Public Health, Oslo). P A Jenum† (Asker and Baerum Hospital, Rud). J E Afset† (St Olav Hospital/Trondheim University Hospital, Trondheim) Poland: H Pituch* (Medical University of Warsaw, Warsaw) J Pawlowska† (Province Hospital Plock, Plock). H Pituch† (The Infant Jesus Teaching Hospital, Warsaw). H Pituch† (Public Central Clinic Hospital, Warsaw). Portugal: C Furtado* (Instituto Nacional de Saude Dr. Ricardo Jorge, Lisboa). R Guiomar Moreira* (Instituto Nacional de Saude Dr Ricardo Jorge, Lisboa); J Machado* (Instituto Nacional de Saude Dr. Ricardo Jorge, Lisboa). L Sancho† (Hospital Fernando Fonseca, Amadora). G Ribeiro† (Hospital da Universidade de Coimbra, Coimbra). H Ramos† (Hospital Geral de Santo António, Porto). Romania: D Lemeni* (Cantacuzino Institute, Bucarest). R Serban* (The Institute for Public Health Bucharest, Bucarest). M Ciocîrlan† (Fundeni Clinic Institute, Bucharest). A Rafila† (Matei Bals Institute for Infectious Diseases, Bucharest). G Lesanu† (Grigore Alexandrescu Emergency Clinical Hospital for Children, Bucharest). E Danaila† (Military Central Emergency Hospital, Bucharest). E Szekely† (Targu-Mures Emergency Hospital, Targu-Mures). Slovakia: R Melková* (Slovak Medical University, Bratislava). E Novakova† (Martinska Fakultna Nemocnica, Martin). L Glosova† (Fakultna Nemocnica Trencin, Trencin). E Novakova† (Nemocnica s Poliklinikou Zilina, Zilina). Slovenia: M Rupnik* (Institute of Public Health, Maribor). T Lejko Zupanc† (University Clinical Centre, Ljubljana). B Kotnik Kevorkijan† (University Clinical Centre, Maribor). G Lesnicar† (Hospital, Celje). Spain: E Bouza* (Hospital General Universitario Gregorio Marañón, Madrid). L Alcalá Hernández† (Hospital General Universitario Gregorio Marañón, Madrid). J Abarca† (Hospital Clínico San Carlos-Complejo Hospitalario, Madrid). A Mena Ribas† (Hospital Son Dureta-Complejo Hospitalario, Palma de Mallorca). I Sanfeliú Sala† (Corporació Sanitària Parc Taulí, Sabadell [Barcelona]). MI García García† (Hospital Universitario de Salamanca-Hospital Clínico y Hospital Virgen de la Vega, Salamanca). Sweden: T Åkerlund* (Swedish Institute of Infectious Disease Control, Solna). M Wullt† (Malmö Allmänna Sjukhus, Malmö). T Norén† (Universitetssjukhuset Örebro, Örebro). A Weintraub† (Södersjukhuset, Stockholm). C Nord† (Södersjukhuset, Stockholm). Switzerland: A F Widmer*† (University Hospital, Basel). A Widmer† (Claraspital, Basel). A Widmer† (Felix Platter Spital, Basel). Turkey: B Levent * (Refik Saydam National Hygiene Center, Ankara). S Kacar† (Yuksek Ihtisas Hospital, Ankara). G Hascelik† (Hacettepe University Medical Faculty, Ankara). B Sener† (Hacettepe University Medical Faculty, Ankara). O Azap (Baskent University Hospital, Ankara). H Arslan† (Baskent University Hospital, Ankara). M. Sinirtas† (Uludag University, Bursa). H Akalin† (Uludag University, Bursa). N Ulger† (Marmara University, Kadikoy, Istanbul). G Soyletir† (Marmara University, Kadikoy, Istanbul). M Petrovska† (Institute of Microbiology and Parasitology, Skopje) United Kingdom–England: M Wilcox*† (Leeds General Infirmary, Leeds). B Patel *† (Health Protection Agency, London; Northwick Park Hospital, Harrow; Central Middlesex Hospital, London). United Kingdom–Northern Ireland: P Rooney† (Belfast City Hospital, Belfast). United Kingdom–Scotland: C Wiuff* (Health Protection Scotland, Glasgow). A Gibb† (Royal Infirmary of Edinburgh, Edinburgh). J Coia† (Stobhill Hospital, Glasgow). European Centre for Disease Prevention and Control: A Navarro Torné (European Centre for Disease Prevention and Control, Stockholm, Sweden). P Tüll (European Centre for Disease Prevention and Control, Stockholm, Sweden). Conflicts of interest The authors declared no conflicts of interest.
www.thelancet.com Vol 377 January 1, 2011
Articles
Acknowledgments We would like to acknowledge Céline Harmanus and Ingrid Sanders for PCR-ribotyping isolates at Leiden University Medical Centre, Jeroen Alblas for support in the development of the web-based questionnaire, Jan Vandenbroucke for advice on multivariate analysis, and the laboratory technicians of all participating laboratories. The study was financed by the European Centre for Disease Prevention and Control through a specific service contract (ECD.894). References 1 Kuijper EJ, Coignard B, Tüll P; the ESCMID Study Group for Clostridium difficile (ESGCD); EU member states and the European Centre for Disease Prevention and Control (ECDC). Emergence of Clostridium difficile-associated disease in North America and Europe. Clin Microbiol Infect 2006; 12 (suppl 6): 2–18. 2 Kelly CP, LaMont JT. Clostridium difficile–more difficult than ever. N Engl J Med 2008; 359: 1932–40. 3 Warny M, Pepin J, Fang A, et al. Toxin production by an emerging strain of Clostridium difficile associated with outbreaks of severe disease in North America and Europe. Lancet 2005; 366: 1079–84. 4 McDonald LC, Killgore GE, Thompson A, et al. An epidemic, toxin gene-variant strain of Clostridium difficile. N Engl J Med 2005; 353: 2433–41. 5 Loo VG, Poirier L, Miller MA, et al. A predominantly clonal multi-institutional outbreak of Clostridium difficile-associated diarrhea with high morbidity and mortality. N Engl J Med 2005; 353: 2442–49. 6 Rupnik M, Wilcox MH, Gerding DN. Clostridium difficile infection: new developments in epidemiology and pathogenesis. Nat Rev Microbiol 2009; 7: 526–36. 7 Smith A. Outbreak of Clostridium difficile infection in an English hospital linked to hypertoxin-producing strains in Canada and the US. Euro Surveill 2005; 10: 2735. 8 Kuijper EJ, van den Berg RJ, Debast S, et al. Clostridium difficile ribotype 027, toxinotype III, the Netherlands. Emerg Infect Dis 2006; 12: 827–30. 9 Kuijper EJ, Barbut F, Brazier JS, et al. Update of Clostridium difficile infection due to PCR ribotype 027 in Europe, 2008. Euro Surveill 2008; 13: 18942. 10 Barbut F, Delmée M, Brazier JS, et al. A European survey of diagnostic methods and testing protocols for Clostridium difficile. Clin Microbiol Infect 2003: 9: 989–96. 11 Bauer MP, Kuijper EJ, van Dissel JT; European Society of Clinical Microbiology and Infectious Diseases. European Society of Clinical Microbiology and Infectious Diseases (ESCMID): treatment guidance document for Clostridium difficile infection (CDI). Clin Microbiol Infect 2009; 15: 1067–79. 12 Paltansing S, van den Berg RJ, Guseinova RA, Visser CE, van der Vorm ER, Kuijper EJ. Characteristics and incidence of Clostridium difficile-associated disease, The Netherlands, 2005. Clin Microbiol Infect 2007; 13: 1058–64. 13 Bidet P, Lalande V, Salauze B, et al. Comparison of PCR-ribotyping, arbitrarily primed PCR, and pulsed-field gel electrophoresis for typing Clostridium difficile. J Clin Microbiol 2000; 38: 2484–87. 14 Kato H, Kato N, Watanabe K, et al. Identification of toxin A-negative, toxin B-positive Clostridium difficile by PCR. J Clin Microbiol 1998; 36: 2178–82. 15 Stubbs S, Rupnik M, Gibert M, Brazier J, Duerden B, Popoff M. Production of actin-specific ADP-ribosyltransferase (binary toxin) by strains of Clostridium difficile. FEMS Microbiol Lett 2000; 186: 307–12. 16 Stubbs SL, Brazier JS, O’Neill GL, Duerden BI. PCR targeted to the 16S-23S rRNA gene intergenic spacer region of Clostridium difficile and construction of a library consisting of 116 different PCR ribotypes. J Clin Microbiol 1999; 37: 461–63. 17 Rupnik M, Avesani V, Janc M, von Eichel-Streiber C, Delmée M. A novel toxinotyping scheme and correlation of toxinotypes with serogroups of Clostridium difficile isolates. J Clin Microbiol 1998; 36: 2240–47. 18 McDonald LC, Coignard B, Dubberke E, Song X, Horan T, Kutty PK, the Ad Hoc Clostridium difficile Surveillance Working Group. Recommendations for surveillance of Clostridium difficile– associated disease. Infect Control Hosp Epidemiol 2007; 28: 140–45.
www.thelancet.com Vol 377 January 1, 2011
19
20 21
22
23
24
25 26
27
28
29
30
31
32
33
34
35
36
37
38
39
Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13: 818–29. Maldonado G, Greenland S. Simulation study of confounder-selection strategies. Am J Epidemiol 1993; 138: 923–36. Barbut F, Mastrantonio P, Delmée M, Brazier J, Kuijper E, Poxton I, on behalf of the European Study Group on Clostridium difficile (ESGCD). Prospective study of Clostridium difficile infections in Europe with phenotypic and genotypic characterisation of the isolates. Clin Microbiol Infect 2007; 13: 1048–57. Lyytikäinen O, Turunen H, Sund R, et al. Hospitalizations and deaths associated with Clostridium difficile infection, Finland, 1996–2004. Emerg Infect Dis 2009; 15: 761–65. Søes L, Mølbak K, Strøbaek S, et al. The emergence of Clostridium difficile PCR ribotype 027 in Denmark—a possible link with the increased consumption of fluoroquinolones and cephalosporins? Euro Surveill 2009; 14: 19176. Soler P, Nogareda F, Cano R. Rates of Clostridium difficile infection in patients discharged from Spanish hospitals, 1997–2005. Infect Control Hosp Epidemiol 2008; 29: 887–89. Vonberg RP, Schwab F, Gastmeier P. Clostridium difficile in discharged inpatients, Germany. Emerg Infect Dis 2007; 13: 179–80. Debast SB, van Leengoed LA, Goorhuis A, Harmanus C, Kuijper EJ, Bergwerff AA. Clostridium difficile PCR ribotype 078 toxinotype V found in diarrhoeal pigs identical to isolates from affected humans. Environ Microbiol 2009; 11: 505–11. Goorhuis A, Bakker D, Corver J, et al. Emergence of Clostridium difficile infection due to a new hypervirulent strain, polymerase chain reaction ribotype 078. Clin Infect Dis 2008; 47: 1162–70. Alvarez-Perez S, Blanco JL, et al. Prevalence of Clostridium difficile in diarrhoeic and non-diarrhoeic piglets. Vet Microbiol 2009; 137: 302–05. Songer JG, Trinh HT, Killgore GE, Thompson AD, McDonald LC, Limbago BM. Clostridium difficile in retail meat products, USA, 2007. Emerg InfectDis 2009; 15: 819–12. Jhung MA, Thompson AD, Killgore GE, et al. Toxinotype V Clostridium difficile in humans and food animals. Emerg Infect Dis 2008; 14: 1039–45. Goossens H, Ferech M, Vander Stichele R, Elseviers M, ESAC Project Group. Outpatient antibiotic use in Europe and association with resistance: a cross-national database study. Lancet 2005; 365: 579–87. Pépin J, Routhier S, Gagnon S, Brazeau I. Management and outcomes of a first recurrence of Clostridium difficile-associated disease in Quebec, Canada. Clin Infect Dis 2006; 42: 758–64. Henrich TJ, Krakower D, Bitton A, Yokoe DS. Clinical risk-factors for severe Clostridium difficile-associated disease. Emerg InfectDis 2009; 15: 415–22. Sailhammer EA, Carson K, Chang Y, et al. Fulminant Clostridium difficile colitis. Patterns of care and predictors of mortality. Arch Surg 2009; 144: 433–39. Goorhuis A, Van der Kooi T, Vaessen N, et al. Spread and epidemiology of Clostridium difficile polymerase chain reaction ribotype 027/toxinotype III in The Netherlands. Clin Infect Dis 2007; 45: 695–703. Sundram F, Guyot A, Carboo I, Green S, Lilaonitkul M, Scourfield A. Clostridium difficile ribotypes 027 and 106: clinical outcomes and risk factors. J Hosp Infect 2009; 72: 111–18. Pépin J, Alary ME, Valiquette L, et al. Increasing risk of relapse after treatment of Clostridium difficile colitis in Quebec, Canada. Clin Infect Dis 2005; 40: 1591–97. McFarland LV, Elmer GW, Surawicz CM. Breaking the cycle: treatment strategies for 163 cases of recurrent Clostridium difficile disease. Am J Gastroenterol 2002; 97: 1769–75. Moshkowitz M, Ben-Baruch E, Kline Z, Shimoni Z, Niven M, Konikoff F. Risk factors for severity and relapse of pseudomembranous colitis in an elderly population. Colorectal Dis 2007; 9: 173–77.
73
Seminar
Borderline personality disorder Falk Leichsenring, Eric Leibing, Johannes Kruse, Antonia S New, Frank Leweke Lancet 2011; 377: 74–84 Department of Psychosomatics and Psychotherapy, University of Giessen, Germany (Prof F Leichsenring DSc, Prof J Kruse MD, Prof F Leweke MD); Department of Psychosomatic Medicine and Psychotherapy, University of Goettingen, Germany (Prof E Leibing DSc); Mental Illness Research, Education and Clinical Center, James J Peters VA Medical Center, Department of Veterans Affairs and Mount Sinai School of Medicine, New York, NY, USA (A S New MD) Correspondence to: Prof Falk Leichsenring, Department of Psychosomatics and Psychotherapy, University of Giessen, Ludwigstrasse 76, 35392 Giessen, Germany
[email protected]. uni-giessen.de
Recent research findings have contributed to an improved understanding and treatment of borderline personality disorder. This disorder is characterised by severe functional impairments, a high risk of suicide, a negative effect on the course of depressive disorders, extensive use of treatment, and high costs to society. The course of this disorder is less stable than expected for personality disorders. The causes are not yet clear, but genetic factors and adverse life events seem to interact to lead to the disorder. Neurobiological research suggests that abnormalities in the frontolimbic networks are associated with many of the symptoms. Data for the effectiveness of pharmacotherapy vary and evidence is not yet robust. Specific forms of psychotherapy seem to be beneficial for at least some of the problems frequently reported in patients with borderline personality disorder. At present, there is no evidence to suggest that one specific form of psychotherapy is more effective than another. Further research is needed on the diagnosis, neurobiology, and treatment of borderline personality disorder.
Introduction Borderline personality disorder is a common mental disorder associated with high rates of suicide, severe functional impairment, high rates of comorbid mental disorders, intensive use of treatment, and high costs to society.1–5 In recent years, research findings have contributed to an improved understanding and therapy of these difficult-to-treat patients. In this Seminar, we provide an up-to-date review of recent research on the diagnosis, epidemiology, course, causes, and treatment of borderline personality disorder in adults.
a prevalence of 10% of all psychiatric outpatients and between 15% and 25% of inpatients.9,10 In a study of a non-clinical sample,7 a high rate of borderline personality disorder was reported (5·9%), indicating that many individuals with this disorder do not seek psychiatric treatment. In primary care, the prevalence reported for borderline personality disorder was four-times higher than that in the general population, suggesting that individuals with this disorder are frequent users of general medical care.11
Diagnosis Epidemiology In epidemiological studies of adults in the USA, prevalances for borderline personality disorder were between 0·5% and 5·9% in the general US population6,7 with a median prevalence of 1·35 as assessed by Torgersen and colleagues.8 There is no evidence that borderline personality disorder is more common in women.7,9 In clinical populations, borderline personality disorder is the most common personality disorder, with
Search strategy and selection criteria We searched Medline, PsycINFO, and Current Contents from their start dates to Dec 31, 2009, with the database-specific search terms such as “borderline personality disorder”, “borderline personality”, or “borderline disorder”. The search was updated in Aug 30, 2010. We mainly selected publications from the past 5 years. Studies had to meet criteria of recent Cochrane reviews on borderline personality disorder;79,107 for example, participants had to be aged 18 years or older, diagnosis of borderline personality disorder was made by use of operational criteria such as that described by DSM-IV or comparable approaches (eg, revised diagnostic interview for borderlines); and outcome measures for which reliability has been indicated were used. In studies of psychotherapy, studies had to also report a clear purpose (eg, pre-defined therapeutic benefits), define a rationale for participant inclusion or exclusion, and include a detailed description of the intervention (eg, treatment manuals or manual-like guidelines).79,107 Pharmacological interventions targeting cognitive-perceptual, affective, and impulsive-behavioural areas of borderline personality disorder were included. With respect to treatment, only randomised controlled trials (RCTs) were included. Two authors (FaL, FrL) independently extracted the necessary information from each article. In cases of disagreement, a third author was included (EL) and disagreements were resolved by consensus.
74
According to the current psychiatric classification system in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV), borderline personality disorder is characterised by a pervasive pattern of instability in interpersonal relationships, identity, impulsivity, and affect (panel).12 For a diagnosis of borderline personality disorder, at least five of the nine criteria must be met. However, suicidal tendency or self-injury are the most useful indications for a correct diagnosis,13 whereas suicidal tendency or self-injury and unstable relationships have been the most predictive features in follow-up studies.14 Accordingly, the rank ordering of criteria as most prototypical of this disorder in DSM-IV was not supported by the evidence. Further research is needed to establish whether some criteria should be given more emphasis than others.5 The nine DSM-IV criteria of borderline personality disorder seem to indicate a statistically coherent construct.15 Because factor analyses have established both a one-factor model and a threefactor model (disturbed relatedness, behavioural dysregulation, affective dysregulation), an underlying multidimensional structure of borderline personality disorder consisting of three homogeneous components might exist.5,15 With nine DSM-IV criteria and a threshold for five positive criteria of a diagnosis of borderline personality disorder, however, there are 151 theoretical possible ways of diagnosing this disorder.2,3 Thus, despite conceptual coherence, borderline personality disorder seems to be a heterogeneous diagnostic www.thelancet.com Vol 377 January 1, 2011
Seminar
category, which can include subtypes of patients, such as those with or without stress-related paranoid ideations or chronic feelings of emptiness.3
Course Although more stable than major depressive disorder, borderline personality disorder seems to be less stable over time than expected for personality disorders.5,16,17 High rates of remission were reported in both short-term and longterm follow-up studies.16 The rate of remission does not seem to be affected by major depressive disorder. By contrast, the rate of remission of major depressive disorder does seem to be significantly reduced by co-occurring borderline personality disorder.18 Affective features (eg, anger, anxiety, depression) and interpersonal features indicative of abandonment and dependency were the most prevalent and stable, whereas impulsive symptoms (eg, suicide efforts, self-injury) and interpersonal features indicative of treatment regressions were the least prevalent and consistent.19 Features of this disorder decline over time and this process seems to be partly affected by temperament.19 With fewer features over time, personality disorders such as borderline personality disorder become more correlated with each other and less distinct as individual disorders.20 Changes in personality traits (defined by the five-factor model) seem to be followed by changes in the personality disorder psychopathology of borderline personality disorder, but not vice versa.21 Traits were more unstable in patients with borderline personality disorder than in patients with other personality disorders indicating a “stable instability”,22 a term introduced by Schmideberg.23 Data from most studies show that patients with borderline personality disorder are not at higher than average risk for schizophrenia or bipolar disorder.24,25
Panel: DSM-IV-TR diagnostic criteria for 301.83 borderline personality disorder The essential feature of borderline personality disorder is that it has a pervasive pattern of instability of interpersonal relationships, self-image, and affect, with notable impulsivity that begins by early adulthood and is present in various contexts, as indicated by five (or more) of the following: 1 Frantic efforts to avoid real or imagined abandonment* 2 A pattern of unstable and intense interpersonal relationships characterised by alternating between extremes of idealisation and devaluation 3 Identity disturbance: notably and persistently unstable self-image or sense of self 4 Impulsivity in at least two areas that are potentially self-damaging (eg, spending, sex, substance misuse, reckless driving, binge eating)* 5 Recurrent suicidal gestures, or threats or self-mutilating behaviour 6 Affective instability caused by a distinct reactivity of mood (eg, intense episodic dysphoria, irritability, or anxiety usually lasting a few hours and only rarely more than a few days) 7 Chronic feelings of emptiness 8 Inappropriate intense anger or difficulty controlling anger (eg, frequent displays of temper, constant anger, recurrent physical fights) 9 Transient, stress-related paranoid ideation or severe dissociative symptoms *Note: does not include suicidal or self-mutilating behaviour covered in criterion 5. DSM-IV=Diagnostic and Statistical Manual of Mental Disorders, fourth edition.
50-times higher than in the general population.3,28 However, there are reports of lower rates.29 Patients with borderline personality disorder have more functional impairment and higher use of treatment than do patients with major depressive disorder.1,30 However, some patients with borderline personality disorder have good psychosocial functioning (25·9%), but most of them (80%) lose this level of functioning over time and do not regain it.31 Personality disorders have a negative effect on the treatment outcomes of various axis I disorders.2
Comorbidity Borderline personality disorder is regularly associated with comorbid axis I and axis II disorders.5–7 84·5% of patients with borderline personality disorder met criteria for having one or more 12-month axis I disorders, and 73·9% met criteria for another lifetime axis II disorder.6,7 Borderline personality disorder is most frequently associated with mood disorders, anxiety disorders, and disorders associated with substance misuse.5–7 With a lifetime prevalence of 39·2%, post-traumatic stress disorder is common but not universal in patients with borderline personality disorder,7 which questions the view of borderline personality disorder as a complex form of post-traumatic stress disorder.26 With respect to comorbid mental disorders, differences between female and male patients with borderline personality disorder have been reported, with disorders associated with substance misuse being more common in men and eating disorders being more common in women.7 Borderline personality disorder is associated with severe and stable functional impairment5–7,27 and characterised by a high risk of suicide.3,28 The mortality rate from suicide is between 8% and 10%, which is www.thelancet.com Vol 377 January 1, 2011
Psychosocial factors in the development of borderline personality disorder Patients with borderline personality disorder report many negative events (eg, trauma, neglect) during childhood32 and substantially more adverse events than do patients with other personality disorders.33 However, no close association between these experiences and the development of psychopathological changes in adulthood has been identified.34,35 For this reason, an interaction between biological (eg, temperamental) and psychosocial factors (eg, adverse childhood events) will probably provide the best explanation of how the condition develops,36 consistent with results from recent studies of gene/environment interaction in this disorder.37,38 The figure shows the biopsychosocial model of borderline personality disorder.
Genetic factors and neurobiology Evidence has emerged that genetic factors contribute to the development of borderline personality disorder;39–41 however, no specific genes have yet been clearly identified as causative. For dimensional representations of 75
Seminar
Genetic factors
Adverse childhood experiences
Biological factors Neurobiological structures Neurobiological dysfunctions
Psychosocial factors Personality traits (eg, neuroticism) Personality functioning (self and interpersonal)
Components of psychopathology in borderline personality disorder Affective dysregulation
Behavioural dysregulation
Disturbed relatedness
Figure: The biopsychosocial model of borderline personality disorder
borderline personality disorder traits (ie, their quantitative intensity), a moderate heritability has been reported.40 In studies of twins, heritability scores for the full diagnosis were 0·65 to 0·75,42 consistent with heritability estimates for personality disorders in general (40%–60%).43 Impulsive aggression, common in patients with borderline personality disorder, is associated with reduced serotonergic responsiveness,44 and some genes that might be linked to psychopathological changes in the disorder are involved in the serotonergic system.42 Thus, the serotonin system is the neurotransmitter system of greatest interest in these patients, and is the assumed site of action for specific selective serotonin-reuptake inhibitors.42 Data from a candidate gene study45 showed an association between a haplotype containing the short allele in the serotonin transporter gene (the serotonintransporter-linked promoter region [5-HTTLPR] in SLC6A4) and development of borderline personality disorder; however, no association between polymorphisms in 5-HTTLPR and this disorder were reported in another study.46 Presence of the short allele of 5-HTTLPR can also indicate a poor treatment response to fluoxetine in patients with borderline personality disorder.47 Polymorphisms in 5-HTTLPR might also modulate the association between serious life events and the development of impulsivity in patients.37 In a study of gene–gene interactions,48 an interaction between the Met158 allele of the catecholamine-Omethyltransferase gene (COMT) and the short allele of 5-HTTLPR was reported in patients with borderline personality disorder, but no association between the 5-HTTLPR genotype alone and development of this disorder was reported. Although the results of these studies suggest a role for polymorphisms in 5-HTTLPR in these patients, there are undoubtedly complex gene–gene and gene–environment interactions, thus making the results inconclusive.42 Another gene that has been implicated in impulsive aggression and suicidal behaviour is the tryptophan hydroxylase gene (TPH), which encodes the first enzyme 76
in serotonin biosynthesis. Two isoforms are known— TPH-1 and TPH-2. Patients with borderline personality disorder have a higher frequency of two of eight polymorphisms in TPH-2 than do controls.49 Finally, in a recent case-control study,50 patients with borderline personality disorder had a greater frequency of polymorphisms in the variable number tandem repeat of the high-activity monoamine oxidase A gene promoter allele than did healthy volunteers. Taken together, the published evidence suggests that there is an abnormality in serotonergic function, which underlies the impulsive aggressive symptoms, and that this defect might be associated with specific genetic risk factors, but the precise molecular nature of this abnormality is not yet clear. Data from early studies of dexamethasonesuppression tests51,52 have suggested that this disorder was associated with excessive production of cortisol and hypothalamic-pituitary-adrenocortical hyperactivity, although this finding was reported in many patients who had post-traumatic stress disorder with severe histories of trauma. However, in more recent studies that investigated borderline personality disorder without comorbid post-traumatic stress disorder, there was evidence of hypersuppression of cortisol, interpreted as increased feedback inhibition of the hypothalamicpituitary-adrenocortical axis.53
Neuroimaging Anatomical MRI findings Although CT studies of the brain did not detect any morphological changes in patients with borderline personality disorder, reduced volume in the amygdala has been reported in some studies with structural MRI.54,55 Excitotoxicity in the course of this disorder was discussed as a possible cause of reduced amygdala volume. Similar to these findings, no morphological changes in the amygdala were reported after the first appearance of symptoms in teenagers.56 Reduced hippocampal volumes,57 but no reductions in the amygdala volume, were observed in patients with post-traumatic stress disorder.58 Although results from a recent meta-analysis59 indicated that the amygdala volume in patients with borderline personality disorder was smaller than that in healthy controls, data from studies not included in that analysis,60–63 including some with hand-tracing of the amygdala, did not show a reduced amygdala volume in these patients. Data from these subsequent studies also highlight the importance of taking into account the effect of comorbid post-traumatic stress disorder on amygdala volume, because patients with borderline personality disorder without posttraumatic stress disorder might not have reduced amygdala volumes.60 Thus, evidence about the specificity of reductions in amygdala volume in patients with borderline personality disorder is inconsistent. Although there were no volumetric differences in the pituitary gland between patients with early-stage borderline personality disorder and controls,64 the www.thelancet.com Vol 377 January 1, 2011
Seminar
pituitary gland volume in teenagers with firstpresentation borderline personality disorder was positively correlated with the number of parasuicidal actions.65 In psychosis, the volume of the pituitary gland seems to be positively correlated with the extent of activation of the hypothalamic-pituitary-adrenocortical axis.66 Thus, an increased volume of the pituitary gland in patients with borderline personality disorder might be an indicator of increased activity of this axis.65 Several reports have revealed morphological changes in other neuroanatomical structures in patients with borderline personality disorder. A reduction in grey matter volume in the anterior cingulate gyrus, posterior cingulate gyrus,63 or hippocampus,55,67 and a reduced size of the right parietal cortex in these patients67 have been reported. Moreover, size abnormalities of the superior parietal cortices were correlated with dissociation in these patients.68 Reduced grey matter volumes in the dorsolateral cortex and in the left orbitofrontal cortex were reported in an early stage of borderline personality disorder.69 No differences in the grey matter volume between patients with borderline personality disorder and patients with other psychiatric disorders were reported.69 The authors assumed an early morphological change in the prefrontal cortex of patients with borderline personality disorder; however, this change might be nonspecific.
PET Results from most PET studies detected changes in frontal glucose metabolism in patients with borderline personality disorder. In these studies, a hypometabolism was more often reported than a hypermetabolism,70 which lends support to the assumption of a frontolimbic dysfunction in these patients. Moreover, autobiographical memories of abuse events in traumatised women without borderline personality disorder were accompanied by large increases in blood flow of the prefrontal cortex (right anterior cingulate, left orbitofrontal, right dorsolateral prefrontal cortex) and a decrease in the left dorsolateral prefrontal cortex compared with traumatised women with borderline personality disorder.60 These findings might be indicative of dysfunctions in prefrontal areas in this disorder. In a recent PET study71 of aggression provocation, patients with borderline personality disorder activated orbital cortex more than did controls, whereas controls activated regions of prefrontal cortex more typically associated with emotion control than did those with borderline personality disorder. This result lends further support to the view that there is an abnormality in prefrontal brain regions associated with emotional control function in borderline personality disorder.
Functional MRI findings The most consistent finding in patients with borderline personality disorder is increased amygdala activation compared with control individuals when viewing aversive emotion-inducing slides72 or when viewing pictures of www.thelancet.com Vol 377 January 1, 2011
human emotional facial expressions.73 During the recall of an unresolved life event, only patients with borderline personality disorder (vs healthy controls) showed bilateral activations of the amygdala.74 The amygdala is assumed to function as a key structure during the processing of anxiety and other affective states. As expected, there was no activation of the amygdala in patients with borderline personality disorder or in healthy individuals when positive visual material was presented.75 Furthermore, activation of the prefrontal cortex was reported for patients with borderline personality disorder after inducing negative emotions, possibly indicating an attempt to control intensive emotions.74,76 When asked to use a cognitive strategy to control their responses to unpleasant pictures by distancing themselves from the image, those with borderline personality disorder had less blood-oxygen-level dependent (BOLD) signal changes in the anterior cingulate and had greater activation in the superior temporal sulcus and superior frontal gyrus than did controls. Thus, patients with borderline personality disorder do not seem to engage the cognitive control regions to the extent that healthy individuals do, which might contribute to the affective instability of this disorder.75 Further findings from functional MRI studies indicate different neuronal patterns of traumatic memory in patients with and without post-traumatic stress disorder.77 Only in patients with post-traumatic stress disorder did a thermic paininduction evoke a decreased amygdala activation.78 These results might indicate the existence of different neural networks in subgroups of patients with borderline personality disorder. These reports lend support to the assumption of a dysfunctional frontolimbic network in borderline personality disorder. This network seems to involve the anterior cingular cortex, the orbitofrontal cortex, the dorsolateral prefrontal cortex, the hippocampus, and the amygdala. However, future studies are needed, including with individuals who do not have borderline personality disorder, including healthy controls, patients with axis I disorders or patients with other personality disorders to investigate the specificity of these findings.
Treatment The American Psychiatric Association’s practice guideline recommends psychotherapy as the main treatment of borderline personality disorder, with pharmacotherapy as an adjunctive component of treatment that targets state symptoms during periods of acute decompensation and trait vulnerabilities.28 This guideline is a set of evidencebased best practice recommendations. The pharmacotherapy algorithms are directed towards three clusters of symptoms: cognitive-perceptual symptoms (with neuroleptics), affective symptoms (with selective serotoninreuptake inhibitors), and impulsive-behavioural dyscontrol (with selective serotonin-reuptake inhibitors and low-dose neuroleptics).28 77
Seminar
Treatment
Mean dose
Bogenschutz et al89
Olanzapine vs placebo
6·9 mg per day
de la Fuente et al92
Carbamazepine vs placebo
Blood concentration 6·4–7·1 μg/mL
Frankenburg et al82
Sodium valproate vs placebo
850 mg per day
Goldberg et al87
Tiotixene vs placebo
8·7 mg per day
Hollander et al93
Sodium valproate vs placebo
Blood concentration 64·6 μg/mL
Hollander et al94
Sodium valproate vs placebo
1325 mg per day
Leone et al88
Loxapine vs placebo
14·4 mg per day
Leone et al88
Chlorpromazine vs placebo
110 mg per day
Pascual et al100
Ziprasidone vs placebo
81 mg per day
Reich et al96
Lamotrigine vs placebo
25–275 mg per day
Rinne et al86
Fluvoxamine vs placebo
150 mg per day
Salzman et al85
Fluoxetine vs placebo
40 mg per day
Schulz et al97
Olanzapine vs placebo
2·5–20 mg per day
Shafti and Shahveisi98
Olanzapine vs haloperidol
2·5–10 mg per day
Soloff et al84
Haloperidol vs amitriptyline vs placebo
4·8 mg per day; 149·1 mg per day
Soloff et al83
Haloperidol vs phenelzine sulfate vs placebo
3·9 mg per day; 60·45 mg per day
Zanarini et al90
Olanzapine vs placebo
5·3 mg per day
Zanarini et al95
Omega-3 fatty acids vs placebo
1 mg per day
Zanarini et al91
Olanzapine vs fluoxetine vs olanzapine plus fluoxetine
3·3 mg per day; 15·0 mg per day; 3·2 mg per day plus 12·7 mg per day
Ziegenhorn et al99
Clonidine vs placebo
0·45 mg per day
Table 1: Randomised controlled trials of pharmacotherapy in patients with borderline personality disorder
Pharmacotherapy
See Online for webappendix
78
Following the Cochrane criteria, evidence for the pharmacological treatment of BPD was reviewed in 2006;79 only ten small randomised controlled trials (RCTs) were identified. The authors concluded that evidence for the pharmacological treatment of BPD was poor.79 In the meantime, several RCTs on the pharmacological treatment of borderline personality disorder have been done and several reviews have been published.80,81 Using the same Cochrane criteria as Binks and colleagues,79 we identified 19 RCTs of pharmacotherapy alone in patients with borderline personality disorder (table 1).82–100 Additionally, the combination of pharmacotherapy with psychotherapy was studied in four RCTs.101–104 As stated by the UK National Institute for Health and Clinical Excellence (NICE) guideline group, funding has turned out to be unclear for five additional RCTs (references 75–82 in the webappendix).105 For this reason, the NICE group deemed the evidence provided by these studies as unreliable and excluded them from analysis. Taking these caveats into account, we decided to exclude these five RCTs from this report. Herein, our review differs from that by Lieb and colleagues,81 which was critically commented on by Kendall and colleagues.106 According to the results of the RCTs included, evidence for the pharmacotherapy of borderline personality disorder is as follows. The antidepressants amitriptyline and imipramine were more effective than placebo for some symptoms of
depression, but not for other symptoms of borderline personality disorder.84 Few differences between monoamine oxidase inhibitors (phenelzine) and placebo were reported.83 Phenelzine reduced hostility, but not depression.83 In one RCT, fluoxetine was not clearly more effective than placebo for depression, but a beneficial effect on anger was reported.85 In another RCT, beneficial effects of fluvoxamine on mood shifts were reported, but not on aggression or impulsivity.86 In some studies, the typical antipsychotic haloperidol was more effective than placebo for several symptoms, but results vary.83,84,87 By contrast with their previous results,84 Soloff and colleagues83 did not find haloperidol more effective than placebo for most outcome measures. Haloperidol seems to be more effective than tricyclic antidepressants (amitripityline) for hostility and schizotypal symptoms.84 Phenelzine was more effective than haloperidol for some symptoms (eg, depression, anxiety, schizotypal symptoms), but not for others (eg, impulse control).83 At present, there is no evidence that one typical antipsychotic is more effective than another (ie, loxapine vs chlorpromazine or tiotixene vs haloperidol).79,88 In two RCTs, the atypical antipsychotic olanzapine was more effective than placebo for several, but not all, symptoms;89,90 however, in another RCT, no superiority of olanzapine over placebo was reported.97 Olanzapine monotherapy and olanzapine combined with fluoxetine were more effective than fluoxetine alone for depression and impulsive aggression.91 This combination had no advantage over olanzapine monotherapy.91 Ziprasidone had no superiority over placebo.100 In a recent small RCT of 28 patients with borderline personality disorder, no differences between olanzapine and haloperidol were reported.98 No studies comparing different atypical antipsychotics are available. The mood stabiliser carbamazepine had no superiority over placebo.92 No effects of sodium valproate were reported on depression.82,93 Beneficial effects on interpersonal sensitivity were reported in one RCT.82 In some, but not all, RCTs, a beneficial of sodium valproate was reported on anger or aggression.82,93,94 Beneficial effects of lamotrigine on affective instability and impulsivity were reported in one RCT.96 Improvements were reported for an omega-3 fatty acid treatment95 in one small RCT of 30 patients with borderline personality disorder, and for clonidine in 18 patients in another small RCT.99 In summary, evidence for pharmacotherapy in borderline personality disorder varies. Beneficial effects on depression, aggression, and other symptoms were reported in some RCTS, but not in others. Many of the findings are based on single RCTs or on small samples of patients. Some studies (eg, of olanzapine) included only female patients. Only a few studies included long-term follow-ups. As stressed by the NICE group, some of the drugs are potentially harmful—eg, valproate semisodium, which is dangerous for women of child-bearing age.106 Antipsychotics have many neurological side-effects and www.thelancet.com Vol 377 January 1, 2011
Seminar
Treatment
Comparison
Bateman and Fonagy109 Mentalisation-based treatment in a partial hospital setting Bateman and Fonagy118 Mentalisation-based psychodynamic treatment
Treatment as usual Structured clinical management
Blum et al120
Brief cognitive-behavioural therapy plus treatment as usual
Treatment as usual
Bohus et al121
Inpatient dialectical behaviour therapy
Treatment as usual
Clarkin et al122
Transference-focused therapy
Dialectical behaviour therapy as supportive therapy
Cottraux et al123
Cognitive-behavioural therapy
Client-centred therapy
Davidson et al119
Brief cognitive-behavioural therapy plus treatment as usual
Treatment as usual
Doering et al132
Transference-focused therapy
Community treatment by experienced therapists
Farrel et al124
Schema-focused therapy plus treatment as usual
Treatment as usual
Giesen-Bloo et al125
Schema-focused therapy
Transference-focused therapy
Gregory et al126
Dynamic deconstructive therapy
Treatment as usual
Harned et al127
Dialectical behaviour therapy
Community treatment by experts
Koons et al114
Dialectical behaviour therapy
Treatment as usual
Linehan et al110
Dialectical behaviour therapy
Treatment as usual
Linehan et al112
Dialectical behaviour therapy
Treatment as usual
Linehan et al111
Dialectical behaviour therapy
Comprehensive validation therapy plus a 12-step substance misuse programme
Linehan et al128
Dialectical behaviour therapy
Therapy by experts
McMain et al129
Dialectical behaviour therapy
Psychodynamically informed clinical management
Munroe-Blum et al116
Psychodynamic therapy
Interpersonal group therapy
Soler et al130
Dialectical behaviour therapy skills training
Standard group therapy
Tyrer et al117
Brief cognitive-behaviour therapy
Treatment as usual
Turner et al115
Dialectical behaviour therapy
Client-centred therapy
Verheul et al131
Dialectical behaviour therapy
Treatment as usual
Weinberg et al133
Brief cognitive-behavioural therapy plus treatment as usual
Treatment as usual
Table 2: Randomised controlled trials of psychotherapy in patients with borderline personality disorder
metabolic effects leading to weight gain and increased risk of diabetes.106 Thus, more evidence is needed before definitive recommendations can be made about the pharmacological treatment of any symptoms of borderline personality disorder.
Psychotherapy Several methods of psychotherapy are available for patients with borderline personality disorder—eg, cognitivebehavioural, interpersonal, or psychodynamic treatments. Two formal reviews on the efficacy of psychotherapy for the disorder exist.107,108 Following Cochrane criteria, Binks and colleagues107 identified only seven RCTs published up to 2002.109–115 Brazier and colleagues108 included two additional RCTs.116,117 In their review on borderline personality disorder, Lieb and colleagues4 identified 11 RCTs. Since 2006, several RCTs on psychotherapy have been published. Following the Cochrane criteria used by Binks and colleagues,107 we identified 24 RCTs on psychotherapy alone in borderline personality disorder (table 2).109–112,114–133 For cognitive behaviour therapy, the results can be summarised as follows. In studies comparing dialectical behaviour therapy with treatment as usual, dialectical behaviour therapy was more effective in several outcome measures (eg, self harm, parasuicidal behaviour, suicidal ideation).110–112,114,128,131 However, the number of patients still www.thelancet.com Vol 377 January 1, 2011
meeting the diagnostic criteria of borderline personality disorder did not differ. For depression, anxiety, admission to hospital, or drop outs, only two trials also reported superiority of dialectical behaviour therapy.121,128 In another RCT,131 dialectical behaviour therapy was more effective than treatment as usual in reducing self-mutilating behaviours and self-damaging impulsive acts, but not in reducing suicidal behaviour. The skills training component for dialectical behaviour therapy was superior to a nonmanualised standard group therapy in several measures (eg, depression, anxiety).130 Furthermore, dialectical behaviour therapy was superior to a community treatment by experts (not further specified non-cognitive behaviour therapy) with respect to remission of co-occurring substance-dependence disorders.127 No superiority was identified for remission of other axis I disorders.127 No clear differences were reported between dialectical behaviour therapy focusing on substance misuse and comprehensive validation therapy plus a 12-step substance misuse programme.111 Compared with client-centred therapy, dialectical behaviour therapy was superior in reducing parasuicidal behaviour, suicidal ideation, and general psychiatric severity.115 No superiority of dialectical behaviour therapy over client-centred therapy was identified for anxiety.115 In the most recent RCT,129 which was sufficiently powered to detect a medium effect size (sample sizes of 90 for both groups), both dialectical behaviour 79
Seminar
therapy and a well specified and psychodynamically informed clinical management provided significant improvements with no differences between treatments. Results from another RCT indicated that schema-focused therapy was superior to transference-focused psychotherapy.125 However, differences in therapist competence ratings indicate that transference-focused psychotherapy might have been less competently implemented than schema-focused therapy. Thus, replications are needed. Adding an 8-month schema-focused therapy group to treatment as usual was more effective than just treatment as usual.124 Adding availability of a therapist by telephone outside office hours to schema-focused therapy did not lead to extra effects.134 Client-centred therapy was as effective as cognitive behaviour therapy with respect to the response criterion, although cognitive behaviour therapy was superior in some other outcome measures.123 Brief cognitive behaviour therapy was not superior to treatment as usual, but was more cost-effective.117 In three studies, new and brief cognitive-behavioural treatments for patients with borderline personality disorder were tested in an addon design.119,120,133 When added to treatment as usual, these strategies were superior to treatment as usual alone in some, but not in all, outcome measures.119,120,133 Seven RCTs have studied psychodynamic psychotherapy.109,116,118,122,125,126,132 In three studies, psychodynamic psychotherapy was more effective than treatment as usual for most outcome measures.109,126,132 In one trial, the effects of mentalisation-based treatment in a partial hospital setting might have been confounded by duration of partial hospitalisation, which was longer in the mentalisation-based treatment condition.109 In a recent RCT,118 however, this treatment was superior to manualdriven structured clinical management for primary (suicidal and self-injurious behaviours, treatment in hospital) and secondary outcome measures (eg, depression, general symptom distress, interpersonal functioning). Psychodynamic psychotherapy was reported in one RCT to be as equally effective as an interpersonal group therapy.116 In another RCT that compared transference-focused psychotherapy, dialectical behaviour therapy, and psychodynamic supportive psychotherapy,122 transference-focused psychotherapy and dialectical behaviour therapy reduced suicidal tendency to the same extent. Transference-focused psychotherapy was superior to dialectical behaviour therapy in some measures of affect regulation, impulsivity, and attachment.122,135 In summary, dialectical behaviour therapy and specific forms of psychodynamic psychotherapy seem to be superior to treatment as usual in some clinically relevant problems of borderline personality disorder. Adding brief cognitive behaviour therapy to treatment as usual seems to be superior to treatment as usual alone. According to follow-up studies, effects of psychotherapy are stable over time.111,112,116,119,120,124,127,128,136–138 At present, there is no clear evidence that one specific form of psychotherapy is 80
superior to another.139 In several studies comparing specific forms of psychotherapy, however, power was not sufficient to detect possible differences.111,116,122 In summary, there is evidence that psychotherapy is beneficial with respect to some clinically relevant problems of patients with borderline personality disorder. However, the available forms of psychotherapy do not yet lead to remission of borderline personality disorder for most patients (ie, no longer fulfilling the criteria of a diagnosis of borderline personality disorder). Thus, further research is needed.
Pharmacotherapy with psychotherapy The benefit of a combination of pharmacotherapy and psychotherapy in borderline personality disorder is unclear. Fluoxetine combined with dialectical behaviour therapy provided no additional benefit compared with dialectical behaviour therapy plus placebo.102 In one study,103 olanzapine added to dialectical behaviour therapy provided an additional benefit compared with dialectical behaviour therapy, although no differences were reported in another study in favour of the combined treatment.104 The combination of interpersonal therapy and fluoxetine was superior to fluoxetine plus clinical management.101
Future perspectives Although much has been learned about borderline personality disorder in recent years, several questions remain. Despite conceptual coherence, borderline personality disorder seems to be a heterogeneous diagnostic category that is less stable and distinct over time than expected. These findings raise questions of both how to conceptualise this disorder and how to implement it in future versions of DSM as a form of personality pathology that is both enduring and distinct from other personality disorders.20 Furthermore, the discussion on whether a categorical or a dimensional model best suits personality disorders is ongoing.140,141 The results of the Collaborative Longitudinal Personality Disorders Study (CLPS)5,142 suggest reconceptualising personality disorders as hybrids of stable personality traits and as intermittently expressed symptomatic behaviours that are attempts to cope with or defend against or compensate for these pathological traits (eg, self-harm to reduce affective tension). Further research is needed on the association between personality traits and personality disorder psychopathological changes as well as on the relation between personality disorders and personality functioning.21,143 Personality might function differently at different ages and in response to different needs.143 Future research on the causes of this disorder should investigate how genetic and psychosocial factors interact with neurotransmitter function to lead to cognitive and emotional regulations and specific traits.44 The available findings of neuroimaging studies lend support to the assumption of a dysfunctional frontolimbic network in borderline www.thelancet.com Vol 377 January 1, 2011
Seminar
personality disorder. The exact molecular nature of this dysfunction is not yet clear. Future studies should include individuals who do not have borderline personality disorder (ie, healthy controls and patients with axis I disorders or other personality disorders) to establish the specificity of findings. Furthermore, further research is needed for childhood and adolescent precursor symptoms of adult borderline personality disorder as well as borderline personality disorder in elderly patients.42 More work is also needed to understand the neurobiology of interpersonal dysfunction and attachment in borderline personality disorder.42 Future studies on pharmacotherapy are needed to improve the empirical support for its use in patients with borderline personality disorder, including studies of long-term effects and studies of the combination of pharmacotherapy and psychotherapy. Psychotherapy is beneficial for some clinically relevant problems of patients with borderline personality disorder. However, further research is needed to also improve other core features of this disorder.5 Improvements in borderline personality disorder are commonly followed by improvements in major depressive disorder.18 Therefore, clinicians are recommended not to focus on the treatment of major depressive disorder in the hope that improvements in major depressive disorder will lead to improvements in borderline personality disorder; instead the personality disorder should be treated.18 Changes in personality traits seem to be followed by changes in personality disorder psychopathology, but not vice versa.21 For this reason, clinicians and future studies should focus on treating the personality traits associated with borderline personality disorder. Focusing on personality traits and on personality functioning associated with personality disorders is consistent with the recent proposals made for DSM-V.144 Contributors All authors conceived the outline of the Seminar and co-wrote the body of the text.
8
9
10 11 12
13
14
15
16
17
18
19
20
21
22
23
Conflicts of interest We declare that we have no conflicts of interest.
24
References 1 Bender DS, Dolan RT, Skodol AE, et al. Treatment utilization by patients with personality disorders. Am J Psychiatry 2001; 158: 295–302. 2 Skodol AE, Gunderson JG, Pfohl B, Widiger TA, Livesley WJ, Siever LJ. The borderline diagnosis I: psychopathology, comorbidity, and personality structure. Biol Psychiatry 2002; 51: 936–50. 3 Oldham JM. Borderline personality disorder and suicidality. Am J Psychiatry 2006; 163: 20–26. 4 Lieb K, Zanarini MC, Schmahl C, Linehan MM, Bohus M. Borderline personality disorder. Lancet 2004; 364: 453–61. 5 Skodol AE, Gunderson JG, Shea MT, et al. The Collaborative Longitudinal Personality Disorders Study (CLPS): overview and implications. J Pers Disord 2005; 19: 487–504. 6 Lenzenweger M, Lane M, Loranger A, Kessler R. Personality disorders in the National Comorbidity Survey Replication. Biol Psychiatry 2007; 62: 553–64. 7 Grant BF, Chou SP, Goldstein RB, et al. Prevalence, correlates, disability, and comorbidity of DSM-IV borderline personality disorder: results from the Wave 2 National Epidemiologic Survey on Alcohol and Related Conditions. J Clin Psychiatry 2008; 69: 533–45.
25
www.thelancet.com Vol 377 January 1, 2011
26
27
28
29
30
Torgersen S, Kringlen, E, Cramer V. The prevalence of personality disorders in a community sample. Arch Gen Psychiatry 2001; 58: 590–96. Torgersen J. Epidemiology. In: Oldham J, Skodol, AE, Bender, DS, eds. Textbook of personality disorders. Washington, DC: American Psychiatric Publishing, 2005: 129–41. Gunderson JG. Borderline personality disorder: ontogeny of a diagnosis. Am J Psychiatry 2009; 166: 530–39. Gross R, Olfson M, Gameroff M, et al. Borderline personality disorder in primary care. Arch Intern Med 2002; 162: 53–60. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, fourth edn. Washington, DC: American Psychiatric Association, 1994. Grilo CM, Becker DF, Anez LM, McGlashan TH. Diagnostic efficiency of DSM-IV criteria for borderline personality disorder: an evaluation in Hispanic men and women with substance use disorders. J Consult Clin Psychol 2004; 72: 126–31. Grilo CM, Sanislow CA, Skodol AE, et al. Longitudinal diagnostic efficiency of DSM-IV criteria for borderline personality disorder: a 2-year prospective study. Can J Psychiatry 2007; 52: 357–62. Sanislow CA, Grilo CM, Morey LC, et al. Confirmatory factor analysis of DSM-IV criteria for borderline personality disorder: findings from the collaborative longitudinal personality disorders study. Am J Psychiatry 2002; 159: 284–90. Zanarini MC, Frankenburg FR, Hennen J, Reich DB, Silk KR. Prediction of the 10-year course of borderline personality disorder. Am J Psychiatry 2006; 163: 827–32. Gunderson JG, Bender D, Sanislow C, et al. Plausibility and possible determinants of sudden “remissions” in borderline patients. Psychiatry 2003; 66: 111–19. Gunderson JG, Morey LC, Stout RL, et al. Major depressive disorder and borderline personality disorder revisited: longitudinal interactions. J Clin Psychiatry 2004; 65: 1049–56. Zanarini MC, Frankenburg FR, Reich DB, Silk KR, Hudson JI, McSweeney LB. The subsyndromal phenomenology of borderline personality disorder: a 10-year follow-up study. Am J Psychiatry 2007; 164: 929–35. Sanislow CA, Little TD, Ansell EB, et al. Ten-year stability and latent structure of the DSM-IV schizotypal, borderline, avoidant, and obsessive-compulsive personality disorders. J Abnorm Psychol 2009; 118: 507–19. Warner MB, Morey LC, Finch JF, et al. The longitudinal relationship of personality traits and disorders. J Abnorm Psychol 2004; 113: 217–27. Hopwood CJ, Newman DA, Donnellan MB, et al. The stability of personality traits in individuals with borderline personality disorder. J Abnorm Psychol 2009; 118: 806–15. Schmideberg M. The borderline patient. In: Arieti S, ed. American handbook of psychiatry New York: Basic Books, 1959: 398–416. Carpenter WT Jr, Gunderson JG. Five year follow-up comparison of borderline and schizophrenic patients. Compr Psychiatry 1977; 18: 567–71. Paris J, Gunderson J, Weinberg I. The interface between borderline personality disorder and bipolar spectrum disorders. Compr Psychiatry 2007; 48: 145–54. Lewis KL, Grenyer BF. Borderline personality or complex posttraumatic stress disorder? An update on the controversy. Harv Rev Psychiatry 2009; 17: 322–28. Skodol AE, Pagano ME, Bender DS, et al. Stability of functional impairment in patients with schizotypal, borderline, avoidant, or obsessive-compulsive personality disorder over two years. Psychol Med 2005; 35: 443–51. American Psychiatric Association Practice Guidelines. Practice guideline for the treatment of patients with borderline personality disorder. Am J Psychiatry 2001; 158: 1–52. Zanarini MC, Frankenburg FR, Vujanovic AA, Hennen J, Reich DB, Silk KR. Axis II comorbidity of borderline personality disorder: description of 6-year course and prediction to time-to-remission. Acta Psychiatr Scand 2004; 110: 416–20. Skodol AE, Gunderson JG, McGlashan TH, et al. Functional impairment in patients with schizotypal, borderline, avoidant, or obsessive-compulsive personality disorder. Am J Psychiatry 2002; 159: 276–83.
81
Seminar
31
32
33
34
35
36 37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
82
Zanarini MC, Frankenburg FR, Reich DB, Fitzmaurice G. The 10-year course of psychosocial functioning among patients with borderline personality disorder and axis II comparison subjects. Acta Psychiatr Scand 2010; 122: 103–09. Zanarini MC, Yong L, Frankenburg FR, et al. Severity of reported childhood sexual abuse and its relationship to severity of borderline psychopathology and psychosocial impairment among borderline inpatients. J Nerv Ment Dis 2002; 190: 381–87. Yen S, Shea MT, Battle CL, et al. Traumatic exposure and posttraumatic stress disorder in borderline, schizotypal, avoidant, and obsessive-compulsive personality disorders: findings from the collaborative longitudinal personality disorders study. J Nerv Ment Dis 2002; 190: 510–18. Fossati A, Madeddu F, Maffei C. Borderline personality disorder and childhood sexual abuse: a meta-analytic study. J Personal Disord 1999; 13: 268–80. Paris J. The nature of borderline personality disorder: multiple dimensions, multiple symptoms, but one category. J Pers Disord 2007; 21: 457–73. Gabbard GO. Mind, brain, and personality disorders. Am J Psychiatry 2005; 162: 648–55. Wagner S, Baskaya O, Lieb K, Dahmen N, Tadic A. The 5-HTTLPR polymorphism modulates the association of serious life events (SLE) and impulsivity in patients with borderline personality disorder. J Psychiatr Res 2009; 43: 1067–72. Wagner S, Baskaya O, Dahmen N, Lieb K, Tadic A. Modulatory role of the brain-derived neurotrophic factor Val66Met polymorphism on the effects of serious life events on impulsive aggression in borderline personality disorder. Genes Brain Behav 2009; 9: 97–102. Siever LJ, Torgersen S, Gunderson JG, Livesley WJ, Kendler KS. The borderline diagnosis III: identifying endophenotypes for genetic studies. Biol Psychiatry 2002; 51: 964–68. Torgersen S, Czajkowski N, Jacobson K, et al. Dimensional representations of DSM-IV cluster B personality disorders in a population-based sample of Norwegian twins: a multivariate study. Psychol Med 2008: 38: 1617–25. Skodol AE, Siever LJ, Livesley WJ, Gunderson JG, Pfohl B, Widiger TA. The borderline diagnosis II: biology, genetics, and clinical course. Biol Psychiatry 2002; 51: 951–63. New AS, Goodman M, Triebwasser J, Siever LJ. Recent advances in the biological study of personality disorders. Psychiatr Clin North Am 2008; 31: 441–61. Cloninger CR. Genetics. In: J. Oldham JM, Skodol AE, Bender DS, eds. Textbook of personality disorders. Washington, DC: American Psychiatric Publishing, 2005: 143–54. Coccaro EF, Siever LJ. Neurobiology. In: Oldham JM, Skodol AE, Bender DS, eds. Textbook of personality disorders. Washington, DC: American Psychiatric Publishing, 2005: 155–69. Ni X, Chan K, Bulgin N, et al. Association between serotonin transporter gene and borderline personality disorder. J Psychiatr Res 2006; 40: 448–53. Pascual JC, Soler J, Barrachina J, et al. Failure to detect an association between the serotonin transporter gene and borderline personality disorder. J Psychiatr Res 2008; 42: 87–88. Silva H, Iturra P, Solari A, et al. Serotonin transporter polymorphism and fluoxetine effect on impulsiveness and aggression in borderline personality disorder. Actas Esp Psiquiatr 2007; 35: 387–92. Tadic A, Victor A, Baskaya O, et al. Interaction between gene variants of the serotonin transporter promoter region (5-HTTLPR) and catechol O-methyltransferase (COMT) in borderline personality disorder. Am J Med Genet B Neuropsychiatr Genet 2009; 150B: 487–95. Ni X, Chan D, Chan K, McMain S, Kennedy JL. Serotonin genes and gene-gene interactions in borderline personality disorder in a matched case-control study. Prog Neuropsychopharmacol Biol Psychiatry 2009; 33: 128–33. Ni X, Sicard T, Bulgin N, et al. Monoamine oxidase a gene is associated with borderline personality disorder. Psychiatr Genet 2007; 17: 153–57. Lahmeyer HW, Reynolds CF 3rd, Kupfer DJ, King R. Biologic markers in borderline personality disorder: a review. J Clin Psychiatry 1989; 50: 217–25. Grossman R, Yehuda R, Siever L. The dexamethasone suppression test and glucocorticoid receptors in borderline personality disorder. Ann N Y Acad Sci 1997; 821: 459–64.
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
Zimmerman DJ, Choi-Kain LW. The hypothalamic-pituitaryadrenal axis in borderline personality disorder: a review. Harv Rev Psychiatry 2009; 17: 167–83. Schmahl CG, Vermetten E, Elzinga BM, Bremner DJ. Magnetic resonance imaging of hippocampal and amygdala volume in women with childhood abuse and borderline personality disorder. Psychiatry Res 2003; 122: 193–98. Weniger G, Lange C, Sachsse U, Irle E. Reduced amygdala and hippocampus size in trauma-exposed women with borderline personality disorder and without posttraumatic stress disorder. J Psychiatry Neurosci 2009; 34: 383–88. Chanen AM, Velakoulis D, Carison K, et al. Orbitofrontal, amygdala and hippocampal volumes in teenagers with first-presentation borderline personality disorder. Psychiatry Res 2008; 163: 116–25. Kitayama N, Vaccarino V, Kutner M, Weiss P, Bremner JD. Magnetic resonance imaging (MRI) measurement of hippocampal volume in posttraumatic stress disorder: a meta-analysis. J Affect Disord 2005; 88: 79–86. Bremner JD, Randall P, Vermetten E, et al. Magnetic resonance imaging-based measurement of hippocampal volume in postttraumatic stress disorder related to childhood physical and sexual abuse—a preliminary report. Biol Psychiatry 1997; 41: 23–32. Nunes PM, Wenzel A, Borges KT, Porto CR, Caminha RM, de Oliveira IR. Volumes of the hippocampus and amygdala in patients with borderline personality disorder: a meta-analysis. J Pers Disord 2009; 23: 333–45. Schmahl C, Berne K, Krause A, et al. Hippocampus and amygdala volumes in patients with borderline personality disorder with or without posttraumatic stress disorder. J Psychiatry Neurosci 2009; 34: 289–95. Chanen AM, Velakoulis D, Carison K, et al. Orbitofrontal, amygdala and hippocampal volumes in teenagers with first-presentation borderline personality disorder. Psychiatry Res 2008; 163: 116–25. New AS, Hazlett EA, Buchsbaum MS, et al. Amygdala-prefrontal disconnection in borderline personality disorder. Neuropsychopharmacology 2007; 32: 1629–40. Minzenberg MJ, Fan J, New AS, Tang CY, Siever LJ. Frontolimbic structural changes in borderline personality disorder. J Psychiatr Res 2008; 42: 727–33. Garner B, Chanen AM, Phillips L, et al. Pituitary volume in teenagers with first-presentation borderline personality disorder. Psychiatry Res 2007; 156: 257–61. Jovev M, Garner B, Phillips L, et al. An MRI study of pituitary volume and parasuicidal behavior in teenagers with first-presentation borderline personality disorder. Psychiatry Res 2008; 162: 273–77. Garner B, Pariante CM, Wood SJ, Velakoulis D, Phillips L, Soulsby B. Pituitary volume predicts future transition to psychosis in individuals at ultra-high risk of developing psychosis. Biol Psychiatry 2005; 58: 417–23. Irle E, Lange C, Sachsse U. Reduced size and abnormal asymmetry of parietal cortex in women with borderline personality disorder. Biol Psychiatry 2005; 57: 173–82. Irle E, Lange C, Weniger G, Sachsse U. Size abnormalities of the superior parietal cortices are related to dissociation in borderline personality disorder. Psychiatry Res 2007; 156: 139–49. Brunner R, Henze R, Parzer P, et al. Reduced prefrontal and orbitofrontal gray matter in female adolescents with borderline personality disorder: is it disorder specific? Neuroimage 2010; 49: 114–20. Juengling FD, Schmahl C, Hesslinger B, et al. Positron emission tomography in female patients with borderline personality disorder. J Psychiatr Res 2003; 37: 109–15. New AS, Hazlett EA, Newmark RE, et al. Laboratory induced aggression: a positron emission tomography study of aggressive individuals with borderline personality disorder. Biol Psychiatry 2009; 66: 1107–14. Koenigsberg HW, Siever LJ, Lee H, et al. Neural correlates of emotion processing in borderline personality disorder. Psychiatry Res 2009; 172: 192–99. Minzenberg MJ, Fan J, New AS, Tang CY, Siever LJ. Fronto-limbic dysfunction in response to facial emotion in borderline personality disorder: an event-related fMRI study. Psychiatry Res 2007; 155: 231–43.
www.thelancet.com Vol 377 January 1, 2011
Seminar
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88 89
90
91
92 93
94
95
Beblo T, Driessen M, Mertens M, et al. Functional MRI correlates of the recall of unresolved life events in borderline personality disorder. Psychol Med 2006; 36: 845–56. Koenigsberg HW, Fan J, Ochsner KN, et al. Neural correlates of the use of psychological distancing to regulate responses to negative social cues: a study of patients with borderline personality disorder. Biol Psychiatry 2009; 66: 854–63. Herpertz SC, Dietrich TM, Wenning B, et al. Evidence of abnormal amygdala functioning in borderline personality disorder: a functional MRI study. Biol Psychiatry 2001; 50: 292–98. Driessen M, Beblo T, Mertens M, et al. Posttraumatic stress disorder and fMRI activation patterns of traumatic memory in patients with borderline personality disorder. Biol Psychiatry 2004; 55: 603–11. Kraus A, Esposito F, Seifritz E, et al. Amygdala deactivation as a neural correlate of pain processing in patients with borderline personality disorder and co-occurrent posttraumatic stress disorder. Biol Psychiatry 2009; 65: 819–22. Binks CA, Fenton M, McCarthy L, Lee T, Adams CE, Duggan C. Pharmacological interventions for people with borderline personality disorder. Cochrane Database Syst Rev 2006; 1: CD005653. Nose M, Cipriani A, Biancosino B, Grassi L, Barbui C. Efficacy of pharmacotherapy against core traits of borderline personality disorder: meta-analysis of randomized controlled trials. Int Clin Psychopharmacol 2006; 21: 345–53. Lieb K, Vollm B, Rucker G, Timmer A, Stoffers JM. Pharmacotherapy for borderline personality disorder: Cochrane systematic review of randomised trials. Br J Psychiatry 2010; 196: 4–12. Frankenburg FR, Zanarini MC. Divalproex sodium treatment of women with borderline personality disorder and bipolar II disorder: a double-blind placebo-controlled pilot study. J Clin Psychiatry 2002; 63: 442–46. Soloff PH, Cornelius J, George A, Nathan S, Perel JM, Ulrich RF. Efficacy of phenelzine and haloperidol in borderline personality disorder. Arch Gen Psychiatry 1993; 50: 377–85. Soloff PH, George A, Nathan S, et al. Amitriptyline versus haloperidol in borderlines: final outcomes and predictors of response. J Clin Psychopharmacol 1989; 9: 238–46. Salzman C, Wolfson AN, Schatzberg A, et al. Effect of fluoxetine on anger in symptomatic volunteers with borderline personality disorder. J Clin Psychopharmacol 1995; 15: 23–29. Rinne T, van den Brink W, Wouters L, van Dyck R. SSRI treatment of borderline personality disorder: a randomized, placebo-controlled clinical trial for female patients with borderline personality disorder. Am J Psychiatry 2002; 159: 2048–54. Goldberg SC, Schulz SC, Schulz PM, Resnick RJ, Hamer RM, Friedel RO. Borderline and schizotypal personality disorders treated with low-dose thiothixene vs placebo. Arch Gen Psychiatry 1986; 43: 680–86. Leone NF. Response of borderline patients to loxapine and chlorpromazine. J Clin Psychiatry 1982; 43: 148–50. Bogenschutz MP, George Nurnberg H. Olanzapine versus placebo in the treatment of borderline personality disorder. J Clin Psychiatry 2004; 65: 104–09. Zanarini MC, Frankenburg FR. Olanzapine treatment of female borderline personality disorder patients: a double-blind, placebo-controlled pilot study. J Clin Psychiatry 2001; 62: 849–54. Zanarini MC, Frankenburg FR, Parachini EA. A preliminary, randomized trial of fluoxetine, olanzapine, and the olanzapine-fluoxetine combination in women with borderline personality disorder. J Clin Psychiatry 2004; 65: 903–07. de la Fuente JM, Lotstra F. A trial of carbamazepine in borderline personality disorder. Eur Neuropsychopharmacol 1994; 4: 479–86. Hollander E, Allen A, Lopez RP, et al. A preliminary double-blind, placebo-controlled trial of divalproex sodium in borderline personality disorder. J Clin Psychiatry 2001; 62: 199–203. Hollander E, Swann AC, Coccaro EF, Jiang P, Smith TB. Impact of trait impulsivity and state aggression on divalproex versus placebo response in borderline personality disorder. Am J Psychiatry 2005; 162: 621–24. Zanarini MC, Frankenburg FR. Omega-3 fatty acid treatment of women with borderline personality disorder: a double-blind, placebo-controlled pilot study. Am J Psychiatry 2003; 160: 167–69.
www.thelancet.com Vol 377 January 1, 2011
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
Reich DB, Zanarini MC, Bieri KA. A preliminary study of lamotrigine in the treatment of affective instability in borderline personality disorder. Int Clin Psychopharmacol 2009; 24: 270–75. Schulz SC, Zanarini MC, Bateman A, et al. Olanzapine for the treatment of borderline personality disorder: variable dose 12-week randomised double-blind placebo-controlled study. Br J Psychiatry 2008; 193: 485–92. Shafti SS, Shahveisi B. Olanzapine versus haloperidol in the management of borderline personality disorder: a randomized double-blind trial. J Clin Psychopharmacol 2010; 30: 44–47. Ziegenhorn AA, Roepke S, Schommer NC, et al. Clonidine improves hyperarousal in borderline personality disorder with or without comorbid posttraumatic stress disorder: a randomized, double-blind, placebo-controlled trial. J Clin Psychopharmacol 2009; 29: 170–73. Pascual JC, Soler J, Puigdemont D, et al. Ziprasidone in the treatment of borderline personality disorder: a double-blind, placebo-controlled, randomized study. J Clin Psychiatry 2008; 69: 603–08. Bellino S, Zizza M, Rinaldi C, Bogetto F. Combined treatment of major depression in patients with borderline personality disorder: a comparison with pharmacotherapy. Can J Psychiatry 2006; 51: 453–60. Simpson EB, Yen S, Costello E, et al. Combined dialectical behavior therapy and fluoxetine in the treatment of borderline personality disorder. J Clin Psychiatry 2004; 65: 379–85. Soler J, Pascual JC, Campins J, et al. Double-blind, placebo-controlled study of dialectical behavior therapy plus olanzapine for borderline personality disorder. Am J Psychiatry 2005; 162: 1221–24. Linehan MM, McDavid JD, Brown MZ, Sayrs JH, Gallop RJ. Olanzapine plus dialectical behavior therapy for women with high irritability who meet criteria for borderline personality disorder: a double-blind, placebo-controlled pilot study. J Clin Psychiatry 2008; 69: 999–1005. Borderline personality disorder. The NICE guideline on treatment and management. The British Psychological Society and The Royal College of Psychiatrists. http://www.nice.org.uk/nicemedia/ live/12125/43045/43045.pdf: p218 (accessed Aug 30, 2010). Kendall T, Burbeck R, Bateman A. Pharmacotherapy for borderline personality disorder: NICE guideline. Br J Psychiatry 2010; 196: 158–59. Binks CA, Fenton M, McCarthy L, Lee T, Adams CE, Duggan C. Psychological therapies for people with borderline personality disorder. Cochrane Database Syst Rev 2006; 1: CD005652. Brazier J, Tumur I, Holmes M, et al. Psychological therapies including dialectical behaviour therapy for borderline personality disorder: a systematic review and preliminary economic evaluation. Health Technol Assess 2006; 10: 1–117. Bateman A, Fonagy P. Effectiveness of partial hospitalization in the treatment of borderline personality disorder: a randomized controlled trial. Am J Psychiatry 1999; 156: 1563–69. Linehan MM, Armstrong HE, Suarez A, Allmon D, Heard HL. Cognitive-behavioral treatment of chronically parasuicidal borderline patients. Arch Gen Psychiatry 1991; 48: 1060–64. Linehan MM, Dimeff LA, Reynolds SK, et al. Dialectical behavior therapy versus comprehensive validation therapy plus 12-step for the treatment of opioid dependent women meeting criteria for borderline personality disorder. Drug Alcohol Depend 2002; 67: 13–26. Linehan MM, Schmidt H, Dimeff LA, Craft JC, Kanter J, Comtois KA. Dialectical behavior therapy for patients with borderline personality disorder and drug-dependence. Am J Addict 1999; 8: 279–92. van den Bosch LM, Verheul R, Schippers GM, van den Brink W. Dialectical behavior therapy of borderline patients with and without substance use problems. Implementation and long-term effects. Addict Behav 2002; 27: 911–23. Koons CR, Robins CJ, Tweed JL, et al. Efficacy of dialectical behavior therapy in women veterans with borderline personality disorder. Behavior Therapy 2001; 32: 371–90. Turner R. Naturalistic evaluation of dialectical behavior therapy-oriented treatment for borderline persoanlity disorder. Cognitive and Behavioral Practice 2000; 7: 413–19.
83
Seminar
116 Munroe-Blum H, Marziali E. A controlled trial of short-term group treatment for borderline personality disorder. J Pers Disord 1995; 9: 190–98. 117 Tyrer PTS, Schmidt U, Jones V, et al. Randomized controlled trial of brief cognitive behaviour therapy versus treatment as usual in recurrent deliberate self-harm: the POPMACT study. Psychol Med 2003; 33: 969–76 118 Bateman A, Fonagy P. Randomized controlled trial of outpatient mentalization-based treatment versus structured clinical management for borderline personality disorder. Am J Psychiatry 2009; 166: 1355–64. 119 Davidson K, Norrie J, Tyrer P, et al. The effectiveness of cognitive behavior therapy for borderline personality disorder: results from the borderline personality disorder study of cognitive therapy (BOSCOT) trial. J Pers Disord 2006; 20: 450–65. 120 Blum N, St John D, Pfohl B, et al. Systems Training for Emotional Predictability and Problem Solving (STEPPS) for outpatients with borderline personality disorder: a randomized controlled trial and 1-year follow-up. Am J Psychiatry 2008; 165: 468–78. 121 Bohus M, Haaf B, Simms T, et al. Effectiveness of inpatient dialectical behavioral therapy for borderline personality disorder: a controlled trial. Behav Res Ther 2004; 42: 487–99. 122 Clarkin JF, Levy KN, Lenzenweger MF, Kernberg OF. Evaluating three treatments for borderline personality disorder: a multiwave study. Am J Psychiatry 2007; 164: 922–28. 123 Cottraux J, Note ID, Boutitie F, et al. Cognitive therapy versus rogerian supportive therapy in borderline personality disorder. Two-year follow-up of a controlled pilot study. Psychother Psychosom 2009; 78: 307–16. 124 Farrell JM, Shaw IA, Webber MA. A schema-focused approach to group psychotherapy for outpatients with borderline personality disorder: a randomized controlled trial. J Behav Ther Exp Psychiatry 2009; 40: 317–28. 125 Giesen-Bloo J, van Dyck R, Spinhoven P, et al. Outpatient psychotherapy for borderline personality disorder: randomized trial of schema-focused therapy vs transference-focused psychotherapy. Arch Gen Psychiatry 2006; 63: 649–58. 126 Gregory R, Chlebowski S, Kang D, et al. A controlled trial of psychodynamic psychotherapy for co-occurring borderline personality disorder and alcohol use disorder. Psychotherapy: Theory/Research/Practice/Training 2008; 45: 28–41. 127 Harned MS, Chapman AL, Dexter-Mazza ET, Murray A, Comtois KA, Linehan MM. Treating co-occurring axis I disorders in recurrently suicidal women with borderline personality disorder: a 2-year randomized trial of dialectical behavior therapy versus community treatment by experts. J Consult Clin Psychol 2008; 76: 1068–75. 128 Linehan MM, Comtois KA, Murray AM, et al. Two-year randomized controlled trial and follow-up of dialectical behavior therapy vs therapy by experts for suicidal behaviors and borderline personality disorder. Arch Gen Psychiatry 2006; 63: 757–66. 129 McMain SF, Links PS, Gnam WH, et al. A randomized trial of dialectical behavior therapy versus general psychiatric management for borderline personality disorder. Am J Psychiatry 2009; 166: 1365–74.
84
130 Soler J, Pascual JC, Tiana T, et al. Dialectical behaviour therapy skills training compared to standard group therapy in borderline personality disorder: a 3-month randomised controlled clinical trial. Behav Res Ther 2009; 47: 353–58. 131 Verheul R, Van Den Bosch LM, Koeter MW, De Ridder MA, Stijnen T, Van Den Brink W. Dialectical behaviour therapy for women with borderline personality disorder: 12-month, randomised clinical trial in The Netherlands. Br J Psychiatry 2003; 182: 135–40. 132 Doering S, Horz S, Rentrop M, et al. Transference-focused psychotherapy v. treatment by community psychotherapists for borderline personality disorder: randomised controlled trial. Br J Psychiatry 2010; 196: 389–95. 133 Weinberg I, Gunderson JG, Hennen J, Cutter CJ Jr. Manual assisted cognitive treatment for deliberate self-harm in borderline personality disorder patients. J Pers Disord 2006; 20: 482–92. 134 Nadort M, Arntz A, Smit JH, et al. Implementation of outpatient schema therapy for borderline personality disorder with versus without crisis support by the therapist outside office hours: a randomized trial. Behav Res Ther 2009; 47: 961–73. 135 Levy KN, Meehan KB, Kelly KM, et al. Change in attachment patterns and reflective function in a randomized control trial of transference-focused psychotherapy for borderline personality disorder. J Consult Clin Psychol 2006; 74: 1027–40. 136 Bateman A, Fonagy P. Treatment of borderline personality disorder with psychoanalytically oriented partial hospitalization: an 18-month follow-up. Am J Psychiatry 2001; 158: 36–42. 137 Linehan MM, Heard HL, Armstrong HE. Naturalistic follow-up of a behavioral treatment for chronically parasuicidal borderline patients. Arch Gen Psychiatry 1993; 50: 971–74. 138 Bateman A, Fonagy P. 8-year follow-up of patients treated for borderline personality disorder: mentalization-based treatment versus treatment as usual. Am J Psychiatry 2008; 165: 631–38. 139 Zanarini MC. Psychotherapy of borderline personality disorder. Acta Psychiatr Scand 2009; 120: 373–77. 140 Skodol AE, Oldham JM, Bender DS, et al. Dimensional representations of DSM-IV personality disorders: relationships to functional impairment. Am J Psychiatry 2005; 162: 1919–25. 141 Widiger TA. Current controversies in nosology and diagnosis of personality disorders. Psychiatric Annals 2007; 32: 93–99. 142 McGlashan TH, Grilo CM, Sanislow CA, et al. Two-year prevalence and stability of individual DSM-IV criteria for schizotypal, borderline, avoidant, and obsessive-compulsive personality disorders: toward a hybrid model of axis II disorders. Am J Psychiatry 2005; 162: 883–89. 143 Tyrer P, Coombs N, Ibrahimi F, et al. Critical developments in the assessment of personality disorder. Br J Psychiatry Suppl 2007; 49: s51–59. 144 American Psychiatric Association. DSM-5: The future of psychiatric diagnosis. http://www.dsm5.org (accessed Aug 30, 2010).
www.thelancet.com Vol 377 January 1, 2011
Health Policy
Measuring impact in the Millennium Development Goal era and beyond: a new approach to large-scale effectiveness evaluations Cesar G Victora, Robert E Black, J Ties Boerma, Jennifer Bryce
Evaluation of large-scale programmes and initiatives aimed at improvement of health in countries of low and middle income needs a new approach. Traditional designs, which compare areas with and without a given programme, are no longer relevant at a time when many programmes are being scaled up in virtually every district in the world. We propose an evolution in evaluation design, a national platform approach that: uses the district as the unit of design and analysis; is based on continuous monitoring of different levels of indicators; gathers additional data before, during, and after the period to be assessed by multiple methods; uses several analytical techniques to deal with various data gaps and biases; and includes interim and summative evaluation analyses. This new approach will promote country ownership, transparency, and donor coordination while providing a rigorous comparison of the cost-effectiveness of different scale-up approaches.
Introduction The Millennium Development Goals (MDGs) have stimulated interest and increased funding for programmes aimed at reduction of maternal and child mortality and the burden of HIV/AIDS, tuberculosis, and malaria. At the same time, the realisation that few programmes and initiatives have been evaluated properly,1–3 and interest in results-based financing approaches,4 are increasing pressure on implementers to undertake effectiveness evaluations. A common evaluation framework is needed to allow future comparison of the performance of different initiatives—measured in terms of increasing coverage and achieving health effects—and their cost.5 Such a framework should include: (1) a conceptual model outlining pathways through which the initiative is expected to affect the MDGs; (2) a list of standard indicators of inputs, processes, outputs, outcomes, and impact, with clear measurement plans; and (3) guidelines for design of evaluations in a compatible way. Much progress has already been made on the first two objectives.6 This Health Policy article focuses on the third topic, reporting on evaluations of large-scale public-health programmes. Our objective is not to establish the efficacy of new biological or behavioural interventions, or of packages of such interventions; these aims are best achieved with randomised controlled trials. We are interested in assessment of how well large-scale complex programmes deliver efficacious interventions, using different delivery channels in various health-system contexts. Programme success is defined as gains in intervention coverage and in health effects under real-world conditions, when implementation tends to be less intense and more variable than in efficacy trials.7 Observational designs are needed because evaluators cannot control where, when, and how rapidly programmes will be implemented at scale by governments, international or bilateral agencies, and private voluntary organisations. www.thelancet.com Vol 377 January 1, 2011
Design of health-programme evaluations has been dominated traditionally by experimental approaches used in medicine, in which specific individuals or clusters of people receive an intervention whereas others do not. Studies tend to be undertaken in controlled environments in which the influence of external factors is kept to a minimum or eliminated. In the real world, however, the intervention or programme of interest usually accounts for only a small part of variability in health outcomes. Figure 1 presents a simplified framework showing that maternal and child health outcomes can also be affected by socioeconomic and contextual factors, by changes in existing health services in the public and private sectors that are outside the scope of the programme of interest, and by other initiatives or interventions in health or other sectors present in the same geographic areas. Because changes in all the above factors can be happening concurrently with implementation of the programme under assessment, real-world effectiveness evaluations present challenges that cannot be properly addressed by the traditional approach of intervention versus comparison group. In addition to the reality that programmes are not scaled up in a vacuum, they also rarely start from a blank sheet. Pre-existing baseline levels and, particularly, trends in key indicators need to be taken into account. We start by describing typical approaches to evaluation of large-scale programmes and move on to propose a new approach that addresses the mosaic of concurrent programmes and initiatives characteristic of most low-income countries with high rates of maternal and child mortality. We draw heavily on an evaluation study published in The Lancet,8 the Multi-Country Evaluation of Integrated Management of Childhood Illness (MCE-IMCI),9,10 and subsequent efforts to develop designs for independent assessments of the Catalytic Initiative to Save a Million Lives,11 including the three-country rapid scale up of maternal, newborn, and child health funded by the Bill & Melinda Gates Foundation via WHO (the Partnership for Maternal, Newborn, and Child Health).
Lancet 2011; 377: 85–95 Published Online July 9, 2010 DOI:10.1016/S01406736(10)60810-0 Post-Graduate Program in Epidemiology, Federal University of Pelotas, Pelotas, Brazil (Prof C G Victora MD); Institute for International Programs, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA (Prof C G Victora, Prof R E Black MD, J Bryce EdD); and World Health Organization, Geneva, Switzerland (J T Boerma MD) Correspondence to: Prof Cesar Victora, Federal University of Pelotas, Pelotas, Brazil
[email protected]
For the Partnership for Maternal, Newborn, and Child Health see http://www.who.int/ pmnch/en/
85
Health Policy
General socioeconomic and other contextual factors (eg, poverty, employment, food production, fertility)
Interventions in other sectors (eg, water, sanitation, education, transportation)
Programme
evaluation finding. Second, in case of several units of analysis, such as districts, provinces, or countries, a dose–response analysis that examines (for instance) the association between funding levels and service coverage could further contribute to the evaluation.
Comparison of programme and non-programme areas Routine health services (eg, health facilities, hospitals, human resources)
Other health programmes (eg, family planning, food distribution, bednets)
Coverage (eg, proportion of the target population receiving the interventions delivered by the programme)
Impact (eg, under-5 mortality, undernutrition, maternal mortality)
Figure 1: Outline of factors affecting maternal and child health and nutrition
Common approaches to large-scale evaluations Before and after assessment Programme areas only This evaluation design is one of the simplest and entails recording changes over time in the intervention area, in terms of indicators of health service provision, utilisation, coverage, and impact. These evaluations usually attempt to note goals in terms of coverage (eg, 80% of children aged 6–59 months receiving two doses of vitamin A in the previous 12 months) or impact (eg, a 25% reduction in mortality over a specific period), and have been described as adequacy designs, because evaluators focus on assessment of how well the stated programme goals are met.7 The counterfactual for this type of evaluation is that changes would not have happened without the programme, that is, indicators would not have improved because of pre-existing routine health services, other programmes in health or related sectors, or in broader determinants of health—eg, socioeconomic or environmental factors (figure 1). The major limitation of before and after designs is that, without a comparison group, there is no way to determine whether changes are attributable to the programme or to other factors. This type of evaluation continues to be popular despite its limitations, particularly with implementers undertaking internal assessments of their own programmes. Such studies are better than no assessment, if results are interpreted with the necessary caveats. They are especially useful when no changes are found in coverage or impact indicators, which suggests that the programme needs to be reformulated or (rarely) managed to offset the negative effect of other determinants of health, such as natural or man-made crises. This design can be extended in several ways. First, if data are available on preintervention trends, the counterfactual is continuation of these trends. Favourable changes in indicators after initiation of interventions are an important 86
This design has been used frequently in rigorous programme evaluations. It allows investigators to increase plausibility of a causal effect by inclusion of an external comparison to assess whether trends in coverage and impact indicators differ in programme and non-programme areas. A variant of this approach is the stepped-wedge design,12 in which the new programme is implemented gradually, ideally with random allocation, and areas without the programme constitute the comparison group up until the time they start implementation. The counterfactual is that programme areas, had the programme not been implemented, would show trends similar to those seen in non-programme areas. Returning to figure 1, this design assumes that background conditions (eg, socioeconomic and environmental factors), levels of existing health services (eg, number of health facilities or staffing patterns), other health programmes (eg, mass immunisation campaigns), or non-health-sector initiatives (eg, water and sanitation) would evolve similarly over time in both programme and non-programme areas. Four scenarios highlight the limitations of this design in view of the current public-health landscape in most low-income countries. These scenarios are illustrative and not mutually exclusive. First, the untouched comparison area scenario assumes that comparison areas do not have the programme under evaluation, and that neither intervention nor comparison areas have other health programmes affecting intervention coverage or mortality. Intervention and comparison areas are assumed to be similar in terms of public and private health services and of non-healthsector interventions, and with respect to the broader determinants of health. Of the four scenarios presented here, this one is the closest to the experimental approach, but it is increasingly uncommon because many programmes are now being implemented in most if not all districts in low-income countries. So-called virgin comparison districts, if they ever existed, are unlikely to exist now. Furthermore, global health partnerships—such as the Global Fund to Fight AIDS, Tuberculosis and Malaria (Global Fund) and the Global Alliance for Vaccines and Immunization (GAVI)—generally support programmes with national coverage. Second, the parallel health programmes scenario takes into account that health programmes other than the one being assessed—but addressing the same causes of morbidity and mortality—are underway in both intervention and comparison districts. An example is evaluation of IMCI in Tanzania.13 The main component of this www.thelancet.com Vol 377 January 1, 2011
Health Policy
intervention was improved case-management of malaria, pneumonia, and diarrhoea in government health facilities. At the same time, other programmes were present in both intervention and comparison areas, including mass distribution of vitamin A and social marketing of insecticide-treated nets. The IMCI evaluation measured coverage of these parallel programmes and took them into account when interpreting impact results, but complete separation of effects was impossible because vitamin A and insecticide-treated nets were also being promoted by IMCI. Third, the rapid socioeconomic progress scenario builds on the previous setting but also recognises that health outcomes are affected strongly by socioeconomic progress. Before introduction of efficacious biological interventions currently promoted in countries of low and middle income, the striking decline of under-5 mortality in high-income countries in the first half of the 20th century attested to the importance of economic progress, nutrition, sanitation, and improved living conditions.14 In today’s world, socioeconomic progress might boost child survival in countries of low and middle income not only through the pathways depicted in figure 1 but also by increasing access to antibiotics and services provided by the private sector, improved transportation (eg, motorcycles and better roads), and communications (eg, mobile phones), all of which can result in better access to health services. A rigorous before-and-after design with a comparison group will measure socioeconomic and other contextual factors at the beginning and end of the evaluation, identify imbalances between the two groups, and attempt to take these into account when analysing results. However, socioeconomic progress and mortality declines can be so rapid that an additional effect of the programme under evaluation might be impossible to detect. The Bangladesh IMCI evaluation provides a good example.10 This study used a cluster-randomised design with intervention and comparison areas to show annual declines in under-5 mortality of 8·6% and 7·8%, respectively. Such massive rates of decrease were not foreseen when the study was designed in 1999, because before that time annual rates of mortality decline were less than 2%. In addition to socioeconomic progress, Bangladesh has wide availability of lifesaving drugs in a thriving private sector, and strong vertical programmes (eg, oral rehydration therapy, vitamin A, vaccination, etc) that were present in both IMCI and comparison areas. This example shows how the effect of a particular intervention can be difficult to detect with rapid socioeconomic and general health service improvement. Contextual epidemiological changes can also affect evaluation. An example is assessment of a set of preventive HIV interventions in rural Zimbabwe during a period of rapid secular decline in HIV prevalence and incidence.15 The secular decrease could have been related to general behavioural changes or a phase of www.thelancet.com Vol 377 January 1, 2011
reduced HIV transmission in the epidemic, because populations at highest risk have died. Under these circumstances, a carefully planned local intervention study is much less likely to be able to record a significant effect on HIV transmission.16 Contextual epidemiological changes can also affect evaluations for other diseases such as malaria, which is affected by rainfall patterns, and measles or meningitis, which have multi-year cyclic patterns of transmission. Finally, the Balkanisation scenario results from the fact that various development partners and international organisations agree—usually with government consent—to support programmes in specific provinces or districts within a country. This process of Balkanisation means that different partners promote and support similar interventions in different geographic areas, sometimes with variable approaches to intervention delivery. Comparison of districts receiving a given programme with either the rest of the country or with another set of neighbouring districts could end up comparing very similar sets of interventions. This situation arose with evaluation of the UNICEF Accelerated Child Survival and Development (ACSD) programme in Mali, which showed a 24% fall in under-5 mortality during the intervention period in the ACSD focus districts and a 31% decline during the same period in the national comparison area.8 Documentation of programme implementation and contextual factors showed that other funding agencies (mainly the US Agency for International Development [USAID]) were supporting scale-up of some of the same maternal and child survival interventions as ACSD in many comparison area districts, suggesting that the presence of ACSD had displaced another programme that was at least as effective, or possibly more so. Because an evaluation focused on one programme almost never gathers information on another, an intervention-comparison design might fail to detect that comparison areas were, in fact, profiting from a more effective programme.
A way forward: the evaluation platform design In the preceding section, we addressed limitations of intervention-comparison designs in contexts for which many programmes and interventions are being scaled up simultaneously. We argued that a reductionist approach to evaluation based on isolation of programme effects is no longer appropriate for scaling up of initiatives to reach the MDGs in most low-income countries. What is the alternative? How can evaluations of large-scale programmes be designed in ways that are scientifically rigorous and yet reflect the real-world context? First, one must stop thinking about comparison of a few programme districts (or populations) with a handful of others that do not have that particular programme—because the comparators are likely to have, to some degree, initiatives similar to those under scrutiny. Second, one must try to understand why certain 87
Health Policy
Tanzania
Cabo Delgado
Zambia
Pemba
Niassa
Nampula
Tete
Zambézia
Manica Zimbabwe Chimoio
Quelimane Sofala
Inhambane Gaza
Hilde De Graeve and Bert Schreuder
South Africa Xai-xai
Stop TB, the US President’s Emergency Plan for AIDS Relief (PEPFAR), etc—attempt to cover the whole country but are implemented with variable intensity in each district or province. All these programmes merit independent evaluation to inform the Mozambican people and programme funders about their effectiveness and, where possible, value for money. Yet, in a review of several evaluations, the effectiveness of most health programmes is not assessed at all, and even when evaluations are undertaken they generally lack scientific rigor.2 In the rare event that more than one programme in the same country is assessed rigorously, results of the different evaluations can be difficult to compare because the same indicators were not used or methods were inconsistent. The new evaluation platform proposed here has several features. First, it uses the district as the unit of design and analysis. Second, it is based on continuous monitoring of the different levels of indicators. Third, additional data are gathered before, during, and after the period to be evaluated by various methods. Fourth, a range of analytical techniques are used to deal with data gaps and biases. Finally, interim and summative evaluation analyses are undertaken. Panel 1 provides details of an evaluation platform approach that was developed for Mozambique. Discussion on the design and applications of the evaluation platform is focused on countries of low and middle income, in contexts for which several agencies and partners are likely to be active in different regions of the country. The main concepts of the platform design are also relevant to developed countries, and indeed have already been incorporated in ecological studies of programme impact on the basis of existing data (eg, vital statistics).
Districts as units
Figure 2: Focus districts for selected development partners in maternal and child health, Mozambique, 2008
programmes are implemented in some areas rather than others, because this reason could affect how likely they are to succeed. Finally, one must be able to answer one of the most important questions of all: which of the various programmes or delivery approaches implemented by different partners works best in a given country? An improved approach to evaluation will build heavily on existing monitoring data and complement them as needed, with thorough quality checks, additional data collection, and enhanced data analyses. Figure 2 shows the geographic distribution by province of health programmes supported by partners in Mozambique in 2008. Most of these programmes cover a few districts within every province. Moreover, several initiatives—such as the Global Fund, GAVI, 88
The preferred unit of study design and analysis for evaluations is usually the district, because this is the core administrative unit for government health and other programmes in many countries. Districts are easily identifiable geographically and typically have some level of sociocultural or economic homogeneity. Population sizes vary from 100 000 to 500 000 in most nations. Some countries have subdistricts as the smallest administrative units and these might serve as units for the evaluation. Data for a range of indicators from programme inputs to service delivery and coverage can be gathered at the district level. Several limitations to this idea exist. Use of services does not follow district boundaries strictly, because people can cross boundaries to obtain the nearest or better services. Some countries redefine district boundaries periodically. District sizes vary widely by country and, in most cases, aggregation of districts is necessary to obtain adequate numbers. Finally, health status indicators, and some coverage indicators, might www.thelancet.com Vol 377 January 1, 2011
Health Policy
Panel 1: Planning the evaluation platform design in Mozambique Mozambique provides us with examples of practical opportunities and challenges of the platform design. Many agencies are present in different parts of the country (figure 2), and several existing databases can provide data at district and provincial level that are useful for evaluation. The platform design can be used to evaluate different initiatives. The reach every district (RED) approach, coordinated by the Ministry of Health, started implementation of a selected set of maternal and child interventions in 33 districts in 2008, with support from different partners. These interventions include longlasting insecticide-treated mosquito nets, immunisations, breastfeeding promotion, vitamin A supplementation, and integrated management of childhood illnesses in health facilities. New cohorts of 33 districts will start implementation every year until all 148 districts are covered, allowing a non-randomised stepped-wedge analysis to be done, building on the evaluation platform, which will include baseline data for contextual factors, coverage, and mortality levels and will allow analyses of coverage and health impact outcomes relative to programme implementation strength. A separate initiative, the rapid scale-up of the partnership for maternal, newborn, and child health, is concentrating its efforts in 12 districts. Since information on all districts in the country will be gathered, the platform approach could be used to evaluate the two programmes described above and other existing or future initiatives. To complement official data available at national level, focal points will be hired in each of the 11 provinces (each with about 13 districts) to report monthly on changes in health facilities, training of facility and community health workers, supervision, and staffing patterns. They will also report on distribution of commodities (eg, insecticide-treated nets, drugs, vaccines, nutrition supplements, etc), provided by governmental and non-governmental sources. For quality control, district-level information obtained by the provincial focal points will be cross-checked with official data provided by sources at national level and with denominator data (eg, population size); discrepancies will be identified and discussed with health managers to contribute to data quality improvement over time. Panel 2 shows an initial list of available variables from different data sources in Mozambique. The proposed network of informants at provincial and district level will help update information on contextual factors in the database and report unpredicted events, such as food shortages or natural or man-made emergencies that could affect maternal and child health. (Continues in next column)
www.thelancet.com Vol 377 January 1, 2011
(Continued from previous column) Coverage data will be obtained from existing and future surveys. The 2003 demographic and health survey (DHS) and 2008 multiple indicator cluster survey included more than 12 000 and 14 000 households, respectively, in their samples. In 2008, over 3000 households belonged to the 33 districts included in the first wave of the RED initiative. This number will provide a sufficient sample for precise measurement of intervention coverage for all key maternal and child health interventions. A new DHS is planned for 2011, which can provide additional coverage information and allow measurement of changes. A major challenge for the evaluation platform in Mozambique is to strengthen local capacity for data collection, management, and analyses. Another, perhaps even greater, challenge is to ensure cooperation and support of the many local and international partners with a role in implementation of health programmes.
only be available or sufficiently precise at higher levels of aggregation, such as provinces or subnational regions. The manner in which districts are selected to receive the programme under evaluation needs to be taken into account. Documentation of criteria that guided this selection is essential.
Monitoring and databases Evaluation needs a continuous, strong monitoring effort that entails careful and systematic documentation of contextual variables, health system inputs and service delivery, intervention coverage, risk factors, and health status. This procedure requires strengthening of country efforts to build district health information systems. Most low-income and middle-income countries already have several databases maintained by governmental, international, or partner institutions (panel 2). These might include data for health inputs (financing, human resources), health facility reports (services provided), facility assessments (geocodes, drug availability, etc), the socioeconomic and demographic situation (including poverty maps), and partner presence. These databases contain information disaggregated at provincial or state level, and sometimes at district level. We propose that relevant information from different existing databases should be integrated in a continuous manner to lay the foundation for evaluations. The main data repository for the evaluation platform is a database in which geographic units constitute the rows, and indicators that are relevant to the evaluation make up the columns—one for every year for which data on that indicator are available. New information about programme implementation by different agencies (government, bilaterals, multilaterals, non-governmental organisations [NGOs]) would also be added. In most countries, the evaluation platform would include 89
Health Policy
Panel 2: Examples of data available from existing databases at district and provincial level in Mozambique, and sources of information Socioeconomic factors • Household assets • Family income and poverty • Parental education and occupation • Unemployment • Land tenure Sources: 2007 census; economic censuses and surveys • Economic crises (inflation rates, crop failures, floods) Source: National Institute of Statistics Demographic factors • Population density • Fertility patterns • Family size • Ethnic groups Source: 2007 census Environmental characteristics • Water supply • Sanitation • Urbanisation • Housing Source: 2007 census • Rainfall • Altitude Source: National Meteorological Institute Baseline health characteristics • Under-5 mortality • Prevalence of malnutrition • HIV prevalence • Malaria transmission patterns Sources: 2007 census; 2008 MICS; malaria and HIV surveys Health services characteristics • Availability of health services (hospitals, clinics, etc) in governmental and private sectors • Population/facility ratio • Health worker staffing patterns • Health worker pay • Drug supply • Baseline utilisation rates • Availability of referral services • Strength of district health management team • District health budget (overall and for child health) Sources: Health Metrics Network; Ministry of Health Information Systems; UNFPA Needs Assessment Survey; WHO Service Availability Mapping (Continues in next column)
information about all districts, thus allowing assessors to understand how programmes are being deployed. However, the platform can also be restricted to a subnational region in which health and development 90
(Continued from previous column) Presence of other projects and programmes that could affect health status • Micronutrients • Indoor residual spraying • Immunisations • HIV programmes • Others Sources: UNICEF; WHO; NAIMA; ODAMOZ MICS=multiple indicator cluster survey. UNFPA=UN Population Fund. NAIMA=Network of Organizations working in Health and HIV/AIDS. ODAMOZ=Official Development Assistance to Mozambique Database.
programmes are concentrated, particularly in the case of large countries with striking regional disparities. For example, a platform-like approach was used in northeast Brazil to evaluate the effectiveness of IMCI.17,18 The evaluation platform will not replace existing databases, but instead will build on and contribute to their improvement. It could be described as a comprehensive database that covers information on all relevant monitoring indicators, pays systematic attention to data quality, and includes contextual and qualitative information to answer evaluation questions.
Additional data collection Although solid monitoring systems provide essential information on trends and indicators, the platform design will need additional data collection to record programme implementation, assess data quality and make statistical adjustments, fill data gaps, and address specific evaluation questions. Various methods might be required, such as health-facility assessments, household surveys and oversampling of districts, longitudinal designs, and qualitative research. The prospective design of the platform provides greater opportunities to gather essential additional data than retrospective studies, when only post scale-up data can be obtained.8 Detailed documentation of inputs, training, supervision, quality of care, and delivery channels is essential for understanding why programmes succeed or fail. This component of health evaluations is frequently neglected because efforts to measure coverage or mortality consume most of the assessors’ time and resources. Data for implementation can be obtained at the national level (eg, Ministry of Health, international agencies, or NGOs supporting the programme), but records of what is actually reaching the population at district level are also important. For example, the IMCI evaluation showed that even simple data for training coverage are sometimes hard to obtain. The numerator—how many health workers were trained in every district—is usually known, but to find out whether workers remained in their posts after training is difficult, because turnover tends to be high in many countries. Even the denominator for training coverage—the total number www.thelancet.com Vol 377 January 1, 2011
Health Policy
of health workers in the cadres eligible for training—can be difficult to ascertain at national level.19 In the planned platform design in Mozambique (panel 1), key informants at provincial level will record programme implementation and check the quality of information available at national level. Documentation of all programmes being implemented in every district permits researchers (with this platform design) to test the possibility that a given initiative could have displaced a more effective one, as discussed in the Mali ACSD example.8 Programme performance scores can provide a quantitative measure of implementation strength on the basis of information gathered through the documentation process. For example, the score for a programme promoting community case-management of pneumonia could be derived from the number of community health workers trained per population, availability of drugs, and the intensity of supervision. These scores can be used to measure the so-called dose of a programme in dose–response analyses. Documentation of costs associated with programmes is also important and can be done by obtaining accurate information on government investments and those by international and bilateral agencies at country level. Donor atlases are available in some nations and can contribute useful information on aid flow by province.20 Disaggregation at district level is also possible.
Analyses of existing survey data Measurement of intervention coverage in low-income countries requires population-based surveys. Demographic and health surveys (DHS)21 or UNICEF’s multiple indicator cluster surveys (MICS) are undertaken every 3–5 years in most low-income countries, and an increasing number of additional surveys are being implemented to assess broader development indicators (eg, living standards measurement studies surveys)22 or specific vertical disease programmes (eg, malaria indicator surveys,23 HIV/AIDS indicators surveys).24 The frequency of such surveys is likely to increase as the MDG 2015 deadline approaches. In most countries, survey samples are insufficient to provide precise estimates of coverage at district level. Rare exceptions include Malawi’s 2010 DHS with 1000 households per district,25 and India’s reproductive and child health surveys with a total sample of more than 600 000 households.26 However, even if the number of sampled households per district is small, few programmes are implemented in just one district, and pooling across several districts could result in sufficient numbers of individuals to assess coverage. Generally, researchers believe that survey results should not be pooled for groups of districts, unless these districts were defined a priori, because few national surveys are designed to provide probability samples at district level. We question this logic, especially since survey reports systematically break down national www.thelancet.com Vol 377 January 1, 2011
results according to age, socioeconomic, and ethnic categories, even though the sample was not designed to be strictly representative of such subgroups. In practice, most nationally representative surveys introduce implicit stratification within every district by listing enumeration areas in a geographic sequence and systematically sampling these areas; as a result, households included in the sample tend to be spread throughout the districts. With due attention to sampling weights, groups of districts in which a programme was implemented can be separated from a national survey for the assessment of coverage.27,28 Dose–response analyses of programme implementation strength and coverage are also possible; these are discussed below (see Analyses for programme enhancement and evaluation). For programmes covering a few districts for which even the pooled sample size is insufficient, coverage estimates will need national DHS or MICS with oversampling, as was done in the ACSD evaluation, or separate surveys will have to be done, using comparable methodologies in programme districts. The primary indicator for MDG 4 (reduction of under-5 mortality by two-thirds) is the mortality rate in children younger than 5 years, and for MDG 5 (reduction of maternal mortality by three-quarters and universal access to reproductive health) it is the maternal mortality ratio. Most DHS and some MICS include full birth or pregnancy histories, in which women of reproductive age report on how many children they ever had, their dates of birth, whether they have died, and if so the dates of death. This information allows retrospective construction of annual mortality rates for young children for a period of up to 10 calendar years before the survey, calculated similarly to the 5-year rates published by DHS.29 The sample size limitations discussed in the context of coverage estimation are even more vital for estimates of under-5 mortality; in this case, oversampling of selected districts with a programme when a national survey is undertaken, pooling results of such surveys across several districts, or doing stand-alone surveys might be needed. In the ACSD evaluation,8 national DHS were oversampled in the programme districts in Benin and Mali, and this action allowed comparison of trends in mortality in these areas to those in the rest of the country. A commonly used impact indicator is prevalence of undernutrition (underweight, stunting, and wasting) in children younger than 5 years, which is related to MDG 1 (eradication of poverty and hunger). Sample size issues are not as important for these indicators because their frequency tends to be high and because their denominator includes all children younger than 5 years in the sample. Use of mean values of weight-for-age, height-for-age, or weight-for-height Z scores further reduces need for large samples, compared with estimation of the prevalence below a given cutoff.30 Maternal mortality ratios can also be estimated on the basis of surveys. However, the required sample sizes are 91
Health Policy
Panel 3: Examples of questions answerable by the platform approach Interim evaluation questions (early in the implementation cycle) Are programmes being deployed where need is greatest? Implementers usually assert that their programmes are being deployed in areas of the country where mortality is highest or poverty most frequent, but this statement is not always true. Indicators of baseline mortality and poverty levels and of the strength of health systems available in the platform database will be linked through the platform to data for implementation, thus supporting assessments of whether programmes are indeed deployed where they are most needed. An example of how useful such simple analyses can be comes from the Integrated Management of Childhood Illness (IMCI) evaluation in Brazil, where implementation was stronger in municipalities that were close to their state capitals than in those that were poorest and had high mortality—an example of placement bias of health programmes.18 In the west African Accelerated Child Survival and Development evaluation, some countries implemented the programme in the neediest areas, but others did not.8 Is implementation strong enough to have an impact? Insufficient implementation is a common reason for absence of impact. Analyses of implementation data gathered at both national and local level will allow linking of inputs and outputs to target populations in the evaluation platform, thus obtaining an estimate of the strength of implementation. Although a high ratio of outputs to population is not necessarily indicative of high coverage, because there could be leakages or other inefficiencies, low ratios surely suggest that implementation is unlikely to lead to a measureable impact. Simulation exercises can contribute to answering this question. The lives saved tool (LiST), which allows users to estimate the impact of changes in coverage for proven interventions on maternal, neonatal, and under-5 mortality,31 is a useful adjunct to these analyses. What approaches lead to rapid coverage increases in the short term? Different implementing agencies and partners generally rely on diverse delivery channels for increasing coverage—eg, facility-based approaches, outreach sessions, community health workers, involving the private sector, community groups, etc.32 Interim analyses of trends in coverage based on mid-term surveys, linked through the platform to implementation data, could help assess which of these approaches are most effective in the short term. Data gathered through the platform might also identify districts that are doing well and those that are lagging behind, and motivate further in-depth analyses with a so-called mixed-methods approach to improve implementation.33 Summative evaluation questions (at the end of a programme cycle) Are coverage rises sustained? Analyses of short-term coverage increases, associated with specific delivery channels (as discussed above), must be complemented by longer term assessments of whether these rises are sustained. This research is especially important in view of findings on the damaging effects of stock-outs of essential commodities (such as drugs or insecticide-treated nets), shifting priorities and funding patterns, and other barriers to continuity of a programme.8,34,35 The evaluation platform design will support analyses of coverage trends in areas with every type of programme, recording associations between the intensity of implementation activities and coverage levels. Indices of programme effort can be used to summarise implementation strength.6 (Continues on next page)
very large and estimates refer to a period that is too far backdated to be useful for evaluation. As an alternative, coverage of maternal interventions can be measured and used as a proxy for mortality. Household surveys are an important source of data relating to progress towards the goals of MDG 6 (to halt 92
and reduce spread of HIV/AIDS, malaria, and other diseases and provide universal access to treatment for HIV/AIDS). They might include data for coverage of malaria interventions (such as use of insecticide-treated bednets, indoor residual spraying, intermittent preventive therapy during pregnancy, and treatment of children with fever) and selected HIV interventions (such as HIV testing coverage in adults and pregnant women attending antenatal clinics). Surveys with biological data collection can provide information about prevalence rates of HIV infection and parasitaemia. Other indicators, such as tuberculosis treatment success rates and use of antiretroviral therapy, are based on reports from health facilities. Disaggregation of survey data at district level might not be possible if information about geographic location of every cluster is deleted from the database—a process known as scrambling. This procedure is common when sensitive data are obtained (such as HIV serology). Deletion of information will severely limit the potential usefulness of surveys for evaluation of all outcomes, including those unrelated to AIDS, and in our view this limitation is strong justification for not including serological status in multipurpose surveys.
Analyses for programme enhancement and evaluation The evaluation platform will ideally summarise all types of available data for factors affecting outcomes of interest (figure 1), including baseline and regular updates on information related to geographic, epidemiological, socioeconomic, demographic, and other relevant characteristics of every district. It will also include variables related to programme intensity—in particular, implementation scores—and coverage and impact data. The new design allows for various analytical techniques. First, data can be pooled from several districts to obtain robust coverage and health outcome estimates. Second, dose–response analyses can be undertaken, in which every district is a datapoint. Third, poor baseline comparability attributable to so-called placement bias (or the fact that districts receiving the programme might differ from other districts in the country) can be dealt with. Fourth, consistency of indicators can be assessed across different levels of the framework (figure 1)—eg, relation of programme implementation scores, coverage, and impact indicators to each other. Fifth, modelling exercises can be implemented, such as those supported by the lives saved tool (LiST).31 Finally, changes in contextual factors can be incorporated into analyses. Districts are the primary units of statistical analysis for the national evaluation platform approach. We envisage two main types of data analysis: (1) interim (or formative) analyses could be undertaken as soon as the database is set up with process indicators and results fed back to implementers to allow mid-course corrections; and (2) endline (or summative) analyses could be done at the www.thelancet.com Vol 377 January 1, 2011
Health Policy
end of a programme cycle to assess how it affected coverage and impact indicators. Interim (or formative) analyses based on the evaluation platform approach can produce results that not only are useful to governmental health managers (and all agencies that have a role in implementation) for making mid-course corrections but also increase the probability that health system barriers to full implementation are identified and addressed. Feedback based on interim analyses is an important departure from traditional evaluation approaches, in which external evaluators are advised to refrain from interfering with programmes under assessment. Double-blinding, which is strongly recommended for testing new interventions, is not feasible in the context of evaluation of scale-up of proven interventions. At the end of a programme cycle, the evaluation platform can enable summative questions to be answered. The timeline for assessment of a programme usually takes 5–7 years to complete, including time needed for full implementation of the programme, for the biological effect of the intervention to take place, and for impact to be measured. Panel 3 shows illustrative questions and approaches to interim (or formative) and summative analyses. The national evaluation platform design will also help answer broader policy-relevant questions that are sometimes difficult to address with traditional designs. Examples include assessment of the equity effects of a programme (which can be undertaken by analysis of coverage and impact, broken down by sex, socioeconomic status, urban or rural residence, and ethnic group) and investigation of unintended effects of programmes (by looking at other health outcomes in the programme area). In view of its multi-programme, multiple-outcome nature, the platform is ideally suited to answering such questions or those related to programme sustainability. If the number of districts permits, researchers could also assess whether the effect of a programme differs according to district characteristics, such as baseline levels of mortality or socioeconomic development. A national evaluation platform, maintained over the long term, can provide answers to questions that cannot be obtained from short-term, single-programme evaluations.
Sample size Calculations of sample size and statistical power for the evaluation platform will depend on the types of comparisons planned. We begin by considering use of the platform to undertake a traditional comparison of districts with and without a given programme. Because the number of districts is fixed and analyses will rely on existing surveys (in which the number of households per district is also fixed), sample size calculations will allow estimation of study power, or the likelihood that a true effect is picked up in analyses. Evidence of low power will www.thelancet.com Vol 377 January 1, 2011
(Continued from previous page) Did programmes have an impact? By linking implementation data to information on mortality and nutritional status, the platform design will allow assessment of a programme’s impact. For example, in IMCI evaluations in Peru and Brazil, training coverage of health professionals was not associated with child mortality indicators in dose–response analyses.17,36 Variable strengths of programme implementation in different districts will favour ecological, dose–response, time-trend analyses in the platform design. However, the more traditional approach of comparing areas with and without each programme is also possible. Was coverage associated with impact? Although the answer to this question might seem obvious, reasons why increased coverage might not have an effect do exist, such as wrong choice of interventions in view of the epidemiological profile of the population, lack of essential cofactors, or low quality of intervention delivery. Dose–response time-series analyses correlating coverage levels with measures of impact across all districts in the country will help answer this question. Simulations with LiST31 will be especially useful, by allowing comparison of estimated impact on the basis of coverage changes with actual reductions in mortality or undernutrition. Do alternative explanations exist for the findings? The wealth of data for contextual factors and their change over time available in the platform can help rule out alternative explanations for impact findings, enhancing the plausibility that the noted effect is attributable to a given programme.7 This process can be done by incorporation of relevant contextual factors as confounding variables in regression models, as was done in the Brazil and Peru IMCI evaluations.17,36 Which programmes were most cost effective? This question will need information on costs gathered as part of programme documentation related to health impact results.
require enrolment of increased numbers of households within each district, or oversampling. Even if oversampling is not feasible and study power is less than what would be desirable, the platform approach can still contribute to interpretation of results by providing additional information on placement bias, confounding variables, and presence of similar programmes in the comparison areas. If dose–response analyses are used, which will usually be the case, then sample sizes will be affected by how much the implementation score varies across the different districts, how much variability there is in baseline coverage, and how strongly implementation affects coverage. Calculations for an evaluation platform design in Malawi suggested that there would be 80% power for detection of a change in coverage of antibiotic treatment for pneumonia in the presence of a community case-management programme, in view of the following assumptions: 28 districts in the country, samples of 1000 households per district, and an increase in coverage of seven percentage points for every ten percentage point increase in the implementation score. Malawi is a special case, with relatively few districts compared with other countries and large survey sample sizes within each district; calculations must be done on a country-by-country basis. 93
Health Policy
Conclusions We propose a systematic approach for evaluation of scale-up of national programmes for maternal and child survival and potentially other public-health programmes that address specific diseases (such as HIV/AIDS, tuberculosis, or malaria). A national evaluation platform shows how ideas have evolved in response to changes in the development landscape in most countries of low and middle income, where governments are working with several partners to implement many simultaneous public-health programmes with overlapping aims and geographic boundaries. Why does this approach make sense? First, setting up a broad platform design that includes documentation of contextual factors and implementation of many programmes—and indicators of coverage, impact, and cost—puts governments in the driving seat and provides them with information needed to make wise decisions about how they should use scarce resources. The platform approach supports country ownership of national programmes and their evaluation, helps build local capacity, and promotes donor coordination in the spirit of the Paris Declaration.37 Second, the approach is likely to be cost effective. Rather than replacing existing databases and data repositories, the platform approach will allow them to communicate with one another, with the specific objective of answering evaluation questions. This strategy can serve as an organising framework for large-scale survey datasets supported by various partners and with several aims, permitting full use of every survey to address broad public-health questions rather than focus on one disease, programme, or population, and complementing other initiatives and proposals for strengthening of district-level data use in support of programmes.38 Although the initial investment in setting up the platform could be fairly small, large surveys will be necessary in some countries to measure coverage and impact indicators with adequate precision for groups of districts implementing a given programme. On the other hand, large surveys that allow assessment of many programmes are likely to be cheaper than stand-alone traditional evaluations of these programmes, each with an intervention and comparison group of districts, which is the modus operandi at present. Even if costs of large surveys needed for the platform approach exceed current evaluation investments, substantial gains will arise in terms of geographic coverage and the scope of findings. Third, the platform design promotes evaluation as a continuous process aimed at improvement of implementation, adapted to the current realities of simultaneous undertaking of several programmes and the resulting absence of untouched comparison areas. The flexible and comprehensive design of the platform allows evaluation to respond to changes in implementation, avoiding potential blows to rigid designs, such as strong implementation of a new 94
programme in a previously defined comparison area. Feedback to all levels and linkage to programming cycles are essential features of our approach. Finally, the platform enhances the independence of evaluators, who are not being commissioned and financed by the one agency whose programme is under scrutiny. It will help governments use the many surveys done in their countries as a source of broader learning about the relative effectiveness and costs of alternative approaches implemented with support from various partners. To ensure independence, the platform should be hosted by a national academic or research institution that is not directly involved in the programmes being evaluated, with a steering committee of national (Ministry of Health, national statistical office, etc) and international (UN agencies, bilateral agencies, NGOs, etc) partners. The research institution would employ and supervise the network of key informants at district level. One limitation of the national evaluation platform approach is its observational design. No other alternative is feasible, however, in view of the focus on evaluation of real-world programmes as they go to scale and the resulting inability of assessors to control the pace and location of implementation. A second limitation is the platform’s reliance on transparency and collaboration between government and development partners, and within many agencies. Funding of an evaluation platform has risks for donors, because the results will reflect the contributions of many and will be in the public domain. Lastly, development of a national information system for programme evaluation that integrates survey results with facility-based and administrative statistics to provide usable district-level information on an ongoing basis is not a trivial matter. It will need an enormous change within countries concerning the way facility and administrative statistics are used. Development of the platform will also require major investments in local capacity building, leading to a shift from stand-alone evaluations commissioned by external parties to country-led assessments that will allow each country to know more about its own programmes. Guidance on how platforms should be made operational is specific to the country and context, and provision of detailed recommendations is beyond the scope of this article. Indeed, implementation of evaluation platforms is an intervention that needs sufficient funding and deserves its own external evaluation. Balancing these limitations is the realism and rigor of the platform design. The current reductionist approach of single-programme, intervention-versus-control-area evaluations has a limited role in the programmatic and development contexts of countries where most avoidable deaths happen. Contributors CGV prepared the first draft of the article, based on discussions with JB and REB arising from their joint work on evaluation of maternal and child health programmes over several years. JTB expanded the article to
www.thelancet.com Vol 377 January 1, 2011
Health Policy
cover other health outcomes, and to place it in the context of national information systems. All authors provided comments on subsequent drafts and approved the final version of the manuscript. Conflicts of interest We declare that we have no conflicts of interest. Acknowledgments We thank Hilde De Graeve and Bert Schreuder for allowing reproduction of figure 2; Baltazar Chilundo, Fatima Abacassamo, and Iná Santos for insights into the adaptation of the method to the Mozambique situation; Jennifer Requejo, Kate Gilroy, and David Peters for comments on the text; Agbessi Amouzou and Larry Moulton for inputs on sample size calculations; and Gareth Jones for comments on the plan of statistical analyses. Members of the Catalytic Initiative Evaluation project group at the Institute for International Programs of Johns Hopkins Bloomberg School of Public Health have contributed in important ways to the experience and thinking described in this report. This work was supported by the Canadian International Development Agency and the Bill & Melinda Gates Foundation through subcontracts to the Institute for International Programs at the Johns Hopkins Bloomberg School of Public Health for independent evaluations of the Catalytic Initiative to Save a Million Lives and of the Rapid Scale Up of the Partnership for Maternal, Newborn and Child Health. References 1 The Lancet. Evaluation: the top priority for global health. Lancet 2010; 375: 526. 2 Evaluation Gap Working Group. When will we ever learn? Improving lives through impact evaluation. Washington, DC: Center for Global Development, 2006. 3 Oxman AD, Bjørndal A, Becerra-Posada F, et al. A framework for mandatory impact evaluation to ensure well informed public policy decisions. Lancet 2010; 375: 427–31. 4 World Bank. Results-based financing (RBF). Nov 11, 2008. http://go.worldbank.org/UDQRQYSTF0 (accessed Feb 8, 2010). 5 Victora CG, Black RE, Bryce J. Evaluating child survival programmes. Bull World Health Organ 2009; 87: 83. 6 Catalytic Initiative to Save One Million Lives. Evaluating the scale-up for maternal and child survival: putting science to work for mothers and children. June 16, 2008. http://www.jhsph.edu/dept/ ih/IIP/projects/catalytic_initiative/Common_evaluation_ framework.pdf (accessed June 29, 2010). 7 Habicht JP, Victora CG, Vaughan JP. Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. Int J Epidemiol 1999; 28: 10–18. 8 Bryce J, Gilroy K, Jones G, Hazel E, Black RE, Victora CG. The Accelerated Child Survival and Development programme in west Africa: a retrospective evaluation. Lancet 2010; 375: 572–82. 9 Bryce J, Victora CG, and the MCE-IMCI Technical Advisors. Ten methodological lessons from the Multi-Country Evaluation of Integrated Management of Childhood Illness. Health Policy Plan 2005; 20 (suppl 1): i94–105. 10 Arifeen SE, Hoque DME, Akter T, et al. Effect of the Integrated Management of Childhood Illness strategy on childhood mortality and nutrition in a rural area in Bangladesh: a cluster randomised trial. Lancet 2009; 374: 393–403. 11 Canadian International Development Agency. The catalytic initiative to save a million lives. Nov 26, 2007. http://www.acdi-cida.gc.ca/ acdi-cida/acdi-cida.nsf/eng/NAD-1249841-JLG (accessed Feb 8, 2010). 12 Brown CA, Lilford RJ. The stepped wedge trial design: a systematic review. BMC Med Res Methodol 2006; 6: 54. 13 Schellenberg JRMA, Adam T, Mshinda H, et al. Effectiveness and cost of facility-based Integrated Management of Childhood Illness (IMCI) in Tanzania. Lancet 2004; 364: 1583–94. 14 McKeown T. The modern rise of population. New York: Academic Press, 1976. 15 Gregson S, Adamson S, Papaya S, et al. Impact and process evaluation of integrated community and clinic-based HIV-1 control: a cluster-randomised trial in eastern Zimbabwe. PLoS Med 2007; 4: e102. 16 Grassly NC, Garnett GP, Schwartländer B, Gregson S, Anderson RM. The effectiveness of HIV prevention and the epidemiological context. Bull World Health Organ 2001; 79: 1121–32.
www.thelancet.com Vol 377 January 1, 2011
17
18
19
20 21
22 23 24 25
26
27
28 29
30
31
32
33
34
35
36
37
38
Amaral J, Leite AJ, Cunha AJ, Victora CG. Impact of IMCI health worker training on routinely collected child health indicators in northeast Brazil. Health Policy Plan 2005; 20 (suppl 1): i42–48. Victora CG, Huicho L, Amaral JJ, et al. Are health interventions implemented where they are most needed? District uptake of the integrated management of childhood illness strategy in Brazil, Peru and the United Republic of Tanzania. Bull World Health Organ 2006; 84: 792–801. Huicho L, Dávila M, Campos M, Drasbek C, Bryce J, Victora CG. Scaling up integrated management of childhood illness to the national level: achievements and challenges in Peru. Health Policy Plan 2005; 20: 14–24. ODAmoz. Mozambique donor atlas 2008. December, 2008. http://mozambique.odadata.ampdev.net/ (accessed June 8, 2010). Measure DHS. Demographic and health surveys: DHS overview. http://www.measuredhs.com/aboutsurveys/dhs/start.cfm (accessed Feb 8, 2010). World Bank. Living Standards Measurement Study. http://go.worldbank.org/IPLXWMCNJ0 (accessed June 8, 2010). Measure DHS. Malaria Indicator Survey. http://www.measuredhs. com/aboutsurveys/mis/start.cfm (accessed Feb 8, 2010). Measure DHS. AIS surveys: AIS overview. http://www.measuredhs. com/aboutsurveys/ais/start.cfm (accessed Feb 8, 2010). Measure DHS. Malawi: standard DHS, 2010. http://www. measuredhs.com/aboutsurveys/search/metadata.cfm?surv_ id=333&ctry_id=24&SrvyTp (accessed Feb 8, 2010). International Institute for Population Sciences (IIPS) and Macro International. National Family Health Survey (NFHS-3), 2005–06: India—volume I. September, 2007. http://www.nfhsindia.org/ NFHS-3%20Data/VOL-1/India_volume_I_corrected_17oct08.pdf (accessed June 28, 2010). West BT, Berglund P, Heeringa SG. A closer examination of subpopulation analysis of complex-sample survey data. Stata J 2008; 8: 520–31. Rao JKN. Small area estimation. Hoboken: Wiley Interscience, 2003. Rutstein SO, Rojas G. Online guide to DHS statistics. October, 2006. http://www.measuredhs.com/help/Datasets/index. htm (accessed June 8, 2010). WHO. Physical status: the use and interpretation of anthropometry: report of a WHO Expert Committee (technical report series no 854). 1995. http://www.who.int/childgrowth/publications/physical_ status/en/index.html (accessed June 8, 2010). Department of International Health, Johns Hopkins Bloomberg School of Public Health. LiST: the Lives Saved Tool—an evidence-based tool for estimating intervention impact. http://www. jhsph.edu/dept/ih/IIP/list/index.html (accessed Feb 8, 2010). Bryce J, el Arifeen S, Pariyo G, Lanata CF, Gwatkin D, Habicht J-P. Reducing child mortality: can public health deliver? Lancet 2003; 362: 159–64. Peters DH, El-Saharty S, Sladat B, Janovsky K, Vujicic M. Improving health service delivery in developing countries: from evidence to action. Washington, DC: World Bank, 2009. Bryce J, Daelmans B, Dwivedi A, et al, on behalf of the Countdown to 2015 Core Group. Countdown to 2015 for maternal, newborn, and child survival: the 2008 report on tracking coverage of interventions. Lancet 2008; 371: 1247–58. Biesma RG, Brugha R, Harmer A, Walsh A, Spicer N, Walt G. The effects of global health initiatives on country health systems: a review of the evidence from HIV/AIDS control. Health Policy Plan 2009; 24: 239–52. Huicho L, Dávila M, Gonzales F, Drasbek C, Bryce J, Victora CG. Implementation of the Integrated Management of Childhood Illness strategy in Peru and its association with health indicators: an ecological analysis. Health Policy Plan 2005; 20 (suppl 1): i32–41. OECD. Paris declaration on aid effectiveness: ownership, harmonisation, alignment, results and mutual accountability. Paris: Organisation for Economic Co-Operation and Development, 2005. Rowe AK. Potential of integrated continuous surveys and quality management to support monitoring, evaluation, and the scale-up of health interventions in developing countries. Am J Trop Med Hyg 2009; 80: 971–79.
95
Case Report
Killing two birds with one stone Gabriel Schembri, Paul Schober Lancet 2010; 377: 96 Department of Genitourinary and HIV Medicine. Leicester Royal Infirmary, Leicester, UK (G Schembri MRCP, P Schober FRCP) Correspondence to: Dr G Schembri, Manchester Centre for Sexual Health, The Hathersage Centre, 280 Upper Brook Street, Manchester, MI3 OFH, UK
[email protected]
For the Patent see http://www. freepatentsonline.com/ y2007/0281911.html
In September, 2008, a 56-year-old man presented to our department with fever, myalgia, pharyngitis, and a maculopapular skin eruption. He reported having unprotected sex with a man 2 months prior to the onset of his symptoms. He was diagnosed with HIV seroconversion illness in view of a high HIV viral load (>14 280 000 copies per ml) and a third generation HIV antibody test that was weakly positive. During routine HIV follow-up in March 2009, his previously normal alkaline phosphatase had risen to 486 IU/L and his alanine aminotransferase was 185 IU/L. His total bilirubin was 15 μmol/L. He had no symptoms except for mild occasional right upper quadrant abdominal discomfort. An abdominal ultrasound scan showed multiple large stones in the gallbladder but was otherwise normal. His antimitochondrial antibody was strongly positive (1:4000 IgG, M2 and M4 positive), consistent with a diagnosis of primary biliary cirrhosis. A liver biopsy showed damaged interlobular bile ducts, with granulomas, and increased amounts of eosinophilic cytoplasm and intraepithelial inflammatory cells, also typical of primary biliary cirrhosis. In April, 2009, his CD4 count was 360 cells per μL, and he was started on tenofovir 245 mg, emtricitabine 200 mg, and efavirenz 600 mg once daily. In June, 2009, owing to central nervous system side-effects, efavirenz was changed to lopinavir 400 mg and ritonavir 100 mg tablets twice daily. Ursodeoxycholic acid was introduced at a dose of 15 mg/kg per day in November, 2009. His liver function tests improved considerably once antiretroviral therapy was introduced (figure). In May, 2010, after 13 months of treatment with anti-retroviral therapy, and 6 months of ursodeoxycholic acid, his liver function tests were normal. His antimitochondrial antibodies decreased to 1:250. Using a branched chain hybridisation assay (Quanti-GeneTM Affymetrix, California, USA) we found that serum from our patient was strongly positive for human betaretrovirus RNA. He was last seen in June, 2010,
ALP (IU/L) ALT (IU/L) Ursodeoxycholic acid started
Se
pt Oc -08 No t-08 v De -08 c Jan-08 Fe -09 b M -09 ar Ap -09 M r-09 a Ju y-09 ne Ju -09 ly Au -09 Se g-09 pt Oc -09 No t-09 v De -09 cJan 09 Fe -10 b M -10 arAp 10 M r-10 ay -1 0
IU/L
Acknowledgments We thank Dr Andrew Mason, and team, University of Alberta, Canada, for performing the test for human betaretrovirus. Contributors GS and PS looked after the patient and wrote the report. Written consent to publish was obtained.
ART started 800 700 600 500 400 300 200 100 0
and was asymptomatic, with no clinical or biochemical evidence of cirrhosis. His HIV virus was completely suppressed and his CD4 count increased to 580 cells per μL. In 2003, a human betaretrovirus closely related to mouse mammary tumour virus was cloned from the biliary epithelium from patients with primary biliary cirrhosis.1 Viral infection of cholangiocytes with human betaretrovirus results in the aberrant expression of the mitochondrial E2 component of the pyruvate dehydrogenase complex. This protein, normally located on the inner mitochondrial membrane, appears on the cell surface in patients with primary biliary cirrhosis, and is thought to trigger the appearance of antimitochondrial antibodies.1 The relation between betaretrovirus infection and primary biliary cirrhosis is controversial. It has been suggested that randomised controlled trials with antiviral therapy may provide supportive evidence for a causal role of viral infection.2 Randomised controlled trials using combination lamivudine 150 mg and zidovudine 300 mg twice a day, have shown biochemical improvements in patients with primary biliary cirrhosis, but failed to achieve significant endpoints, possibly because of the development of viral resistance to therapy.3 Other antiretrovirals that have proven effective against betaretrovirus in vitro, include: adefovir, tenofovir, and lopinavir.4 Here, we report that the combination of tenofovir, emtricitabine, lopinavir, and ritonavir used to manage HIV infection, coincided with marked reduction of cholestatic liver function tests, before the institution of ursodeoxycholic acid. If an association between primary biliary cirrhosis and human betaretrovirus infection can be established, the efficacy of this highly active antiviral regimen can be tested in future controlled trials in patients with primary biliary cirrhosis unresponsive to ursodeoxycholic acid.
Time (months)
References 1 Xu L, Shen Z, Linsheng G, et al. Does a betaretrovirus infection trigger primary biliary cirrhosis? PNAS 2003; 100: 845–59. 2 Poupon R. Retrovirus infection as a trigger for primary biliary cirrhosis? Lancet 2004; 363: 260–61. 3 Mason AL, Wasilenko ST. Other potential medical therapies: the use of antiviral agents to investigate and treat primary ciliary cirrhosis. Clin Liver Dis 2008; 12: 445–60. 4 Montano-Loza AJ, Wasilenko S, Bintner J, Mason AL. Cyclosporine A inhibits in vitro replication of betaretrovirus associated with primary biliary cirrhosis. Liver Int 2010; 30: 871–77.
Figure: Alkaline phophatase (ALP) and alanine aminotransferase (ALT) levels showing improvement after anti-retroviral treatment started
96
www.thelancet.com Vol 377 January 1, 2011