LIST OF CONTRIBUTORS Brandie S. Adams
Department of Veterans Affairs Medical Center, St. Louis, MO, USA
Matthew E. Archibald
Department of Sociology, Emory University, Atlanta, GA, USA
Stephanie L. Ayers
Sociology Program, School of Social and Family Dynamics, Arizona State University, Arizona, USA
Cecilia Benoit
Department of Sociology, University of Victoria, Victoria, B.C., Canada
Lisa Cox Hall
Department of Sociology, University of Colorado at Denver and Health Sciences Center, Denver, CO, USA
Lucille C. Dauz
Department of Veterans Affairs Medical Center, St. Louis, MO, USA
Sangita Devarajan
Department of Surgery, Saint Louis University Medical Center, St. Louis, MO, USA
Jemima A. Frimpong
Health Care Systems Department, The Wharton School, University of Pennsylvania, Philadelphia, PA, USA
Jean Giles-Sims
Department of Sociology, Texas Christian University, Fort Worth, TX, USA
Greg A. Greenberg
Northeast Program Evaluation Center, Department of Veterans Affairs Medical Center, New Haven, CT, USA ix
x
LIST OF CONTRIBUTORS
Jennie Jacobs Kronenfeld
Sociology Program, School of Social Family Dynamics, Arizona State University, Tempe, AZ, USA
Frank E. Johnson
Department of Surgery, Saint Louis University Medical Center, St. Louis, MO, USA and Department of Veterans Affairs Medical Center, St. Louis, MO, USA
Barbara W. Kim
Department of Asian and Asian American Studies, California State University, Long Beach, CA, USA
Sam S. Kim
Sociology Program, School of Social and Family Dynamics, Arizona State University, Tempe, AZ, USA
Charles Lockhart
Department of Political Science, Texas Christian University, Fort Worth, TX, USA
Walter E. Longo
Division of Colon and Rectal Surgery, Yale University, New Haven, CT, USA
Lan H. Marietta
Department of Veterans Affairs Medical Center, St. Louis, MO, UK
Marc A. Musick
Department of Sociology, The University of Texas at Austin, Austin, TX, USA
Patrick A. Rivers
College of Applied Sciences & Arts, Southern Illinois University, Carbondale, IL, USA
Anne Rogers
National Primary Care Research and Development Centre (NPCRDC), University of Manchester, Manchester, UK
Leah Rohlfsen
Sociology Program, School of Social and Family Dynamics, Arizona State University, Tempe, AZ, USA
Caroline Sanders
National Primary Care Research and Development Centre (NPCRDC), The University of Manchester, Manchester, UK
List of Contributors
xi
Leah Shumka
Department of Anthropology, University of Toronto, Toronto, ON, Canada
Deborah Sullivan
Sociology Program, School of Social and Family Dynamics, Arizona State University, Tempe, AZ, USA
Mary P. Valentine
New York University, New York, USA
Katherine S. Virgo
Department of Surgery, Saint Louis University Medical Center, St. Louis, MO, USA and Department of Veterans Affairs Medical Center, St. Louis, MO, USA
Tarynn M. Witten
Department of Gerontology, Virginia Commonwealth University Randolph Minor Access, Richmond, VA, USA
Meredith G. F. Worthen
Department of Sociology, The University of Texas at Austin, Austin, TX, USA
Grace J. Yoo
Department of Asian American Studies, San Francisco State University, San Francisco, CA, USA
HEALTH INEQUALITIES AND HEALTH DISPARITIES: A SOCIOLOGICAL PERSPECTIVE$ Jennie Jacobs Kronenfeld This chapter provides an introduction to Volume 25 of the Research in the Sociology of Health Care series. This volume is entitled Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers. The overall volume is divided into five sections. Section 1 includes this chapter and another chapter that provides an introductory and overall approach to issues related to health inequalities and health disparities. Section 2 examines racial and ethnic inequalities and disparities. Section 3 includes chapters that focus on the topic of health inequalities and disparities from the perspective of research about health care providers and health care facilities. The last two sections of the book focus on consumers and topics of health care disparities. Section 4 focuses on issues $
This chapter provides an introduction to Volume 25, Research in the Sociology of Health Care, Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers. It introduces the topic of health inequalities and health disparities and discusses the approach to this issue in the United States based on federal government efforts as well as based on research by medical sociologists, political scientists and researchers in health care more generally, such as those in public health. This chapter also serves as an introduction to the volume. As such, the chapter explains the organization of the volume and briefly comments on each of the chapters included in the volume.
Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 3–14 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00001-4
3
4
JENNIE JACOBS KRONENFELD
related to substance abuse, mental health and related concerns. Section 5 focuses more generally on issues of disparities and consumers of health care, with chapters looking at issues of vulnerable women, breast cancer and colorectal cancer. This chapter first discusses issues of inequalities and disparities in health care and health and then provides an overview of each of the chapters in the book.
HEALTH CARE INEQUALITIES AND DISPARITIES While Americans have often believed that the United States has the best health care system in the world and that, as one of the wealthiest nations, we therefore must have the best health care available to our citizens, researchers in medical sociology, public health and health services research have emphasized for decades that America tolerates extremes of wealth and poverty much greater than in many European countries. This toleration of extremes extends to the approach to the delivery of social and health services, as well as to consumer goods. Over 40 million Americans do not have health insurance and thus have limited access to expensive health care services (Morone & Jacobs, 2005b). Even more may have very poor health coverage, so that if a serious illness were to occur, the person would have a very hard time finding care and paying for that care. Even if people have coverage for major health care problems, many people do not have insurance that covers areas of health care such as vision care, dental care and audiology services. While these are not life threatening health care concerns, they are health care concerns that impact quality of life and even ability to achieve. A child who cannot see well has trouble succeeding in school. A person in pain from tooth problems has trouble concentrating on tasks, and poor oral health is one contributor to nutrition concerns among the elderly. Lack of access to hearing aids increases the social isolation of the elderly, but these services are not covered by Medicare, the federal program that does provide access to health care services for most of the elderly in the United States. From a societal perspective, the issues of inequality in the overall economy lead to what some have described as an American Dilemma of why the United States is not number one in overall health statistics (Morone & Jacobs, 2005a; Kawachi, 2005; Jacobs, 2005). Even though the United States spends a greater share of its gross domestic product on health care than other nations, many people (estimates are 15.9 percent of the U.S. population) lack health insurance coverage (Luft, 2007). On many health
Health Inequalities and Health Disparities
5
measures, the United States is often no better and at times worse than many other nations (World Health Organization, 2000). Kawachi (2005) points out that on 16 different indicators of health status, the United States ranks 12th overall when compared with the 13 most economically advanced countries, behind such countries as Japan, Sweden, Canada, France, Australia, Spain, the United Kingdom and Denmark. This is true for percentage of low birth weight (13th), neonatal and infant mortality (13th) and years of potential life lost (13th). The United States does a little bit better on life expectancy measures, ranking 11th for life expectancy at age 1 for females but only 12th for males and 7th at life expectancy at age 65 for females and 7th for males. By age 80, the U.S. life expectancy is near the top (3rd for males and females). Worse yet, these mediocre comparative results are not the result of limited spending on health care services. The United States spends more than any other nation on health care, whether the figure examined is the dollars spent per capita or the percent of gross domestic product spent on health care. Many health care experts agree that part of the reason for disparities in health and health care use in the United States are large and exceptional levels of economic inequality and poverty (Kawachi, 2005; Jacobs, 2005). Race, income and market ideology are all sources of health care inequalities (Stone, 2005). Stone points out that there are disparities by characteristics such as gender, state residence, immigrant status and type of illness. She also points out that these different characteristics of people, their location and diseases that disadvantage people in receipt of health care services become cross-cutting divisions that weaken political support for reforms to make the system more just (that is, exhibiting less inequalities across groups and social characteristics). Race and income are not the only fault lines that divide people. She does argue, however, that market ideology is the most important obstacle to health care equity because, under market theory, distribution will follow economic demand rather than need (Stone, 2005). Gottschalk (2005) argues that one of the major reasons in the United States why health care inequalities have not resulted in a social movement or reform coalition is the weak role of organized labor, especially in recent years. While compared to European countries, labor unions have historically been weak in the United States and there has not been a major political party specifically identified with the labor movement, Gottschalk (2005) and others have pointed out that labor has been important in health care reforms in the United States (Rosner & Markowitz, 1997). The development of early prepaid group practices and some important pushes
6
JENNIE JACOBS KRONENFELD
for reform in the U.S. health care system such as Medicare were helped greatly by the labor movement in the United States. The current system of employment-based health benefits in the United States today resulted from collective bargaining agreements during World War II and in the 1950s. While historically organized labor in the United States was an advocate of national health insurance, beginning in 1978, organized labor was willing to push private sector solutions that used government mandates and employer mandates as the way to ensure health insurance coverage for more people. Gottschalk argues that divisions within the labor movement and the distractions of the NAFTA (North American Free Trade Agreement) debates made it difficult for the labor movement to be aggressive in its support of the proposed Clinton health care reforms in the early 1990s. This allowed the opponents of the proposed Clinton legislation to define and dominate the public debate. Legal issues also impact health inequalities and health disparities. In other areas of social policy, the right to receive services is much clearer. Welfare is a government entitlement program and education is a legal requirement, but health care has traditionally been provided by the private sector with voluntary contractual obligations and in recent decades the courts have been reluctant to confront health care inequalities (Jacobson & Selvin, 2005). There are some legal provisions to ensure care and thus limit inequities in health care in the U.S. Medicare, a federal program, provides health care coverage for almost all Americans who are 65 and older. Medicaid, a joint federal-state program, provides health care coverage for many but not all poor people who meet eligibility requirements. These requirements are mostly defined by states. SCHIP (State Child Health Insurance Program) is another joint federal-state program that provides coverage for many children of the working poor. In recent decades, the courts tend to defer to the political branch for most issues of health care policy. There have been some expansions of the federal government role in helping to reduce inequities in access to health care services (Grogan & Patashnik, 2005; Kronenfeld, 2006). The two major ones have been expansions for children or pregnant women and expansions for the elderly. Incremental expansions to cover some children and pregnant women were enacted between 1984 and 1990 and SCHIP expanded the coverage for children of the working poor, although in operation there is important variability in how the program operates from state to state. Medicaid has become the most important payer of long-term care services, and is providing nursing home care for many people who were not ‘‘poor’’ for most of their working lives, but are unable to meet the expenses of nursing home coverage.
Health Inequalities and Health Disparities
7
Not all issues of health care inequality occur within the United States, although that has been the focus of most of this review. Research has documented differences in other countries as well, and some of this research has been sociological, following long-standing interests in medical sociology in examination of race, ethnicity and social class as factors that impact variation in health status and receipt of health care services (Smaje, 2000). A recent Canadian study that examined direct cardiac care found that socioeconomic status was at the heart of health care inequalities in Canada, even within a system that assures access to basic care for all (Basky, 2000). In Ontario, Canada, patients living in neighborhoods with the highest average incomes received coronary angiography 23 percent more often and had 45 percent shorter waiting times for treatment than did patients living in the lowest-income neighborhoods.
THE FEDERAL EFFORT TO ADDRESS HEALTH AND HEALTH CARE DISPARITIES In the past decade, a great deal of attention at both the federal government level and in some of the large private foundations that are important funders of research in health care has been given to the issue of health care disparities and inequities (Kaiser Family Foundation, 2003; National Healthcare Disparities Report, 2004; Smedley, Stith, & Nelson, 2003). The National Healthcare Disparities Report (2004) shows that individuals from lower socioeconomic backgrounds and racial and ethnic minorities with varying backgrounds are more likely to report unmet health care needs and less likely to have a consistent source of health care, receive routine care and benefit from insurance coverage. A focus on some of these issues is much older. For example, about 20 years ago the federal government began to release reports that examined issues of black and minority health. These early efforts within the government as well as within sociological, social science and public health research led to legislation that mandates a greater interest in issues of health care disparities and inequalities. Within the federal government, one of the pushes for more research on health care inequalities came from the passage of Public Law 106-129, the Healthcare Research and Quality Act of 1999. This law directed the Agency for Healthcare Research and Quality (AHRQ) to develop two annual reports, one focused on quality and one focused on disparities. AHRQ was directed to track prevailing disparities in health care delivery as they relate
8
JENNIE JACOBS KRONENFELD
to racial and socioeconomic factors among priority populations. Priority populations include low-income groups, racial and ethnic minorities, women, children, the elderly, individuals with special health care needs, the disabled, people in need of long-term care, people requiring end-of-life care and places of residence (rural communities). The first National Healthcare Disparities Report (2004) was built on some previous efforts in the federal government, especially Healthy People 2010 (U.S. Department of Health & Human Services, 2000) and the IOM 2002 Report, Unequal Treatment: Confronting Racial and Economic Disparities in Healthcare (Smedley et al., 2003). The elimination of disparities in health was one of the goals of Healthy People, 2010. Unequal Treatment extensively documented health care disparities in the United States, with a focus on those related to race and ethnicity. One of the weaknesses of this report was that there was not any focus on disparities related to socioeconomic status. The National Healthcare Disparities Report (2004) did have a focus on the ability of Americans to access health care and variation in the quality of care. Disparities related to socioeconomic status were included, as were disparities linked to race and ethnicity and the report also tried to explore the relationship between race/ethnicity and socioeconomic position. There were seven key findings from the report. Firstly, inequality in quality of care continues to exist and often these are particularly true for some more serious health care problems, such as minorities being diagnosed with cancer at later stages, less often receiving optimal care when hospitalized for cardiac problems and higher rates of avoidable hospital admissions among blacks and poorer patients. Secondly, disparities come at a personal and societal price. Thirdly, differential access to health care often leads to disparities in quality of care actually received. As a fourth issue, opportunities to provide preventive care are often missed. The last three points all relate to the need for more data, more research and the linkage of those to policy within the United States. The knowledge about why disparities continue to exist is still limited, and data limitations may limit improvement efforts. Despite these concerns, improvement is possible and some examples are provided using California subpopulation data that demonstrate how targeted some prevention efforts to specific groups can yield useful results. In 2005, the third National Healthcare Disparities Report (2005) was released. One advantage of continuing reports is that they not only provide an overview of the current situation, but also allow a comparison to previous years. This 2005 report focuses on findings from a set of core report measures. The two measures of access covered in this report are facilitators
Health Inequalities and Health Disparities
9
and barriers to care and health care utilization. The overall summary indicates that disparities still exist, but some disparities are diminishing, an encouraging result, but one that clearly leaves opportunities for further improvement. Disparities remain in both areas of access, all areas of quality, and across many levels and types of care including preventive care, treatment of acute conditions and management of chronic disease. This applies to a variety of specific clinical conditions including cancer, diabetes, end stage renal disease, heart disease, HIV disease, mental health and substance abuse and respiratory diseases. Looking at access more specifically, major issues of disparity occur for poor people and Hispanics, with lesser but important issues for Blacks, American Indians and Asians. Poor people have worse access to care than high-income people for all eight core report measures. Hispanics have worse access for 88 percent of the core report measures, while for Blacks and American Indians, they have worse access on half of the measures. For Asian Americans, they have worse access on 43 percent of the measures. The 2005 report also tracks changes in the core measures over time. For each core report measure, racial, ethnic and socioeconomic groups are compared with a designated comparison group at various points in time. For racial minorities, more disparities in quality of care are becoming smaller rather than larger, while for Hispanics, 59 percent were becoming larger and 41 percent smaller. For poor people, half of disparities were becoming smaller and half were becoming larger. All groups faced some disparities in quality of care, and some disparities in quality were important for multiple groups, such as new AIDS cases, problems with timeliness of care and problems with patient-provider communications. Reducing disparities is an important overall goal. There are many ways to work towards these goals, some of which include greater understanding of disparities, as may come from the other chapters in this volume. The reaction to the federal reports is not always simple, and the interpretation of the information from the reports can itself cause controversy. For example, there was a controversy about the some rewrites of the 2003 report that downplayed whether racial differences in care resulted in adverse health outcomes (Bloche, 2004). Some newer approaches have focused on looking at disparities from the angle of how law can be used to push policy changes. Rosenbaum and Teitelbaum (2005) view law as a powerful instrument in American life and believe that classic civil rights law can be expanded to address discrimination in health and health care. One way they suggest approaching inequality is through the lens of unequal quality of care rather than only a focus on unequal access. There are many innovative ways
10
JENNIE JACOBS KRONENFELD
to examine inequalities and disparities and the chapters in this book begin to provide some new data and ways to examine issues also. Several recent chapters in the Journal of the American Medical Association discuss issues of universal health care coverage and link these to health disparities and access to care (Luft, 2007; Lurie & Dubowitz, 2007; Fontanarosa, Rennie, & DeAngelis, 2007). Lurie and Dubowitz (2007) argue that social factors related to disparities in care relate directly to access to care, which then leads to better health. They also argue that universal access to care might be called universal access to health, given this link. Luft (2007) calls for fundamental restructuring of the payment system within health care to achieve both universal coverage and improved efficiency. Lurie and Dubowitz (2007) point out that one key contributor to disparities in health is differential access to care, and this is linked to differences in rates of uninsurance. Hispanics and Blacks have higher rates of uninsurance (34 percent and 21 percent), as compared to rates on 13 percent among whites (Lurie & Dubowitz, 2007). Given the early start of campaigning for the Democratic and Republican nominations for the Presidency in 2008, there are already beginning discussions about plans to reform the health care insurance system in the United States. As these recent chapters point out, such reform (if it occurs) may play an important role in reduction of health care disparities and inequalities in the future in the United States.
REVIEW OF ORGANIZATION OF THE BOOK As mentioned previously, this volume is divided into five sections. The first section, Health Inequalities and Health Disparities: Overall Perspectives, includes this chapter and another chapter by Sanders and Rogers. The Sanders and Roger chapter, entitled ‘‘Theorising Inequalities in the Experience and Management of Chronic Illness: Bringing Social Networks and Social Capital Back in’’ applies the subject of social networks to chronic illness and disability and helps to explore the issue of social networks as a way to think about health inequalities and health disparities. The second section includes two chapters that examine racial and ethnic inequalities and disparities. The first chapter, by Greenberg, looks at U.S. minorities’ access to health care under managed care. Organizational theory is used to clarify and synthesize the large and diverse literature on the relationship between managed care (MC) and ethnic differences in access to health services. MC practices are classified by whether they are used by
Health Inequalities and Health Disparities
11
health care organizations to define their boundaries or to coordinate care. The second chapter by Yoo and Kim focuses on health care issues of one specific minority group, Korean Americans. Korean Americans have one of the highest rates of not having health care insurance in the United States. Through in-depth interviews and surveys, this study found that that high premiums prevented the uninsured from purchasing health insurance and high deductibles prevented insured persons from utilizing health services. The third section includes chapters that focus on the topic of health inequalities and disparities from the perspective of research about health care providers and health care facilities. This section includes three chapters each looking at issues linked to physicians, long-term care facilities and nurse anesthetists. Musick and Worthen examine issues linked to trust in physicians. Using data from the 1998 General Social Survey, the chapter shows that social resources, vulnerability in finances and in perceptions about the end of life and exposure to unstable environments all are fairly consistent predictors of physician trust. Giles-Sims and Lockhart review disparities in Medicaid support for and quality of nursing facility long-term care across states. The chapter uses regression models that provisionally explain the sources of these inequalities and comments on the social implications of these disparities. Sullivan and Rohlfsen examine issues about certified registered nurse anesthetists (CRNAs) through a comparison of data from two states, Arizona and Iowa. CRNAs are the sole anesthesia providers in about two-thirds of the nation’s rural hospitals. This study focuses on the impact of the turf battles between CRNAs and anesthesiology over the requirement for CRNA supervision. The chapter links this issue to policy issues of Medicare requirements for CRNA supervision and its relationship to availability of care in rural versus urban areas. The fourth section of the book focuses on consumers of health care and issues of mental health, substance abuse and related concerns. This section includes three chapters. The first chapter by Archibald looks at socioeconomic and racial/ethnic disparities in substance abuse treatment provision, treatment needs and utilization. The chapter begins by noting that the National Health Disparities Report documents parity in substance abuse treatment provision among individuals of varying socioeconomic and racial/ethnic backgrounds. This study investigates that finding by analyzing the relationship between community socioeconomic and racial/ethnic disadvantage and organizational provision of substance abuse treatment, treatment need and utilization across United States counties in 2000, 2002 and 2003. Results confirm equity in service provision in poorer communities and those with higher concentrations of African Americans. The study also
12
JENNIE JACOBS KRONENFELD
demonstrates that significant disparities remain in communities with higher concentrations of Hispanics, youth and female-headed households. The second chapter in this section is by Ayers and her coauthors and explores geographical disparities in variation in utilization of and need for mental health services. People in 13 different states are included in the study. Although individual sociodemographic characteristics are important in examining mental health utilization, some state characteristics (especially the percentage of African Americans in the state) are also important predictors of variations in mental health utilization and need for services. The final chapter in this section is Witten’s chapter ‘‘Transgender Bodies, Identities & Healthcare: Effects of Perceived and Actual Violence and Abuse’’. This chapter argues that some groups, due to elevated stigma associated with membership in that group, are invisible as a disparate minority. She focuses on issues of health care for the transgender-identified population and shows how the normative viewpoint of mental illness and unacceptable religious status, along with lifelong perceived and actual abuse and violence, creates a socially sanctioned inequality in health care for this population. The last section of the book focused on consumers and topics of health care disparities more generally, with chapters with more specific foci such as issues of vulnerable women, breast cancer and colorectal cancer. This section includes three chapters. The first chapter by Shumka and Benoit looks at issues of some more specialized health care for vulnerable women workers (those working in socially and economically marginalized ‘‘frontline’’ service occupations) in Canada. Participants identified a number of health concerns that they link to the everyday suffering they endure – i.e., feeling inadequate, incompetent, lonely, self-conscious, disenfranchised or dissatisfied. Within the context of an overall national health care system, these women often articulated a desire for alternative/complementary care, something not funded by the Canadian health care system. This care would require out-of-pocket costs that the women often cannot afford. The second chapter in this section deals with issues of age, breast cancer and disparities in health knowledge and treatment for women. Hall argues that older women have differential experiences that, in part, stem from our youth-oriented culture. Older women are less empowered than their younger counterparts to display the same degree of agency and this limits aspects of their care. The last chapter by Virgo and her colleagues looks at the influence of dual use (people who are eligible for multiple sources of government-reimbursed care such as the Department of Veterans Affairs (VA) and Medicare) on
Health Inequalities and Health Disparities
13
survivorship patterns of a nationwide cohort of elderly veterans with colorectal cancer. Based on past research, it is not clear whether combined eligibility translates into increased access to care and/or improved outcomes of care. While there are reasons to believe this might be true there are also reasons why continuity of care may suffer leading to worse outcomes when patients receive health services from multiple unrelated sources of care. Using retrospective analyses of 13 years of nationwide Medicare and VA inpatient and institutional outpatient data, Virgo and her colleagues find an important benefit of dual eligibility for government-reimbursed health care services that have significantly increased likelihood of survival. In addition, there may be a higher curative treatment rate per detected recurrence for people with dual eligibility. Overall, this volume presents a varied array of chapters that deal with differing aspects of sociology of health care and presents a sociological perspective on health inequalities and health disparities. Chapters look at overall perspectives, racial and ethnic variations, variations related to providers of care and specialized aspects of health disparities. This book adds to the knowledge in the medical sociological literature about health care inequalities and health disparities. Whether a reader chooses to read each chapter in order, or pick and choose specific topics of the greatest interest, this volume adds to the ongoing discussion in many health disciplines, including medical sociology, about health inequalities, health disparities and how they can be reduced in the future in the United States and in other countries.
REFERENCES Basky, G. (2000). Socioeconomic status at the heart of health care inequality. Canadian Medical Association Journal, 162(2), 253–254. Bloche, M. G. (2004). Health care disparities – science, politics, and race. The New England Journal of Medicine, 350, 1568–1570. Fontanarosa, P. B., Rennie, D., & DeAngelis, C. D. (2007). Access to care as a component of health system reform. Journal of the American Medical Association, 297(10), 1128–1130. Gottschalk, M. (2005). Organized labor’s incredible shrinking social vision. In: J. A. Morone & L. R. Jacobs (Eds), Healthy, wealthy and fair: Health care and the good society (pp. 137–175). New York: Oxford University Press. Grogan, C., & Patashnik, E. (2005). Medicaid at the crossroads. In: J. A. Morone & L. R. Jacobs (Eds), Healthy, wealthy and fair: Health care and the good society (pp. 267–295). New York: Oxford University Press.
14
JENNIE JACOBS KRONENFELD
Jacobs, L. R. (2005). Health disparities in the land of equality. In: J. A. Morone & L. R. Jacobs (Eds), Healthy, wealthy and fair: Health care and the good society (pp. 37–62). New York: Oxford University Press. Jacobson, P. D., & Selvin, E. (2005). Courts, inequality and health care. In: J. A. Morone & L. R. Jacobs (Eds), Healthy, wealthy and fair: Health care and the good society (pp. 235–266). New York: Oxford University Press. Kaiser Family Foundation. (2003). Key facts: Race ethnicity and medical care. Washington, DC: Kaiser Family Foundation. Kawachi, I. (2005). Why the United States is not number one in health. In: J. A. Morone & L. R. Jacobs (Eds), Healthy, wealthy and fair: Health care and the good society (pp. 19–36). New York: Oxford University Press. Kronenfeld, J. J. (2006). Expansion of publicly funded health insurance in the United States: The children’s health insurance program and its implications. Lanham, MD: Lexington Books. Luft, H. S. (2007). Universal health care coverage: A potential hybrid solution. Journal of the American Medical Association, 297(10), 1115–1118. Lurie, N., & Dubowitz, T. (2007). Health disparities and access to health. Journal of the American Medical Association, 297(10), 1118–1121. Morone, J. A., & Jacobs, L. R. (Eds). (2005a). Healthy, wealthy and fair: Health care and the good society. New York: Oxford University Press. Morone, J. A., & Jacobs, L. R. (2005b). Introduction: Health and wealth in the good society. In: J. A. Morone & L. R. Jacobs (Eds), Healthy, wealthy and fair: Health care and the good society (pp. 3–19). New York: Oxford University Press. National Healthcare Disparities Report. (2005). National Healthcare Disparities Report. AHRQ Publication no. 06-0017. December, Agency for Healthcare Research and Quality, Rockville, MD, www.ahrq.gov/qual/nhdr05/nhdr05.pdf National Healthcare Disparities Report: Summary. (2004). National Healthcare Disparities Report: Summary. Agency for Healthcare Research and Quality, Rockville, MD, February. http://www.ahrq.gov/qual/nhdr03/nhdrsum03.htm Rosenbaum, S., & Teitelbaum, J. (2005). Addressing racial inequality in health care. In: D. Mechanic, L. B. Rogut & D. C. Colby (Eds), Policy challenges in modern health care (pp. 135–150). Piscataway, NJ: Rutgers University Press. Rosner, D., & Markowitz, G. (1997). Hospitals, insurance and the American labor movement: The case of New York in the postwar decades. Journal of Policy History, 9, 74–95. Smaje, C. (2000). Race, ethnicity, and health. In: C. E. Bird, P. Conrad & A. M. Fremont (Eds), Handbook of medical sociology (5th ed.). Upper Saddle River, NJ: Prentice Hall. Smedley, B. D., Stith, A. Y., & Nelson, A. R. (Eds). (2003). Unequal treatment: Confronting racial and ethnic disparities in health care (Institute of Medicine). Washington, DC: National Academies Press. Stone, D. (2005). How market ideology guarantees racial inequality. In: J. A. Morone & L. R. Jacobs (Eds), Healthy, wealthy and fair: Health care and the good society (pp. 65–89). New York: Oxford University Press. U.S. Department of Health and Human Services. (2000). Healthy people 2010: With understanding and improving health and objectives for improving health (2nd ed., Vols). Washington, DC: U.S. Government Printing Office, November, 1(11). World Health Organization. (2000). World Health Report 2000. Geneva, Switzerland: World Health Organization.
THEORISING INEQUALITIES IN THE EXPERIENCE AND MANAGEMENT OF CHRONIC ILLNESS: BRINGING SOCIAL NETWORKS AND SOCIAL CAPITAL BACK IN (CRITICALLY) Caroline Sanders and Anne Rogers ABSTRACT Social networks have been a central focus of sociological research on inequalities but less has focused specifically on chronic illness and disability despite a policy emphasis on resources necessary to support selfmanagement. In this chapter, we seek to unpack overlaps and distinctions between social network approaches and research on the experience and management of chronic illness. We outline four main areas viewed as central in articulating the potential for future work consistent with a critical realist perspective: (1) body–society connections and realist/ relativist tensions; (2) the controversy of ‘variables’ and accounting for social and cultural context in studying networks for chronic illness support; (3) conceptualising social support, network ties and the Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 15–42 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00002-6
15
16
CAROLINE SANDERS AND ANNE ROGERS
significance of organizations and technology; and (4) translating theory into method.
INTRODUCTION Social networks have figured prominently in sociological research on inequalities within various aspects of social life including education (Bagnall, Longhurst, & Savage, 2003), crime (Kennedy, Kawachi, Prothrow-Smith, Lochner, & Gupta, 1998) and family relationships (Bott, 1957; Edwards, 2004). A particular focus has been on the role of social networks in constructing a valuable commodity for individuals and groups that has been conceptualised as social capital. The role of social networks and the associated production of social capital in the inequitable genesis of illness and mortality has also been a key focus of sociologists and epidemiologists interested in public health and the social causes of ill health (Davey Smith & Lynch, 2004). Within Western democracies, policy changes have increasingly placed an emphasis on community care and the maximisation of self-management capacity for chronic conditions promoting an interest in available support and social networks (Pescosolido & Kronenfeld, 1995; Walker, 1987; Rogers, Hassell, & Nicolaas, 1999). However, relatively little conceptual and empirical work on social networks and social capital has focussed specifically on chronic illness and disability with some notable exceptions emerging mainly in the United States (e.g. Pescosolido, 2001; Levy & Pescosolido, 2002). As Pescosolido (2001) states, ‘‘Too often we have neglected to consider that what makes people’s experience in the community and treatment systems ‘success’ or ‘failure’ are intimately tied to the kind of relationships forged and maintained in those contexts’’ (p. 468). There has been a tendency for those devising and evaluating selfmanagement interventions inside formal health and social care settings, to bracket off the role played by social networks and resources in favor of a focus on discrete programs of guided self-management skills training (Kendall & Rogers, 2007). Social networks and support may not be seen as being the mainstream business of policy makers, health care providers and professionals; however, understanding the role these play has implications for how people with long-term conditions are supported inside and outside of formal health service provision. An over-emphasis on the provision of formalised self-care programs which prioritise psychological change, may
Theorising Inequalities in the Experience and Management of Chronic Illness
17
mask or detract from an awareness of the role of support (or lack of it) from existing networks in adapting to chronic illness. Similarly a focus on the selfmanagement of symptoms (directed from within health services) may detract from the importance of maintaining ‘‘valued social roles, coherent identities and a ‘normal life’’’ (Townsend, Wyke, & Hunt, 2006, p. 185). The potential value of developing a richer analytical and empirical understanding in this area is implicated in the large body of work highlighting the relevance of social support to chronic illness adaptation and management. Much of this springs from qualitative research on the experience of illness within family and other collective contexts (Anderson & Bury, 1988; Gregory, 2005; Williams, 1989; Rogers & Pilgrim, 1991). A number of quantitative studies have also attempted to measure the impact of social support on recovery from chronic illness episodes (e.g. see review by Kaplan & Toshima, 1990). For some individuals and groups, aspects of social support are not easily separated out from individual selfmanagement. Consequently, any evaluation of formalised self-management interventions requires an awareness of the complexities of deriving (or failing to derive) tangible support and ontological security from network relations (e.g. Gregory, 2005). An in-depth understanding of social networks and social capital has relevance for understanding the way health service resources can complement, substitute or enhance patients’ existing resources in managing a long-term condition. Appreciation and understanding of these aspects are relevant for ensuring appropriate professional and service input into self-care support. Turning to inequalities, the structure and content of social networks may impact on who gains access to, and engages in, self-management programs. There is evidence that such programs may be operating an ‘inverse care’ law (Hart, 1971), with recruitment skewed in favor of those who are more affluent and see themselves as good self managers; whilst those who could benefit the most are least likely to engage with such programs (Kennedy, Rogers, & Gately, 2005). Additionally, some network ties are known to advantage those who are already ‘rich’ in terms of social capital, whereas other ties (or lack of them) serve to disadvantage those who are already lacking social capital. This makes the subject particularly salient to the inequalities in health agenda and is important for understanding the failure to engage specific groups who might benefit most from health interventions. An array of concepts has been associated with social networks including: social support, neighbourhood attachment, civic engagement and community trust (Veiel & Baumann, 1992). Some have commented on the confusion and conflation of meaning within this group of concepts which
18
CAROLINE SANDERS AND ANNE ROGERS
have sometimes been used inter-changeably (Berkman, Glass, Brissette, & Seeman, 2000). This has prompted theoretical and empirical work as a means of promoting clarity and distinction (Li, Pickles, & Savage, 2005; Szreter & Woolcock, 2004). All of these concepts have been drawn upon in studying the impact of social relationships on individuals and groups for various aspects of social life. Within the health field, a key focus has been the relationship between societal structure (conceived as having ‘upstream’ influence) and the micro-level influence of networks involving interpersonal behaviour and psychobiological processes (having ‘downstream’ influence) as dynamically linked to health status (Berkman et al., 2000). Researchers have often pointed to the work of early sociologists such as that of Durkheim (1951) on suicide, and Simmel (1955) on the nature of group affiliations, to illustrate these links at the root of a social networks perspective (Berkman et al., 2000; Pescosolido & Levy, 2002; Pescosolido & Rubin, 2000). Social networks have mainly been viewed to convey social advantage to those positioned within them, and the distinctions between types of network and social engagement have been drawn upon to account for differentials in the levels of capital attained. Researchers have also studied the negative impact of specific social relationships (e.g. Rook, 1992), and there has been an emphasis on the general erosion of social capital in deprived areas caused by factors such as poverty, suburbanisation, and changes in family and economic structure (Putnam, Leonardi, & Nanetti, 1993). In this chapter we outline some key themes arising from recent theoretical debates and empirical research on social networks and social capital which appear to readily connect with previous research on the experience and management of chronic illness. Although overlaps are apparent, there are also distinctions and ongoing debates that are particularly evident when exploring the dominant theoretical perspectives. Whilst social network research has mainly emerged within a positivist and ‘realist’ framework, much of the sociological research on illness experience and management can be defined as post-positivist, and declared perspectives vary including versions of realism or relativism. In addition, the focus on ‘networks’ as a means of studying social relations has evolved in a distinct way within Science and Technology Studies (STS) via Actor Network Theory (ANT) that adopts a strong relativist stance. However, there has been very little dialogue between this approach and the mainstream approach of Social Network Analysis, save some acknowledgement of common anthropological roots (Knox, Savage, & Harvey, 2006). We seek to unpack some of the overlaps and distinctions between relevant bodies of work
Theorising Inequalities in the Experience and Management of Chronic Illness
19
as a means of articulating the potential for future research. We outline four main areas comprising: (1) body–society connections and realist/relativist tensions; (2) the controversy of ‘variables’ and accounting for social and cultural context in studying networks for chronic illness support; (3) conceptualising social support, network ties and the significance of organisations and technology; and (4) translating theory into method. We argue that this might serve as a useful framework for examining the impact of contemporary social change on inequalities in the experience and management of illness, enabling critical insights on policy concerns to promote self-management and equitable health care for those with chronic conditions. We present the view that a social network approach consistent with a critical realist perspective is beneficial for connecting research on the narrative experience of illness, with the political and economic issues of primary importance for those with chronic conditions or disabilities. Bridging this divide is important to facilitate a specific focus on inequalities and associated implications for meeting current and future health care needs.
BODY–SOCIETY CONNECTIONS AND REALIST/RELATIVIST TENSIONS Social network research applied to health and inequalities has highlighted the connections between society and the body in the aetiology of disease within a social causation model. Here, social relationships as mediated by structural variables are viewed as essential; for example, in the transmission of infectious diseases and the genesis of mental illness such as depression. Where some have focussed on the indirect connections between the body and society via health-related behaviour, others have focussed more directly on the psychobiological processes associated with aspects of social relations. In the latter case, research has drawn attention to the impact of emotional stress on the body associated with specific life events such as bereavement and social roles according to factors such as gender and employment status (Parkes, Benjamin, & Fitzgerald, 1969; Haney, 1980; Gove, 1984; Marmot, Rose, Shipley, & Hamilton, 1978; Brown & Harris, 1989). More recently, notions of the embodiment of social capital (Freund, 2006) and emotional capital (Williams, 1998) have been drawn upon to examine the way structural and interactional factors impact on the body, increasing vulnerability to illness in socially patterned ways.
20
CAROLINE SANDERS AND ANNE ROGERS
Social network research has been adept at exploring the impact of social exclusion on the body in creating illness inequalities. The flip side of this is the impact of the body on social exclusion which has important implications for inequalities in the experience and management of existing chronic conditions. As noted by Pescosolido (2001), the ability to form and maintain social ties may be the result of health, illness and disability problems, not simply factors implicated in their cause (p. 484). The relationship between the body and social structure has also been a focus of discussion and debate for examining the way impairment is associated with social roles, societal participation and exclusion (Williams, 1996, 1999a, b; Williams & Busby, 2000). Moreover, the interactional dynamics connecting impairment to perceptions of ‘self ’ and identity (Kelly & Field, 1996) have been elicited, particularly where impairment is associated with stigma (Reidpath, Chan, Gifford, & Allotey, 2005; Coleman, 1997). This work has emphasised the cultural significance and day-to-day consequences of illness as a fundamentally embodied experience within a societal context (Williams, 1996). It seems that an explicit focus on relations as ‘networks’ around illness experience and management could elaborate and extend such work. As Berkman et al. (2000) state, ‘‘those roles that provide each individual with a coherent and consistent sense of identity are only possible because of the network context which provides the theatre in which role performance takes place’’(p. 849). The network context of role performance has been well illustrated within social network research in examining the impact of stigma (an issue of considerable importance in relation to chronic illness and disability) on opportunities for accessing social capital (Carter & Feld, 2004; Warr, 2005). Some disability theorists have been critical of the focus of medical sociologists on the body and impairment arguing that it conforms to an individualistic model of disability (Oliver, 1996); a view that has been contested (Bury, 1997). In contrast, the work of disability theorists is presented as a ‘social’ model with many adopting a neo-marxist perspective to emphasise the political and economic factors that define disability experience (Barnes, 1998). Such debates have prompted greater exploration of the divide between illness and disability (Barnes & Mercer, 1996; Williams, 1996; Zola, 1991) with some advocating the need for a ‘critical realist’ approach that recognises both the corporeal aspects of experience and the constraints of social structure (Williams, 1999b; Pilgrim & Rogers, 1994, Rogers & Pilgrim, 2005). Within social network research, there have also been parallel debates, and whilst most mainstream social network research is regarded as both structural and realist (Knox et al., 2006), some
Theorising Inequalities in the Experience and Management of Chronic Illness
21
argue that the ‘centre of gravity’ has tended toward the ‘micro’ end (Szreter & Woolcock, 2004). Szreter and Woolcock review perspectives adopted in previous research on social capital and health to highlight the importance of an historical materialist approach accounting for the political aspects of network structure that link state and society relations to health resources. The tensions between relativist (social constructivism) and realist (social/ critical realism) theoretical perspectives are particularly relevant to researching inequalities in illness experience because of the differing views adopted regarding power and agency. This is important for conceptualising relationships between bodies and society. Here, we briefly discuss some key aspects of ongoing debates regarding these perspectives and the particular value of a critical realist approach to researching inequalities in the experience and management of chronic conditions. This forms a precursor for the remaining sections with particular implications for translating theory into method as a proposed means of furthering this area of research. As previous authors have noted, there are multiple versions of social constructionism in both weak and strong form, subsuming various elements which are sometimes criticised for being contradictory (Bury, 1986; Nettleton, 1995; Timmermans & Berg, 2003; Rogers & Pilgrim, 2005). Whilst some strands of constructionism accept an external reality, others do not. For example, ‘radical constructionism’ presents biology as a social invention rather than accepting it as a universal reality (e.g. Armstrong, 1983). However, most sociologists accept some degree of external reality. As Lock argues, ‘‘There is, of course a biological reality, but the moment that efforts are made to explain, order, and manipulate that reality, then a process of contextualisation takes place in which the dynamic relationship of biology with cultural values and the social order has to be considered’’ (Lock, 1988 cited in Nettleton, 1995, p. 29). Some have argued for a middle ground, and as Rogers and Pilgrim (2005) state ‘‘social constructivism y does not necessarily have to be set in opposition to social realism (the view that there is an independent existing reality) or social causationism (the view that social forces cause measurable phenomena to really exist)’’ (p. 16). Turner also claims that it is possible to adopt a foundationalist ontology with a constructionist epistemology, advocating a pragmatic stance contingent upon the questions we want to address regarding illness and disability (Turner, 1992, 2001). Turner highlights a rationalistic bias in sociology which has until recently treated the social actor as a disembodied rational agent and points to Bordieu’s concepts of habitus and practice as valuable in drawing attention to the
22
CAROLINE SANDERS AND ANNE ROGERS
symbolic power of the body. These concepts have also been drawn upon as a valuable means of studying social relations and the generation of ‘cultural’ capital within specific contexts.
THE CONTROVERSY OF ‘VARIABLES’ AND ACCOUNTING FOR SOCIAL AND CULTURAL CONTEXT IN STUDYING NETWORKS FOR CHRONIC ILLNESS SUPPORT The relationships between social networks and social capital generation associated with specific variables, such as gender, age ethnicity and socioeconomic status (SES), have been a focus of previous social network research, although there has been a decline in the emphasis placed on such variables alongside contemporary cultural changes (Pescosolido & Rubin, 2000). Pescosolido (1996) states that ‘‘using socio-demographics as a proxy rather than struggling with a direct conceptutalisation of what social structure is and how it operates to affect individuals’ lives has left us at a loss in attempting to forsee what might happen to individuals under massive social change’’ (p. 173). Scambler, Higgs, and Jones (2002) also point out that the continuing medical sociological commitment to the social causation model and (neo-) positivist methodologies centred on ‘variable analysis’ has been responsible for a lack of sociological imagination in much inequalities research. Below, we outline some previous research that connects specific ‘variables’ with social inequalities as relevant to chronic illness experience and management. We also discuss some new and salient issues for contemporary research. Previous research has focussed on the connections between social roles and relationships associated with gendered inequalities in health (Brown & Harris, 1978; Gove, 1984). Gender has been particularly salient in addressing the management of chronic conditions in relation to caring responsibilities and practices which predominantly fall to women (Finch & Mason, 1993; Graham, 1983; Gregory, 2005; Thomas, 1995). Additionally, disabled women have highlighted the structural factors and cultural assumptions that impact upon social participation and can compound social exclusion amongst this group (French, 1993; Thomas, 1997). Social network research has specified differences in the nature of network ties between men and women, finding that women tend to form more extensive networks; whereas, men tend to have limited networks depending more on
Theorising Inequalities in the Experience and Management of Chronic Illness
23
emotional support from partners. This established view has also been challenged in previous research on social networks and social support among disabled people (Morgan, Patrick, & Charlton, 1984). How social networks change over the life-course in association with gender differences (Akiyama & Antonucci, 2002; Antonucci & Akliyama, 1987; Kahn, 2006) is pertinent to exploring the management of chronic conditions that are more common amongst older people (Adamson, Lawlor, & Ebrahim, 2004; Shippy & Karpiak, 2005). Consideration of contingencies specific to other stages of the life-course and conditions are also important as exemplified in the work of Roche and Tucker (2003) who focus on day-to-day lives of two groups of young people: young carers and young people with ME. They outline common factors that serve to socially isolate and exclude young people who are reliant on, or drawn into supporting, home-based caring relationships. Roche and Tucker argue that the current social exclusion debate’s primary focus on the public sphere (e.g. homelessness and social exclusion) neglects the ways in which young people can experience similar forms of disadvantage in the private sphere. They suggest the need for a wider critical perspective to clarify the interconnectedness of young people’s exclusion in the private and public sphere. This analysis has wider resonance for the experiences of other groups with chronic conditions, where social exclusion from the public sphere can be rendered invisible by concentration on the private domain of illness experience. Discussions by sociologists of a move to a ‘late’ modern or ‘postmodern’ society, highlight the relationship between contemporary cultural changes and consumption. For example, Warde and Tampubolon (2002) draw on secondary analysis of the British Household Panel Survey to investigate leisure consumption in relation to associational involvement and friendship network ties. They report a need for greater appreciation of the complexity and diversity of network ties (including the contextual effects of differences in friendship related to socio-economic status) is required to understand how personal connections influence consumption. Consumption (and resistance) associated with the self-management of long-term health problems has also been exemplified by Fox and Ward (2006) in a study of how specific groups of people and patients’ use of the Internet crystallises around notions of identity. They studied discussion forums, online clinics regarding the use of weight loss drugs and Viagra, as well as a discussion group focussed on anorexia nervosa to explore a range of identities from ‘expert patient’ to ‘resisting consumer’. The nature of identity formation in tandem with an increasingly consumer-oriented and technology-driven society hold
24
CAROLINE SANDERS AND ANNE ROGERS
implications for changes in the nature of networks and support for the experience and management of illness and disability. Contemporary cultural changes also give rise to other important dynamics to explore in relation to traditional variables, such as gender, ethnicity and SES. For example, Edwards (2004) provides a critical review of network research associated with changes in family structure. She claims that the breakdown of the traditional family and the rise of diverse family forms (e.g. lone mothers, same sex, co-habiting, step, dual worker) are regarded as creating social fragmentation and the erosion of social capital associated with the work of Putnam and others. However, she highlights the potential increase in resources arising from new forms of association created with the rise of diversity and fluidity in family life. The latter suggests the need to explore further the impact of these changes in family structure on the experience and management of chronic conditions. Social network research has also been particularly influential in highlighting the importance of place of residence in the creation of various social inequalities (Fischer, 1982; MacDonald, Shildrick, Webster, & Simpson, 2005; Warr, 2005). Such work has often focussed on changes in urban environments that are linked to socio-economic status, especially the association of urban decay with deprivation. Others have focussed on the impact of socio-economic decline in rural areas, or have contrasted communities living in deprived areas versus those in affluent areas to examine differentials in social capital generation (e.g. see Bagnall et al., 2003 on educational disparities). The relevance of ‘place’ for the experience and management of chronic illness and disability is highlighted by Shaw et al., who examined self-care support for people with diabetes in two contrasting and medically underserved communities (one rural and one urban). They examined the amount of support provided by key sources (including family and friends, community organisations, neighbours and neighbourhood, and resources in the wider community), and the associations between support from these sources and adherence to recommended diabetes self-care behaviours. Findings revealed different problematic health behaviours for the two communities with low rates of exercise in rural residents that was strongly associated with lack of neighbourhood resources. Whilst urban respondents reported more social support from all sources than rural respondents, there was a high level of smoking (40%) compared to the rural community (9%). The authors conclude that community-based interventions designed to improve diabetes self-care behaviour in underserved communities may be most effective if they include a focus not only on support from family and friends, but also on enhancing neighbourhood resources and creating links
Theorising Inequalities in the Experience and Management of Chronic Illness
25
between adults with diabetes and relevant community organisations (Shaw, Gallant, Riley-Jacome, & Spokane, 2006). The interface between organisational and informal networks for selfcare support is illustrated in studies of community pharmacies that are frequently used by proxy consulters on behalf of others (Rogers et al., 1999). Inequalities connected to place are indicated via different levels of engagement and receipt of advice and support from community pharmacists located in rural versus inner city deprived areas. In the former, the nature of contact and support is found to be more integrated with other community networks; whereas, barriers to such integration in the latter setting stemmed from perceived locality threat (Rogers, Hassell, Noyce, & Harris, 1998). Other examples of research focussing on the importance of ‘place’ for chronic illness experience and management have also considered the intersection of additional variables such as age. For example, Wen, Cagney, and Christakis (2005) examined the contextual effects of urban community environment on mortality among older individuals with existing serious illness living in Chicago. Analysis demonstrated that advantageous socioeconomic context contributed to lower mortality risk, but that social network density (size and frequency of interaction) was detrimental rather than protective; thus illustrating the complex relationship between social environments and health. The authors conclude that community-level interventions based on social capital/social cohesion models are unlikely to succeed without changes in the economic and health care realm for older people who are already ill. Schulz et al. (2006) also consider variables including ‘place’ in a study focussing on psychosocial stress and social support as mediators of relationships between income, length of residence and depressive symptoms among African American women on the eastside of Detroit. For women living in this racially segregated area with a high concentration of poverty, relationships between household income and symptoms of depression were found to be partially mediated by financial stress and social support, but neighbourhood disorder and discrimination influenced depressive symptoms independent of household income (Schulz et al., 2006). These studies point to the complex interaction of structural variables located within specific places but multiple questions remain under-explored: What is it like to live with a chronic illness in a specific neighbourhood (affluent or deprived)? How does place of residence impact upon social relationships (supportive and otherwise) and opportunities for accessing local health and social care resources? What is the relationship between social capital, ‘place’ and other contextual variables on inequalities in the experience and management of
26
CAROLINE SANDERS AND ANNE ROGERS
illness and disability? These questions raise additional issues for the conceptualisation and empirical study of network ties that are discussed in subsequent sections.
CONCEPTUALISING SOCIAL SUPPORT, NETWORK TIES, AND THE SIGNIFICANCE OF ORGANISATIONS AND TECHNOLOGY Distinctions have been drawn between discrete aspects of social support and network ties associated with informal relations (such as those based on kinship) and to examine how people are connected to (or excluded from) resources from formal organisations. Moreover, the creation of networks around the development and implementation of technologies has been a focus of organisational research and STS. New technologies are increasingly being deployed to provide support in the management of long-term conditions. However, this deployment is generally divorced from what is known about social support needs and existing support for chronic illness management. The early work of Weiss (1974) provides a useful typology of support comprising four aspects that are relevant to the experience and management of illness: emotional (love, caring, understanding), instrumental (practical help), appraisal (help with decision-making and agreeing on course of action) and informational (availability of advice and information) support. One study of social support and adjustment in three phases of illness (diagnostic, treatment and post-treatment) for patients with nasopharyngeal carcinoma specified four forms of social support similar to the Weiss typology: emotional, instrumental, informational and affiliational (Ma, 1998). Findings indicated that patients may need different types of social support at different stages of the illness (see also Morgan & Swann, 2004). Informational support may be more important in moderating the stress of the diagnostic phase, which is characterised by uncertainty and anxiety. In the treatment phase, instrumental support could help to relieve stress brought about by the physical effects of radiotherapy. Suitor, Wellman, and Morgan (1997) also highlight the importance of studying how and why social networks change and discuss the impact of rapid ‘catastrophic’ change (e.g. marital change) associated with social circumstances around which networks are organised. Similarly, various episodes in trajectories of illness are associated with ‘catastrophic’ change, such as the diagnosis of
Theorising Inequalities in the Experience and Management of Chronic Illness
27
serious (or even ‘life-threatening’) chronic illness. Less catastrophic changes may also occur with exacerbation of symptoms or problems related to illness or disability with implications for changing support needs. Social network researchers have examined the effects of both the structure (number of ties) and the content (the closeness of ties) of networks. A distinction has been drawn between strong ties developed through close personal relationships such as those within families, versus weaker ties with civil organizations and other more formal groups. Weak ties have been associated with the generation of ‘bridging’ capital because of the opportunities for mobility they provide across a relatively distal social space. In contrast, strong ties have been conceptualised as providing ‘bonding’ capital, illustrating the proximal nature of ties where actors are viewed to be metaphorically ‘glued’ together. These ties serve very different functions and are socially patterned in that groups with low socio-economic status tend to depend more on strong ties whilst the middle classes are more adept at making use of the strength of weak ties (Bagnall et al., 2003; Li et al., 2005). This is important in relation to access of material resources, because whilst ‘weak ties’ lack intimacy, they facilitate the diffusion of influence and information (Granovetter, 1973). Both early and later sociological work has focussed on the nature of reciprocal and trusting relations amongst neighbours within community settings (Bulmer, 1986; Li et al., 2005) and could be a focus for future work on illness experience and management given the implications for selfmanagement capacity. Network theorists have commented on the significance of specific forms of relationships such as friendships, and the distinctions between friends and acquaintances associated with contemporary social changes (Allan, 1996; Fischer, 1982; Pahl, 2000). ‘Naturally’ occurring informal relationships have been regarded as a source of social support and a potential resource that can be harnessed for supporting the management of chronic conditions, or even more formally as part of health care interventions in the form of ‘lay helpers’ or ‘be-friending’ programmes. The potential value of the latter is implied from evidence of the substitutive use of professional networks as a means of accessing social support. For example, loneliness has been significantly associated with frequency of consultation with general practitioners in British primary care (Ellaway, Wood, & Macintyre, 1999) and identified as a driver for recruitment to guided self-management programs run by formal health service organisations (Kennedy et al., 2005). Additionally, the notion of ‘fringe work’ refers to activities that professionals engage in as a response to perceived deficiencies in health and social care support systems; operating through the
28
CAROLINE SANDERS AND ANNE ROGERS
mobilisation and substitution of resources. This in turn is viewed as shaping reciprocal and trusting relationships between professionals and nonprofessional clients (De la Cuesta, 1993). In addition to the ‘substitution’ outlined above, the more formal ties that connect individuals with chronic conditions to professional and organisational support are relevant for changing systems of care within community settings (Pescosolido, 1996), where community institutions might act as ‘brokers’ for access to care (e.g. see Small, 2006). Recent work has distinguished between domains of social support including civic participation, neighbourhood attachment as well as networks of close personal relationships (Li et al., 2005). These distinctions are likely to assist with mapping the nature of support mechanisms and range of resources linked to everyday ways of living with chronic conditions within specific residential areas. Szreter and Woolcock (2004) review the distinctions between bonding and bridging capital, and the more recent concept of ‘linking’ capital. They define the latter as norms of respect and networks of trusting relationships between people who are interacting across explicit, formal or institutionalised power or authority gradients in society. Bridging is essentially a horizontal metaphor for connections between individuals that are more-orless equal in terms of status and power. In contrast, linking is presented as a ‘vertical’ metaphor to explore links across unequal power differentials, and viewed to have utility for exploring access to public and private services that can only be delivered through ongoing face-to-face interaction, such as the case of general practice medicine (2004, p. 655). They cite examples from previous research to illustrate the importance of linking capital, showing how in poor communities it is the nature and extent (or lack) of respectful and trusting ties to representatives of formal institutions (such as health care providers) that has a major bearing on peoples’ welfare. The concept of ‘linking’ capital seems especially relevant to studying the organisational context where informal self-management intersects with formal delivery of health services in meeting needs for those with chronic conditions. The example of general practice drawn above is one of many areas of concern in the management of chronic conditions where the focus on interaction, access and exclusion afforded by the network approach could be further exploited. Looman (2004) discusses the significance of horizontal versus vertical relationships with health care professionals in a qualitative study focussing on definitions of social capital emerging from research amongst family caregivers of children with chronic conditions. She found that families who were successful in ‘the system’ used ‘horizontal’
Theorising Inequalities in the Experience and Management of Chronic Illness
29
metaphors such as ‘teamwork’ to define their relationships with professionals as well as other community residents. Whilst there has been a policy shift in emphasis to notions of ‘patient-centredness’ and ‘partnership’ that seem consistent with horizontal network metaphors, how these social ties and relationships work in practice warrants further research. Pescosolido (2001) provides a useful model to study the ‘network episodes’ associated with the management of chronic conditions that necessarily involves movement across multiple care settings. At the heart of Pescosolido’s Multi-Level Network Model lies the idea that the ‘social structure’ which influences health and illness behaviour and outcomes is ‘‘the operation of professional, organisational, and community network ties’’ (Pescosolido, 1996, p. 176). Hirdes and Scott (1998) also provide an example of the network context of care within a chronic care hospital. However, there seems to be potential to extend such work across settings to take account of patients’ own self-management practices and experiences of managed care. This might usefully be achieved drawing upon Bordieu’s notion of habitus. Lin’s social capital theory emerged from social network analysis and extended Bordieu’s conception to identify three discrete processes: investing in, accessing, and gaining a return from social capital. These dynamic processes distinguish the concept of social capital from the more passive notion of social support often used in the context of chronic illness (Webber, 2007). Moreover, Bordieu’s concept of habitus makes it possible to further elaborate the connections between bodies and social structure in studying the experience and management of illness. Angus, Kontos, Dyck, McKeever, and Poland (2005) draw on Bordieu’s concept of habitus and field in their ethnography of long-term home care in Ontario; focussing on the home as a site where resources are processed, consumed and produced as in the case of family caregiving. Amongst participants, they found that changes in the concordance between the body and the field are prompted by impairments associated with illness and by an influx of caregivers: ‘‘This loss of an assumptive flow to everyday activity disrupts a practical sense long sedimented in the relationship between the care recipient’s bodies and the spaces and meanings of home’’ (2005, p. 166). They state that the logics and conflicts of the field of health -care become active within the home, which, as with the domestic field, already possesses its own logics and hierarchical arrangements. Bordieu argues that skilful action does not arise from an intellectual operation, but is a kind of creative response that comes from being embedded in a situation. Thus, Bordieu cites habitus as the generating source of practical creativity and innovation, as well as the embodiment of social structure (Crossley, 2001).
30
CAROLINE SANDERS AND ANNE ROGERS
Jinnett, Coulter, and Koegel (2002) highlight the need for a network approach that takes account of organisational environments which have a major impact on trajectories of care in throwing up various contingencies: ‘‘Providers in one sector may defer to the other sector (whether it be mental health or physical health); they may ignore the problem, which results in the individual client seeking care pretty late in the game; they may work together in a concerted fashion to address problems but still operate as separate entities, thus duplicating efforts; they may merge functions under one roof both in terms of single organisation and a single location; or they may create multi-role teams to address complex issues across agencies’’ (p. 105). As they state, the nature of mental health treatment is people intensive and coordination involves multiple providers (case manager, nurse, psychiatrist, psychologist, social worker) sharing views about patient treatment (p. 106). They also discuss the ‘slippages’ that can occur when trying to integrate new technology into role relationships between specific health professionals that can throw up issues of conflict. This is likely to have relevance to the provision of services for chronic illnesses more generally, and is exemplified by evaluation of attempts to bring new technologies into defined areas of practice (e.g. see May et al., 2001 on the case of telepsychiatry). The proliferation and implementation of technologies to enhance self-care support bring new issues of equity and access to the fore. The expansion of information and communication technologies (ICTs) has increased forms of online support for those with chronic conditions including conditionspecific support groups and telehealth services such as online consultations, remote monitoring of signs and symptoms to aid self-management, and the provisions of self-management advice. However, there are concerns about inequitable access where deprived people are least likely to have Internet access in their own homes (Mead, Varnam, Rogers, & Roland, 2003). Drentea and Moren-Cross (2005) report the generation of heterogeneous ties and social capital via an Internet mother site, although benefits were restricted to the more affluent mothers who used the site. In some cases, new technological interventions to improve self-management of chronic conditions have been targeted in deprived communities. For example, Lindsay, Smith, Bell, and Bellaby (2007) conducted a qualitative pilot study to evaluate whether facilitated access to the Internet may improve capacity of older men living in a deprived inner city area to manage their heart conditions. They found the intervention to be worthwhile in strengthening social support and facilitating behaviour changes that enhanced selfmanagement.
Theorising Inequalities in the Experience and Management of Chronic Illness
31
The integration of new assistive technologies into self-care support necessarily requires sociologists to engage in the evaluation of such interventions and this raises new challenges with theoretical and methodological implications. Some of the distinctions and overlaps between relativist and realist approaches to studying body–society connections once again come to the fore in reviewing the varied approaches to studying technologies in society. May, Mort, Williams, Mair, and Gask (2003), May and Ellis (2001), May et al. (2001), May, Rapley, Moreira, Finch, and Heaven (2006) have successfully drawn upon theoretical, conceptual and empirical insights emerging from research in the field of STS where previous research has sought to examine the ‘translation’ of new technologies into medical practice. An important focus of such work has been on the interaction within networks evolved via the design and implementation of new technologies that is termed actor network theory (ANT). As discussed earlier this approach is distinguished from traditional social network analysis by its strong relativist stance (Knox et al., 2006). Such an approach is deemed valuable for researching the social relations that have a significant impact on shaping the trajectory of how telecare interventions come to be adopted or rejected within the context of existing organisation of care and caring relationships (May et al., 2003). Recent studies of ‘technology in practice’ have attempted to avoid a deterministic and essentialist focus on negative assumptions about new technologies and also avoid privileging specific social categories such as ‘gender’ or ‘class’ within analyses (Timmermans & Berg, 2003). Moreover, researchers adopting an ANT approach have often used ethnographic methods to inform fine-grained analyses of the subtle dynamics and networks of relations inherent in the implementation of new medical technologies. Examples have focused on technologies used in Accident and Emergency departments (Timmermans, 1998), diagnostics (Mol & Elsman, 1996), IVF clinics (Cussins, 1998) and the use of medical records (Berg, 1996). This approach has also been used successfully to study the role of support groups in settling controversy around the contested diagnostic category of Repetitive Strain Injury (RSI) (Arksey, 1994). The conceptualisation of networks within ANT differs from traditional social network analysis with the ‘rhizome’ metaphor (from the work of Deleuze & Guattari, 1988) sometimes drawn upon by ANT theorists to depict ‘a multiplex, heterogeneous and robust web of relations’ (Grabher, 2006; p. 166). Grabher describes this as a valuable alternative to the ‘tie and node’ imagery often presented in more traditional approaches. We suggest that the theoretical and methodological approach of ANT has much to offer in studying new and important issues for inequalities in the experience and management of
32
CAROLINE SANDERS AND ANNE ROGERS
chronic conditions, including the deployment of new information and assistive technologies, as well as the growth of ‘self-help’ support groups for specific (and contested) conditions. It is to the methodological issues we now turn in discussing possible ways forward for research in this field.
TRANSLATING THEORY TO METHOD: A WAY FORWARD FOR RESEARCHING SOCIAL NETWORKS AND INEQUALITIES IN THE EXPERIENCE AND MANAGEMENT OF CHRONIC ILLNESS Much of the existing work studying health inequalities in relation to social capital and health has provided quantitative evidence to support findings related to disparities. Such work has served a valuable purpose in placing goals to reduce health inequalities on policy agendas. However, the dominance of a positivist perspective within mainstream social network analysis has resulted in a lack of imagination regarding theory and methods for researching health inequalities (Scambler et al., 2002) and social networks (Knox et al., 2006). Moreover, many early measures of social capital have been derived from secondary sources (mainly surveys) that were not explicitly designed to measure ‘social capital’ (Szreter & Woolcock, 2004). Lynch, Due, Muntaner, and Davey-Smith (2000) also state that quantitative aggregate measures are often blunt and unsophisticated, tending to mask some of the effects and actions of social capital. Scambler et al. (2002) support a ‘critical realist’ perspective for researching health inequalities based on two modes of inference: ‘retroduction’ and ‘abduction’. Quantitative methodology is appropriate for the former, because retroduction involves moving from a knowledge of events to a knowledge of mechanisms contributing to the generation of events. Whilst identification of mechanisms is also implicated in abduction, qualitative research is most appropriate because it allows casting of familiar events (via ‘redescription’ or ‘retextualisation’) within a new and innovative frame of reference (Scambler et al., 2002, p. 59). Here Scambler et al. also flag the importance of recent research drawing on the work of Bordieu and his concept of habitus to highlight the ‘indebtedness of agency to structure’ (p. 62) as one fruitful approach for further work on the subject of inequalities. Qualitative studies afford major potential for examining aspects of relationships and community structures that are often hard to measure (Swann & Morgan, 2006), and Jinnett et al. (2002) point to the notion of a
Theorising Inequalities in the Experience and Management of Chronic Illness
33
‘negotiated order’ (after Strauss et al., 1963) as a way of avoiding a crude structural determinist bias. This is important to ensure that the impact of structure on embedded cases is not exaggerated whilst neglecting the impact of cases on context itself: ‘‘In this way network analysis can also be usefully wedded to the notion of a grounded research methodology’’ (Jinnett et al., 2002, p. 108). Blaxter and Poland (2006) claim that survey questions traditionally used in social network analyses risk tautology because they appear to measure both the things thought to create social capital and the things which are the consequences of disparities in social capital. For example, social networks create social capital, but high social capital creates cohesive societies. They also found that there are conflicts between individualising and structural approaches because variables might be significant at the community level but not at the individual level. Their qualitative research found that some items featured in surveys to examine network ties were meaningless amongst older urban citizens, and responses illustrated the complexities often masked by such questions. For example, they found that whilst there was a tendency for older respondents to claim they had ‘a good family’ and ‘good neighbours’, many then contradicted this by describing conflict and neglect. Such complexities have also been revealed in research on illness experience where older people present an optimistic view of their illness experience and management, whilst also revealing extensive problems and suffering (Sanders, Donovan, & Dieppe, 2002). Some have drawn upon mixed methods to study the relationship between social networks and health inequalities (e.g. Ziersch, Baum, MacDougall, & Putland, 2005). Pescosolido, Brooks-Gardner, and Lubell (1998) used mixed methods to study the pathways followed into mental health services for people with mental health problems and presented qualitative data alongside quantitative data illustrating that even where people report that they made an expressed choice to seek formal care, they were often ‘supported’ in such choices through their social networks. The qualitative data was also important in exploring unusual cases that did not conform to the overall quantitative findings. The ‘complexities’ of chronic care trajectories have also been explored using an ethnographic approach (Allen, Griffiths, & Lyne, 2004). A particular strength of previous qualitative research on the experience of chronic illness has emerged from the exploration of narrative accounts (Bury, 2001; Kelly & Dickinson, 1997; Robinson, 1990; Williams, 1984; Radley, 1993). This narrative focus might usefully be extended to research the nature of network ties associated with SES and ‘place’ in relation to inequalities in the
34
CAROLINE SANDERS AND ANNE ROGERS
experience and management of chronic conditions, whilst drawing theoretically from network research. As Jinnett et al. (2002) argue, there is considerable potential to bring together micro and macro analytic approaches within network research. They propose that an ‘embedded’ qualitative case study design is most appropriate for the careful analysis of linkages between individuals within multilevel contexts incorporating social systems, organisations and environments. Previous research has also focussed on the need for considering chronic illness experience and self-management within a ‘whole system’ context distinguishing between three levels: the patient, the professional and the organisational structure (Kennedy & Rogers, 2001; Pescosolido, 1996). As Jinnett et al. claim, quantitative studies can identify the way the system is organised structurally, and how the clinics are related to one another and the system as a whole; however, it cannot capture what motivates these structural relationships. The latter can be captured by qualitative methods, as can the patient’s experiences of day-to-day living with illness within (or excluded from) specific social networks.
CONCLUSION To date there has been a limited body of work focussing directly on social networks and chronic illness and disability although some has focused on aspects of social support. The wider sociological literature on chronic illness and disability has long drawn attention to the importance of social relationships in mediating experiences and management. However, the recent conceptual and empirical developments stemming from studies focussed specifically on networks of social relations holds considerable potential for researching inequalities in the experience and management of chronic conditions. Knox et al. (2006) contrast the positivist approaches emerging from traditional forms of social network analysis with the postpositivist (and relativist) work of actor network theorists and point to opportunities for new kinds of network thinking drawing from both, whilst cautioning ‘a need to recognise the limits of any appeal to networks as a kind of holy grail’ (p. 129). We argue that network thinking could add important insights in the context of policy concerns to capitalise on social networks for the purpose of maximising self-management capacity. A postpositivist orientation in this area is potentially of value in accounting for both structure and action that are at the root of macro and micro concerns of chronic illness and disability experience.
Theorising Inequalities in the Experience and Management of Chronic Illness
35
Network approaches prompt a critical focus on social and cultural change that inevitably impacts upon experience and management. Contemporary changes associated with consumption and technological developments bring new issues to the fore for the research of chronic conditions that might usefully be explored drawing upon social network perspectives. This is evident in changing networks of support facilitated by the Internet via chat rooms and online support groups, and via new systems of telecare that bring new assistive technologies into the homes of people living with chronic conditions. The use of such technologies will necessarily influence and be influenced by networks of care around those using them. There are additional questions regarding equity around access and use of such technologies that require further research. Rural and urban changes have implications for the nature of support networks within specific communities and have been shown to be important contributors to inequalities in other aspects of social life. Where places are known to disadvantage people in accessing other social resources such as education, it seems likely that the experiences associated with social exclusion are likely to be compounded for people with chronic conditions. Similarly, place of residence and the nature of network ties fostered within specific settings are obviously important for access to material resources as is the role of the state and professionals in the provision of services. For example, Szreter and Woolcock (2004) draw attention to the value of an historical materialist perspective in this respect, stating that social capital is as much about styles and forms of leadership and activism among health workers as it is about the seemingly abstract properties of social cohesion among communities or social collectivities of various kinds. The notion of ‘linking’ capital seems a particularly important concept in this respect for exploring inequalities related to chronic illness. The use of voluntary associations and support groups focusing on particular illness conditions are a growing and reportedly valuable part of the social and informational support provided to a range of groups in the population experiencing chronic illness. However, the extent to which these organisations mobilise and utilise lay resources in practice is a moot point (De la Cuesta, 1993). Thus, the distribution, access to and use of such resources constitutes an axis of inequality that warrants exploration in promoting our understanding of support for living with long-term conditions. Finally, this review demonstrates that there is scope for further research to illuminate inequalities in the experience and management of chronic conditions from a social networks perspective drawing as much on
36
CAROLINE SANDERS AND ANNE ROGERS
qualitative approaches as quantitative ones. Future work might be conducted most productively by drawing on the conceptual and theoretical developments within network research, whilst also drawing on the theoretical and empirical strengths of previous qualitative research on the experience and management of chronic conditions.
REFERENCES Adamson, J., Lawlor, D. A., & Ebrahim, S. (2004). Chronic diseases, locomotor activity limitation and social participation in older women: Cross sectional survey of British Women’s Heart and Health Study. Age and Ageing, 33, 293–298. Akiyama, H., & Antonucci, T. (2002). Gender differences in depressive symptoms: Insights from a life span perspective on life stages and social networks. In: B. A. Pescosolido & J. A. Levy (Eds), Social networks and health. New York: JAI Press. Allan, G. (1996). Kinship and friendship in modern Britain. Oxford: Oxford University Press. Allen, D., Griffiths, L., & Lyne, P. (2004). Understanding complex trajectories in health and social care provision. Sociology of Health and Illness, 26, 1008–1030. Anderson, R., & Bury, M. (1988). Living with chronic illness: The experience of patients and their families. London: Unwin Hyman. Angus, J., Kontos, P., Dyck, I., McKeever, P., & Poland, B. (2005). The personal significance of home: Habitus and the experience of receiving long-term home care. Sociology of Health and Illness, 27(March), 161–187. Antonucci, T. C., & Akliyama, H. (1987). An examination of sex differences in social support among older men and women. Sex Roles, 17, 737–749. Arksey, H. (1994). Expert and lay participation in the construction of medical knowledge. Sociology of Health and Illness, 16, 448–468. Armstrong, D. (1983). Political anatomy of the body. Cambridge: Cambridge University Press. Bagnall, G., Longhurst, B., & Savage, M. (2003). Children, belonging and social capital: The PTA and middle class narratives of social involvement in the north-west of England. Sociological Research Online, 8. Barnes, C. (1998). The social model of disability: A sociological phenomenon ignored by sociologists? In: T. Shakespeare (Ed.), The disability reader: Social science perspectives (pp. 65–78). London: Cassell. Barnes, C., & Mercer, G. (1996). Exploring the divide: Illness and disability. Leeds: The Disability Press. Berg, M. (1996). Practices of reading and writing: The constitutive role of the patient record in medical work. Sociology of Health and Illness, 18, 499–562. Berkman, L. F., Glass, T., Brissette, I., & Seeman, T. E. (2000). From integration to health: Durkheim in the new millenium. Social Science and Medicine, 51, 843–857. Blaxter, M., & Poland, F. (2006). Moving beyond the survey in exploring social capital. In: C. Swann & A. Morgan (Eds), Social capital for health. Insights from qualitative research (pp. 87–107). London: NHS Health Development Agency. Bott, E. (1957). Family and social network: Roles, norms and external relationships in urban families. London: Tavistock.
Theorising Inequalities in the Experience and Management of Chronic Illness
37
Brown, G. W., & Harris, T. O. (1978). Social origins of depression: A study of psychiatric disorder in women. New York: Free Press. Brown, G. W., & Harris, T. O. (1989). Life events and illness. London: Guilford Press. Bulmer, M. (1986). Neighbours: The work of Philip Abrams. Cambridge: Cambridge University Press. Bury, M. (1986). Social constructionism and the development of sociology. Sociology of Health and Illness, 8, 137–169. Bury, M. (1997). Health and illness in a changing society. London: Routledge. Bury, M. (2001). Illness narratives: Fact or fiction? Sociology of Health and Illness, 23, 263–285. Carter, W. C., & Feld, S. L. (2004). Principles relating social regard to size and density of personal networks, with applications to stigma. Social Networks, 26, 323–329. Coleman, L. M. (1997). Stigma-An Enigma Demystified. In: L. Davis (Ed.), The disability studies reader (pp. 216–231). London: Routledge. Crossley, N. (2001). The phenomenological habitus and its construction. Theory and Society, 30, 81–120. Cussins, C. M. (1998). Ontological choreography: Agency for women patients in an infertility clinic. In: M. Berg & A. Mol (Eds), Differences in medicine: Unravelling practices, techniques and bodies. London: Duke University Press. Davey Smith, G., & Lynch, J. (2004). Commentary: Social capital, social epidemiology and disease aetiology. International Journal of Epidemiology, 33, 691–700. De la Cuesta, C. (1993). Fringe work-peripheral work in health visiting. Sociology of Health and Illness, 15, 665–682. Deleuze, G., & Guattari, F. (1988). A thousand plateaus. Capitalism and schizophrenia, London: Athlone. Drentea, P., & Moren-Cross, J. L. (2005). Social capital and social support on the web: The case of an Internet mother site. Sociology of Health and Illness, 27, 920–943. Durkheim, E. (1951). Suicide. New York: Free Press. Edwards, R. (2004). Present and absent in troubling ways: Families and social capital debates. The Sociological Review, 1–21. Ellaway, A., Wood, S., & Macintyre, S. (1999). Someone to talk to? The role of lonliness as a factor in the frequency of GP consultations. British Journal of General Practice, 49, 363–367. Finch, J., & Mason, J. (1993). Negotiating family responsibilities. London: Routledge. Fischer, C. S. (1982). To dwell among friends: Personal netwroks in town and city. Chicago: The University of Chicago Press. Fox, N., & Ward, K. (2006). Health identities: From expert patient to resisting consumer. Health, 10, 461–479. French, S. (1993). Disability, impairment or something in between? In: J. Swain, V. Finkelstein, S. French & O. Oliver (Eds), Disabling barriers-enabling environments (pp. 17–25). London: Sage. Freund, P. (2006). Socially constructed embodiment: Neurohormonal connections as resources for theorizing about health inequalities. Social Theory and Health, 4, 85–108. Gove, W. R. (1984). Gender differences in mental and physical illness: The effects of fixed roles and nurturant roles. Social Science and Medicine, 19, 77–91. Grabher, G. (2006). Trading routes, bypasses, and risky intersections: Mapping the travels of ‘netwroks’ between economic sociology and economic geography. Progress in Human Geography, 30, 163–189.
38
CAROLINE SANDERS AND ANNE ROGERS
Graham, H. (1983). Caring: A labour of love. In: J. Finch & D. Groves (Eds), A labour of love: Women, work and caring. London: Routledge and Kegan Paul. Granovetter, M. S. (1973). The strength of weak ties. The American Journal of Sociology, 78, 1360–1380. Gregory, S. (2005). Living with chronic illness in the family setting. Sociology of Health and Illness, 27, 372–392. Haney, C. A. (1980). Life events as precursors of coronary heart disease. Social Science and Medicine, 14A, 119–126. Hart, T. (1971). The inverse care law. The Lancet, 405–412. Hirdes, J. P., & Scott, K. A. (1998). Social relations in a chronic care hospital: A whole network study of patients, family and employees. Social Networks, 20, 119–133. Jinnett, K., Coulter, I., & Koegel, P. (2002). Cases, contexts and care: The need for grounded network analysis. In: B. A. Pescosolido & J. A. Levy (Eds), Social networks and health. New York: JAI Press. Kahn, R. L. (2006). Aging and social support. In: M. W. Riley (Ed.), Aging from birth to death: Interdisciplinary perspectives. Boulder, CO: Westview Press. Kaplan, R. M., & Toshima, M. T. (1990). The functional effects of social relationships on chronic illnesses and disability. In: B. K. Sarason, I. G. Sarason & G. R. Pierce (Eds), Social support: An interactional view. Chichester: Wiley. Kelly, M. P., & Dickinson, H. (1997). The narrative self in autobiographical accounts of illness. The Sociological Review, 42, 254–278. Kelly, M. P., & Field, D. (1996). Medical sociology, chronic illness and the body. Sociology of Health and Illness, 18, 241–257. Kendall, E., & Rogers, A. (2007). Extinguishing the social? State sponsored self-care policy and the chronic disease self-management programme. Disability and Society, 22, 129–143. Kennedy, A., & Rogers, A. (2001). Improving self-management skills: A whole systems approach. British Journal of Nursing, 10, 734–737. Kennedy, A., Rogers, A., & Gately, C. (2005). Assessing the introduction of the expert patients programme into the NHS. Primary Care Research and Development, 137–148. Kennedy, B. P., Kawachi, I., Prothrow-Smith, D., Lochner, K., & Gupta, V. (1998). Social capital, income inequality, and firearm violent crime. Social Science and Medicine, 47, 7–17. Knox, H., Savage, M., & Harvey, P. (2006). Social networks and the study of relations: Networks as method, metaphor and form. Economy and Society, 35, 113–140. Levy, J. A., & Pescosolido, B. A. (2002). Social networks and health. New York: JAI Press. Li, Y., Pickles, A., & Savage, M. (2005). Social capital and social trust in Britain. European Sociological Review, 21, 109–123. Lindsay, S., Smith, S., Bell, F., & Bellaby, P. (2007). Tackling the digital divide: Exploring the impact of ICT on managing heart conditions in a deprived area. Information, Communication and Society, 10, 95–114. Looman, W. S. (2004). Defining social capital for nursing: Experiences of family caregivers of children with chronic conditions. Journal of Family Nursing, 10, 412–428. Lynch, J., Due, P., Muntaner, C., & Davey-Smith, G. (2000). Social capital-is it a good investment strategy for public health? Journal of Epidemiology and Community Health, 54, 404–408. Ma, J. L. C. (1998). Effect of perceived social support on adjustment of patients suffering from nasopharyngeal carcinoma. Health and Social Work, 23, 167–174.
Theorising Inequalities in the Experience and Management of Chronic Illness
39
MacDonald, R., Shildrick, T., Webster, C., & Simpson, D. (2005). Growing up in poor neighbourhoods: The significance of class and place in the extended transitions of ‘socially excluded’ young adults. Sociology, 39, 873–891. Marmot, M. G., Rose, G., Shipley, M., & Hamilton, P. J. (1978). Employment grade and coronary heart disease in British civil servants. Journal of Epidemiology and Community Health, 32, 244–249. May, C., & Ellis, N. T. (2001). When protocols fail: Technical evaluation, biomedical knowledge, and the social production of ‘facts’ about a telemedicine clinic. Social Science & Medicine, 53, 989–1002. May, C., Gask, L., Atkinson, T., Ellis, N., Mair, F., & Esmail, A. (2001). Resisting and promoting new technologies in clinical practice: The case of telepsychiatry. Social Science & Medicine, 52, 1889–1901. May, C., Mort, M., Williams, T., Mair, F., & Gask, L. (2003). Health technology assessment in its local contexts: Studies of telehealthcare. Social Science & Medicine, 57, 697–710. May, C., Rapley, T., Moreira, T., Finch, T., & Heaven, B. (2006). Technogovernance: Evidence, subjectivity, and the clinical encounter in primary care medicine. Social Science & Medicine, 62, 1022–1030. Mead, N., Varnam, R., Rogers, A., & Roland, M. (2003). What predicts patients’ interest in the Internet as a health resource in primary care in England? Journal of Health Services Research and Policy, 8, 33–39. Mol, A., & Elsman, B. (1996). Detecting disease and designing treatment: Duplex and the diagnosis of diseased leg vessels. Sociology of Health and Illness, 18, 609–631. Morgan, A., & Swann, C. (2004). Social capital for health: Issues of definition, measurement and links to health. London: NHS Health Development Agency. Morgan, M., Patrick, D. L., & Charlton, J. R. (1984). Social networks and psychosocial support among disabled people. Social Science and Medicine, 19, 489–497. Nettleton, S. (1995). The sociology of health and illness. Cambridge: Polity. Oliver, M. (1996). Understanding disability: From theory to practice. London: Macmillan. Pahl, R. (2000). On friendship. Cambridge: Polity. Parkes, C. M., Benjamin, B., & Fitzgerald, R. G. (1969). Broken heart: A statistical study of increased mortality among widowers. British Medical Journal, 1, 740–743. Pescosolido, B. A. (1996). Bringing the ‘Community’ into utilization models: How social networks link individuals to changing systems of care. In: J. J. Kronenfeld (Ed.), Research in the sociology of health care (pp. 171–198). Greenwich, CT: JAI Press. Pescosolido, B. A. (2001). The role of social networks in the lives of persons with disabilities. In: G. L. Albrecht, K. D. Seelman & M. Bury (Eds), Handbook of disability studies (pp. 468–489). London: Sage. Pescosolido, B. A., Brooks-Gardner, C., & Lubell, K. M. (1998). How people get into mental health services: Stories of choice, coercion and muddling through from first timers. Social science and medicine, 46, 275–286. Pescosolido, B. A., & Kronenfeld, J. J. (1995). Health, illness, and healing in an uncertain era: Challenges from and for medical sociology. Journal of Health and Social Behavior, 35, 5–33. Pescosolido, B. A., & Levy, J. A. (2002). The role of social networks in health, illness, disease and healing: The accepting present, the forgotten past, and the dangerous potential for a complacent future. In: J. A. Levy & B. A. Pescosolido (Eds), Social networks and health (pp. 3–28). New York: JAI Press.
40
CAROLINE SANDERS AND ANNE ROGERS
Pescosolido, B. A., & Rubin, B. A. (2000). The web of group affiliations revisited: Social life, postmodernism and sociology. American Sociological Review, 65, 52–76. Pilgrim, D., & Rogers, A. (1994). Something old, something new y: Sociology and the organisation of psychiatry. Sociology, 28, 521–538. Putnam, R. D., Leonardi, R., & Nanetti, R. Y. (1993). Making democracy work: Civic traditions in modern italy. Princeton: Princeton University Press. Radley, A. (1993). Worlds of illness: Biographical and cultural perspectives on health and disease. London: Routledge. Reidpath, D. D., Chan, K. Y., Gifford, S. M., & Allotey, P. (2005). ‘He hath the French pox’: Stigma, social value and social exclusion. Sociology of Health and Illness, 27, 468–489. Robinson, I. (1990). Personal narratives, social careers and medical courses: Analysing life trajectories in autobiographies of people with multiple sclerosis. Social Science & Medicine, 30, 1173–1186. Roche, J., & Tucker, S. (2003). Extending the social exclusion debate: An exploration of the family lives of young carers and young people with ME. Childhood, 10, 439–456. Rogers, A., Hassell, K., & Nicolaas, G. (1999). Demanding patients? analysing the use of primary care. Buckingham: Open University Press. Rogers, A., Hassell, K., Noyce, P., & Harris, J. (1998). Advice-giving in community pharmacy: Variations between pharmacies in different locations. Health and Place, 4, 365–373. Rogers, A., & Pilgrim, D. (1991). Pulling down churches: Accounting for the British mental health service users movement. Sociology of Health and Illness, 13, 129–148. Rogers, A., & Pilgrim, D. (2005). A sociology of mental health and illness. Milton Keynes: Open University Press. Rook, K. (1992). Detrimental aspects of social relationships: Taking stock of an emerging literature. In: H. O. Viel & U. Baumann (Eds), The meaning and measurement of social support. New York: Hemisphere. Sanders, C., Donovan, J., & Dieppe, P. (2002). The significance and consequences of having painful and disabled joints in older age: Co-existing accounts of normal and disrupted biographies. Sociology of Health and Illness, 24, 227–253. Scambler, G., Higgs, P., & Jones, I. R. (2002). A critical realist perspective on class relations and health inequalities. In: J. J. Kronenfeld (Ed.), Social inequalities, health and health care delivery (20th ed., pp. 57–75). Greenwich, CT: JAI Press. Schulz, A. J., Israel, B. A., Zenk, S. N., Parker, E. A., Lichtenstein, R., Shellman-Weir, S. et al. (2006). Psychosocial stress and social support as mediators of relationships between income, length of residence and depressive symptoms among African American women on Detroit’s eastside. Social Science and Medicine, 62, 510–522. Shaw, B. A., Gallant, M. P., Riley-Jacome, M., & Spokane, L. S. (2006). Assessing sources of support for diabetes self-care in urban and rural underserved communities. Journal of Community Health, 31, 393–412. Shippy, R. A., & Karpiak, S. E. (2005). Perceptions of support among older adults with HIV. Research on Aging, 27, 290–306. Simmel, G. (1955). Conflict and the web of group affiliations. New York: Free Press. Small, M. L. (2006). Neighbourhood institutions as resource brokers: Childcare centers, interorganisational ties, and resource access among the poor. Social Problems, 53, 274–292. Strauss, A., Schtzman, L., Ehrlich, D., Bucher, R., & Sabshin, M. (1963). The hospital and its negotiated order. In: E Freidson (Ed.), The hospital in modern society. New York: Free Press.
Theorising Inequalities in the Experience and Management of Chronic Illness
41
Suitor, J. J., Wellman, B., & Morgan, D. L. (1997). It’s about time: How, why, and when networks change. Social Networks, 19, 1–7. Swann, C., & Morgan, A. (2006). Social capital for health. London: NHS Health Development Agency. Szreter, S., & Woolcock, M. (2004). Health by association? Social capital, social theory, and the political economy of public health. International Journal of Epidemiology, 33, 650–667. Thomas, C. (1995). Domestic labour and health: Bringing it all back home. Sociology of Health and Illness, 17, 328–352. Thomas, C. (1997). The baby and the bathwater: Disabled women and motherhood in social context. Sociology of Health and Illness, 19, 622–643. Townsend, A., Wyke, S., & Hunt, K. (2006). Self-managing and managing self: Practical and moral delimmas in accounts of living with chronic illness. Chronic Illness, 2(3), 185–194. Timmermans, S. (1998). Resuscitation technology in the emergency department: Towards a dignified death. Sociology of Health and Illness, 20, 144–167. Timmermans, S., & Berg, M. (2003). The practice of medical technology. Sociology of Health and Illness, 25, 97–114. Turner, B. S. (1992). Regulating bodies: Essays in medical sociology. London: Routledge. Turner, B. S. (2001). Disability and the sociology of the body. In: G. L. Albrecht, K. D. Seelman & M. Bury (Eds), Handbook of disability studies (pp. 252–266). London: Sage. Veiel, H. O. F., & Baumann, U. (1992). The many meanings of social support. In: H. O. F. Veiel & U. Baumann (Eds), The meaning and measurement of social support. London: Hemisphere publishing. Walker, A. (1987). Enlarging the caring capacity of the community: Informal support networks and the welfare state. International Journal of Health Services, 17, 369–386. Warde, A., & Tampubolon, G. (2002). Social capital, networks and leisure consumption. The Sociological Review, 155–180. Warr, D. J. (2005). Social networks in a ‘discredited’ neighbourhood. Journal of Sociology, 41, 285–308. Webber, M. (2007). Individual social capital and depression: Findings from a primary care cohort study. Conference presentation at The University of Manchester: Mental health, social exclusion and inequality. (Funded by ESRC, March, 2007). Weiss, R. S. (1974). The provisions of social relationships. In: Z. Rubin (Ed.), Doing unto others. Englewood Cliffs, NJ: Prentice Hall. Wen, M., Cagney, K. A., & Christakis, N. A. (2005). Effect of specific aspects of community social environment on mortality of individuals diagnosed with serious illness. Social Science and Medicine, 61, 1119–1134. Williams, G. (1984). The genesis of chronic illness: Narrative re-construction. Sociology of Health and Illness, 6, 175–200. Williams, G. (1989). Hope for the humblest? The role of self-help in chronic illness: The case of ankylosing spondylitis. Sociology of Health and Illness, 11, 135–159. Williams, G. (1996). Representing disability: Some questions of phenomonology and politics. In: C. Barnes & G. Mercer (Eds), Exploring the divide: Illness and disability (pp. 194–212). Leeds: The Disability Press. Williams, G. (1999a). Review article: Bodies on the battlefield. The dialectics of disability. Sociology of Health and Illness, 21, 242–252.
42
CAROLINE SANDERS AND ANNE ROGERS
Williams, G., & Busby, H. (2000). The politics of ‘disabled’ bodies. In: S. J. Williams, J. Gabe & M. Calnan (Eds), Health, medicine & society (pp. 169–185). London: Routledge. Williams, S. J. (1998). Capitalising on emotions? Rethinking the inequalities in health debate. Sociology, 32, 121–139. Williams, S. J. (1999b). Is anybody there? Critical realism, chronic illness and the disability debate. Sociology of Health and Illness, 21, 797–819. Ziersch, A. M., Baum, F. E., MacDougall, C., & Putland, C. (2005). Neighbourhood life and social capital: The implications for health. Social Science and Medicine, 60, 71–86. Zola, I. K. (1991). Bringing our bodies and ourselves back in: Reflections on a past, present, and future ‘‘medical sociology’’. Journal of Health & Social Behavior, 32, 1–16.
U.S. MINORITIES’ ACCESS TO HEALTH CARE UNDER MANAGED CARE: A SYNTHESIS OF THE LITERATURE Greg A. Greenberg ABSTRACT In this chapter organizational theory is used to clarify and synthesize the large and diverse literature on the relationship between managed care (MC) and ethnic differences in access to health services. MC practices are classified by whether they are used by health care organizations to define their boundaries or to coordinate care. MC practices used to coordinate care are further categorized as one of five types: rules and programs, authority, goal setting, culture, or client coordination. This review also presents hypotheses derived from this literature that specify the predicted effects of MC practices on ethnic differences in access to health services. It was found that few of these hypotheses had been empirically investigated and although some evidence was found that MC boundary-setting practices disadvantage minorities, there were not consistent findings with respect to those practices used to coordinate care.
Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 45–76 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00003-8
45
46
GREG A. GREENBERG
Equity in the delivery of health care services is a major concern of policy makers as well as purchasers, providers, and recipients of care. A recent report by the Institute of Medicine documented the existence of substantial barriers to obtaining appropriate health care among ethnic and racial minorities in the U.S., even when insurance and health status, income and age are held constant (IOM, 2003). Of particular concern is whether the increasing implementation of managed care (MC) practices, such as utilization review (UR) and capitation, that often constrain the delivery of health services, has a greater impact on minorities than on other Americans. There is now a very large body of work that addresses various aspects of this issue: how MC practices affect ethnic1 differences in access to health care. However, this literature remains disorganized and social scientific theorizing on the topic remains undeveloped. In this chapter we use organization theory to synthesize and clarify this literature. Specifically, a typology from organization theory is used to guide the discussion, allowing for some meaningful order to be imposed on the diverse literature, bringing attention to disagreements and agreements in the literature, and facilitating the application of other concepts and theoretical tools developed in the study of organizations. Exploratory hypotheses are also specified so as to facilitate an examination of the relevant empirical evidence and to bring attention to subject areas where research is lacking. Synthesizing and further theorizing the literature and reviewing the empirical evidence that addresses how MC practices affect ethnic differences in access to health services should facilitate and encourage research on this important subject.
BACKGROUND The American Medical Association very broadly defines MC practices as the arrangements of any entity that delivers, administers, or assumes risk for health services that are implemented in order to control or influence the quality, accessibility, utilization, costs, prices, or outcomes of services provided to a defined population (Elder, 2002). Many types of health care organization make intensive use of MC practices, including independent provider organizations, provider groups, and physician-hospital organizations. MC practices have spread even to integrated public health systems such as the Veterans Health Administration, which in 1995 divided into 22 semi-autonomous networks, introduced prospective reimbursement, and implemented performance monitoring systems tied to administrators’ compensation (Greenberg & Rosenheck, 2003). However, to simplify the
U.S. Minorities’ Access to Health Care under MC
47
discussion, as does the literature, the focus here will be on a large variety of organizations that are referred to as managed care organizations (MCOs), such as health maintenance organizations, preferred provider organizations, and point of service plans. By the 1990s MCOs provided health services to 85 per cent of insured employees (Kuttner, 1999) and by 2000 to over half of all Medicaid recipients (Rosenbaum, 2003). MCOs have in common that they take responsibility for coordinating both the financing and delivery of health services. However, MCOs vary in many ways, including how much risk they share with other actors (such as payers and providers), the restrictiveness of their networks, and whether providers are contractors or employees. The most common explanation for the spread of MC practices is that they offer effective mechanisms for cost containment in an era of rising health care costs and declining government support for public health services (Grembowski, Cook, Patrick, & Roussel, 2002; Schlesinger & Gray, 1999). The assumption that MCOs improve service quality has also facilitated the spread of MC practices. For instance, many advocates and administrators of public insurance programs believe that MCOs offer beneficiaries access to the same services as high income people or at least increase beneficiaries choice of providers, availability of preventive care, and coordination of services. For the public payer MCOs are also believed to have the advantage of delivering services through an accountable organization (Fosset & Thompson, 1999; Grembowski et al., 2002; Schlesinger & Gray, 1999). With respect to the issue of whether MC practices disproportionately disadvantage minorities, while some observers of the health sector suggest that several MC practices have positive effects on minorities’ access to care, a much larger group of researchers suggests MC practices either: (1) reduce the amount of health services that are available to minorities; (2) reduce the ability of minorities to get private health insurance coverage; or (3) make it more difficult for minority providers to obtain or maintain membership in a MCO.2 In this review the focus will be on all three of these possible detrimental outcomes for minorities’ access to care.
CATEGORIZING MANAGED CARE PRACTICES MC practices can be categorized by whether a health care organization uses the practice to coordinate care or to define its boundaries. Drawing on Scott’s (2003) classification of the types of structures organizations use to respond to their environment we further divide the practices used to
48
GREG A. GREENBERG
coordinate care into five types – rules and programs, authority, goal setting, client coordination, and culture – that are examined below. Organizational boundaries are usually one of the most distinguishing features of an organization, but also one of the most difficult to define. Individuals can be characterized as being part of an organization based on such factors as their interests, training, social relations, activities, or behaviors (Scott, 2003). The focus here will be on organizational boundaries defined by who is and who is not an organizational member. More narrowly, only two types of MCO members will be examined in this review – providers who perform the primary care-giving activities (either employees or contractors) and clients who benefit from those activities. Thus, the boundary-setting practices of interest in this review are the techniques by which a MCO recruits and retains its providers and clients or ends the membership of either. Discussion of the typology will consist of two sections, first a section that examines how the literature suggests the five types of MC practices used to coordinate care affect ethnic differences in access to care and next a section that has similar analysis of MC boundary-setting practices. Three briefer sections follow that present: (1) the methodology used to obtain and select those empirical studies that appropriately address the hypotheses presented in the first two sections; (2) an examination of the degree to which the empirical studies verify the hypotheses; and (3) the implications and limitations of this review.
COORDINATION OF CARE Rules and Programs Rules can be defined as agreements about how decisions are to be made or work is to be processed. A program is a set of rules, i.e., a set of contingent procedures to be followed given specific conditions. For example, cybernetic systems are programs in which the same rule is used repeatedly in a closed system but the decision made varies based on existing conditions. Another example are decision-trees in which the resulting outcome that follows the use of the procedures specified by the rule governing existing conditions leads to a new rule, procedures, and set of possible outcomes. All organizations, even those performing very simple tasks, use rules and programs. It has been suggested that the application of evidence-based UR standards and guidelines by increasing consistency across patient groups, predictability, and at least the appearance of objectivity may increase service use by
U.S. Minorities’ Access to Health Care under MC
49
minorities (IOM, 2003; Rice, 2003). Also suggested is that MCO requirements that all members have a provider (who can implement rules) will increase minorities likelihood of having a regular provider and usual source of care (Haas, Phillips, Sonneborn, McCulloch, & Liang, 2002; Hargraves, Cunningham, & Hughes, 2001; Reschovsky, 1999). However, other researchers suggest that rules and programs may be biased against minorities (Chin, 2000; Elder, 2002; Davis, 1997b; La Roche & Turner, 2002; Randall, 1994; Randall et al., 1996; Snowden, 1998) and that when decisions are made that restrict access to care (i.e., rules are applied in particular ways) minorities and their providers may be less willing and able advocates in overturning such decisions (Randall, 1994; Balsa, Seller, MgGuire, & Bloche, 2003; Bloche, 2001; Lowe et al., 2001; Schlesinger, 1987). First, I will describe the most relevant MC programs, i.e., UR and guidelines. I will then present the ways in which these programs are implemented (prospectively, concurrently, and retrospectively), and then lastly, discuss the ways in which researchers suggest that these programs affect ethnic differences in access to care. Programs The most important programs (i.e., sets of rules) in the health sector are UR standards and practice guidelines (hereafter referred to as guidelines). UR standards are sets of rules about the type and amount of services that will be provided for clients given their condition, coverage, medical history, demographic characteristics (usually age and sex), and possibly other factors (Knight, 1998). Utilization management is the employment of these standards so as to monitor the use of services before, during, or after their provision to ensure their necessity, efficacy, and appropriateness (given insurance coverage, illness, and other factors) (Knight, 1998). Utilization management or UR (the terms are generally used interchangeably) is also a method of rationing care. In contrast, guidelines are sets of rules whose perceived purpose is to reduce inappropriate variation in the provision and use of health services. Another difference is that guidelines are more detailed than UR standards; they specify alternative appropriate treatments or clinical protocols, paths, and/or algorithms (i.e., the timing and coordination of treatment) (Fauman, 2002; Knight, 1998; Rice, 2003). Guidelines can result in either more or less services, depending on how current practice patterns differ from what the guidelines suggest and as the name implies they are meant to offer guidance (Rice, 2003; Stensen, 2000). Most MCOs engage in some form of UR and according to the American Association of Health Plans, by 1995 more than
50
GREG A. GREENBERG
85 per cent of MCOs promoted the use of guidelines (Knight, 1998). Since in implementation guidelines may to some degree be mandated and used to ration rather than enhance the quality of care, guidelines and UR standards will be jointly discussed. A wide variety of sources for UR standards and guidelines exist. In 1996 there were already almost 1,800 guidelines released by 75 national groups (Saxton & Leaman, 1998). More widely accepted UR standards and guidelines are based on experts review of the clinical literature and evaluation of relevant clinical data. They may be modified or supplemented by specialty medical societies, local medical associations, consultant firms, and MCO medical advisory boards, that may also develop their own UR standards and guidelines (Bloche, 2001; Knight, 1998). Program Implementation Prospective review refers to denying or modifying the availability of services before they are provided. Two primary types of prospective review exist, preauthorization (or pre-certification) and gatekeeping. In the case of preauthorization an MCO authorizes treatment based on the criteria mentioned above (i.e., patient condition, medical history, etc.). It depends on the MCO, whether the member or provider needs to seek permission for the service, such as obtaining a preadmission certification for a hospital stay or an authorization for elective surgery. MCOs also differ by whether the authorization is granted by the MCO directly or another entity employed by the MCO (such as a UR agency or a group practice) (Kerr et al., 1995; Knight, 1998). A utilization reviewer employed by the MCO (or other entity) will use guidelines or UR standards to assess whether the case meets the criteria for the service and may also specify the scope or intensity of the service (Knight, 1998). Gatekeepers are providers who are compensated by MCOs to make decisions for patients regarding the appropriateness of care that they may not do themselves (such as diagnostic testing, hospitalization, and specialist consultation). They are also expected to provide preventive care and promote healthy behaviors (Saxton & Leaman, 1998). The techniques, such as financial incentives, that health care organizations use to encourage gatekeepers and other providers to adhere to UR standards and guidelines will be discussed below. Concurrent review is the review of ongoing treatment at regular intervals (conducted during a patient’s hospital stay or a series of outpatient visits) to determine whether treatment is still necessary while retrospective review is the evaluation of services that have already been provided. Retrospective
U.S. Minorities’ Access to Health Care under MC
51
review could involve the use of aggregate data to evaluate levels and types of services provided by individual or group practices (often called provider profiling) or to identify under and over-utilization of services for a specific population. At the individual client level retrospective review can lead to the denial of reimbursement for treatment after the fact. For instance, there have been controversies and legal cases with respect to the circumstances in which a MCO must pay for emergency room care (Saxton & Leaman, 1998). Implications for Ethnic Disparities The standardization associated with UR has been criticized on several grounds. First, UR standards and guidelines could pose problems for minorities because they may be based on biased reference points, i.e., data obtained from the treatment of middle class white males, and minorities are more likely to have: (1) illnesses that have gone without medical treatment and that are more severe and require longer treatment due to the lack of health care during childhood and such resources as proper food and housing and (2) medical needs created by being exposed to risk factors for illness or injury, such as homelessness, violence, drug abuse, air and water pollution, racism, etc. (Davis, 1997b; Elder, 2002; Randall, 1994; Randall et al., 1996). Other commentators also suggest that minorities may have different culturally determined preferences for the type and amount of care they seek and that, at least in the case of mental illness, following guidelines may lead practitioners to misjudge culturally specific symptoms resulting in service underutilization (Chin, 2000; La Roche & Turner, 2002; Snowden, 1998). Thus, according to these arguments, the use of UR standards and guidelines without adjustment for client ethnicity may lead to less than appropriate service use for minority clients. Some researchers also suggest that UR may cause ethnic disparities in access to care because minority patients are less willing and effective advocates for themselves. Reasons given for minorities lower willingness and ability to appeal (informally or formally) the decisions of utilization reviewers and gatekeepers, include minorities relative lack of access to resources (i.e., financial, social networks, education, etc.) (Randall, 1994; Bloche, 2001; Schlesinger, 1987), inexperience with the cultural environment of health care organizations (Balsa et al., 2003; Randall, 1994), and historical mistreatment and poor current experiences in health care and other types of organizations that create feelings of distrust and disempowerment (Bloche, 2001). As important as the degree to which minority patients advocate for themselves is the degree to which providers are willing and able to advocate
52
GREG A. GREENBERG
for minority patients. Several researchers have suggested that the willingness and ability of a provider to advocate for minority patients will be reduced to the degree that they have: (1) less clarity about the conditions of minority patients due to communication difficulties; (2) less personal engagement with minority patients; and (3) conscious or unconscious biases and stereotypes regarding minority patients (Balsa et al., 2003; Bloche, 2001; IOM, 2003; Lowe et al., 2001). The locations, such as hospital clinics, where minorities are more likely to receive care may also result in the providers who serve minorities not having the influence, reputation, skills, or relationships with patients that are needed to effectively advocate. Because minorities are less likely to have either a regular provider or a usual source of care (Hargraveset al., 2001; IOM, 2003; Phillips, Mayer, & Aday, 2000) they are more likely to depend on, for example, hospital staff physicians rather than physicians with a regular practice. Physicians in the latter group have more influence as well as more knowledge of and a stronger relationship with the patient (Bloche, 2001; IOM, 2003). Settings where minorities receive care are also likely to have higher staff turnover, resulting in minorities having weaker relationship with their providers and their providers not having strong relationships with other providers or administrators (Bloche, 2001). Rice (2003) also suggests that more specialized and trained doctors who are more influential have a greater probability of pricing out of the type of insurance that minorities are more likely to have, such as Medicaid. To summarize: Hypothesis 1. Utilization review (UR) will lead to a greater failure of minorities to receive services compared with whites due to: (1) biases in the way guidelines or UR Standards are developed and (2) the lower willingness and ability of minority patients and their providers to contest UR decisions. As discussed earlier, several researchers have instead hypothesized that UR increases minorities’ access to care because: Hypothesis 2. The use of gatekeepers increases the relative likelihood of minorities having a regular provider and a usual source of care (Haas et al., 2002; Hargraves et al., 2001; Reschovsky, 1999). Hypothesis 3. The greater use of rules and programs associated with UR increases consistency across patient groups (by reducing provider and other decision-makers discretion) resulting in reduced ethnic differences in access to care (IOM, 2003; Rice, 2003).
U.S. Minorities’ Access to Health Care under MC
53
The arguments of the next section contrast with Hypothesis 3 in that they focus on how MCOs reduce rather than increase the relevance or use of rules and programs and thus increase the discretion of providers and other decision-makers. Hypothesis 3 and the arguments of the next section do have in common though, the underlying optimistic assumption that appropriately designed and implemented rules and programs by reducing the discretion of providers and other decision-makers can reduce ethnic differences in access to health services. Thus, Hypothesis 3 and the arguments below offers an interesting contrast with most of the arguments of this section, in which rather than a lack of rules, the existence of rules is viewed as problematic (due to their inherent biases and the lower ability of minorities to contest them).
Authority and Discretion Hierarchy allows for individuals in positions of authority to decide when rules apply. For example, preauthorization might be deemed necessary for all referrals to a cardiac surgeon but not for obstetricians. These individuals also decide when rules cannot easily be applied, i.e., when the unexpected or irregular occurs and when ambiguity and uncertainty exist, which are common occurrences in the delivery of medical services. Decision-makers also interpret how rules apply, develop new rules, and refine rules. Most importantly, when rules do not exist, they make decisions, i.e., they have discretionary authority. I will first discuss how this discretion in the delivery of health services may contribute to ethnic disparities in access to health services. Next, I will discuss how MCOs unintentionally and intentionally implement rules in ways that increase space for discretion. Discretion and Ethnic Disparities In any type of medical system providers and other decision-makers have a large amount of discretion because of a variety of factors; such as the lack of resources for expensive diagnostic tests, ambiguity about the diagnostic implications of symptoms, lack of agreement on how to view particular client outcomes, and uncertainty about the efficacy of many diagnostic and therapeutic alternatives. These factors prevent the codification of treatment rules for many conditions as well as their implementation. Additionally, because of such factors, if rules are used, they are easily challenged and hard to defend (Balsa et al., 2003; Bloche, 2001). In situations in which rules cannot be developed, applied, or defended, the combination of uncertainty and complexity with decisions-makers cognitive
54
GREG A. GREENBERG
limitations and the costs of obtaining information results in providers, utilization reviewers, and other decision-makers using ‘‘heuristics.’’ Heuristics are decision rules that simplify decisions or act as mental shortcuts. They may work much of the time but are not rigorously researched and are usually not made explicit (Balsa et al., 2003; Bloche, 2001; Scott, 2003). When heuristics are applied to ethnic groups, they can become stereotypes, i.e., unfavorable generalizations about minorities that result in their disparate treatment. An example would be the assumption that a particular minority group is less likely to comply with more complex and long-term treatment regimens and the assumption resulting in that group not getting the most up-to-date treatment. Some researchers have added that provider propensity to use stereotypes that favor white patients is increased by providers greater difficulty in obtaining information from minority patients (Balsa et al., 2003; Bloche, 2001; Rice, 2003). Such stereotypes may be unrelated to provider bias against a particular ethnic group and can operate consciously or unconsciously (Balsa et al., 2003; Bloche, 2001). Rule Implementation and Discretion Although not an argument that MC systems differently disadvantage minorities in comparison to other systems, several researchers have concluded that ample room exists in the more rule intensive MC systems for biased decision-making and stereotypes (Balsa et al., 2003; Bloche, 2001; IOM, 2003; Randall, 1994; Randall et al., 1996). A reason is that MCOs, like most other organizations, imperfectly implement rules. Those who are implementing rules often differ from those who developed the rules and the degree and type of interaction by these two groups varies by MCO. Additionally, utilization reviewers may vary a lot in their experience and training as well as their emphasis; some reviewers may focus on reducing costs and only approve care for the sickest patients while other reviewers may be more ethical and open to compromise (Fauman, 2002). This imperfect rule implementation allows for the suggestion that much room remains for discretion, and thus the use of stereotypes and biased decisionmaking in MC systems. More important than imperfect rule implementation is how MCOs purposively implement rules in ways that increase provider discretion, and thus potentially, the role of stereotypes and biased decision-making. For example, MCOs often do not release information regarding previous but similar UR decisions and commonly treat guidelines and UR standards as trade secrets. The reason MCOs withhold such information is that MCO managers worry that it would help providers to ‘‘game the system’’ by
U.S. Minorities’ Access to Health Care under MC
55
allowing them to alter patients’ medical conditions so as to conform to UR standards (Knight, 1998). MCOs withholding of such information hinders the standardization of medical rules (or the development of sound medical practice) slowing the reduction of provider discretion (Bloche, 2001; IOM, 2003; Knight, 1998). To summarize: Hypothesis 4. MCOs failure to disclose rules and prior rule implementation decisions increases provider discretion (or slows its reduction) resulting in their greater use of stereotypes and biased decision-making, hindering minorities’ access to care. Several researchers (Balsa et al., 2003; Bloche, 2001; IOM, 2003; Randall et al., 1996) have also suggested: Hypothesis 5. The use of additional decision-makers (who are often not familiar with the patient) due to UR increases the role of stereotypes in allocating services, disadvantaging minorities. For instance, utilization reviewers, who have less familiarity with clients, may also use stereotypes to guide their treatment and reimbursement decision-making (or make decisions based on assumptions regarding who is most likely to appeal a UR ruling or sue) (Balsa et al., 2003; Bloche, 2001; IOM, 2003; Randall et al., 1996). A related argument is that the growing use of providers to ration care (i.e., as gatekeepers) in a weak rule environment has possibly allowed for a greater role of stereotypes in service provision decisions (IOM, 2003; Balsa et al., 2003). Decision-Making Constraints and Discretion It has also been suggested: Hypothesis 6. MC practices implemented to constrain and influence provider decision-making reduce the time providers have to collect and analyze patient information increasing their dependence on stereotypes, thereby hindering minorities’ access to care (Balsa et al., 2003; Bloche, 2001; Randall et al., 1996). Such MC practices will now be discussed.
Goal Setting Rather than specifying and enforcing how tasks are to be accomplished or vetting their decisions the MC techniques discussed here encourage
56
GREG A. GREENBERG
providers to accomplish desired outcomes. Such techniques are referred to as ‘‘targeting’’ or ‘‘goal setting’’ (Scott, 2003). I will review two types of commonly used goal-setting techniques, performance monitoring and financial risk shifting, and then discuss the suggested implications of these techniques for ethnic disparities in access to health services. Performance Monitoring A variety of measures and techniques may be used to evaluate and inform providers of their performance. The most important variant of provider performance monitoring is ‘‘provider profiling,’’ a form of retrospective UR. Provider profiling involves the quantification and then the ranking (or comparison to set standards) of providers (or group practices) consumption of resources for the treatment of particular illnesses (Elder, 2002; Knight, 1998; Randall, 1994). Providers are informed of these ranking to encourage changes in their practice style so that they meet particular performance targets. As with other types of performance indicators these rankings may be tied to a provider’s compensation and their employment or contract status (Anonymous, 1995; Elder, 2002; Knight, 1998). Stoddard, Reed, and Hadley (2003) found from analysis of a nationally representative physician survey that approximately 67 per cent of physicians are subject to profiling without financial incentives and 14 per cent with financial incentives. Financial Risk Shifting Financial risk shifting is a goal-setting technique in which various forms of compensation, such as interest in an MCO, bonuses, and capitation, are used to expose providers to some risk. Rather than just specifying goals and informing providers of whether they are achieving them, such pay incentive plans can be viewed as a method by which managers and owners change the ‘‘cognitive map’’ of providers, such that their interests are more closely aligned with managers and owners. In other words, if successful, these compensation arrangements may result in providers internalizing the goals of managers and owners, a much stronger form of goal setting. This is not to suggest that performance monitoring is ineffective. Additionally, some overlap exists between performance-based compensation (i.e., tied to providers’ profiles or other indicators) and risk-based compensation. For instance, bonuses are arrangements by which providers only receive additional compensation if they reach performance targets, such as providing enough services or, in the case of gatekeepers, meeting individual or group utilization goals. The degree to which bonuses put providers at risk depends on how they are implemented.
U.S. Minorities’ Access to Health Care under MC
57
Risk sharing through capitation or other techniques can occur at more than one administrative level in a single MCO (for instance, the health insurance plan,3 provider group, or individual provider level). Under capitation individual providers or organizations are paid a fixed amount for a patient or set of defined services for a specific period. Whoever is capitated is at risk because they only profit if client needs are met at a cost below the agreed upon rate and will suffer losses if service costs exceed the payment rate (Randall, 1994; Snowden, 1998). Payment rates are usually adjusted for age and sex, but adjustment for other factors is uncommon (Knight, 1998; Kongstvedt, 2001). An individual provider may be responsible not only for their service costs but also for costs arising from referrals to hospitals and specialists (Rice, 2003). Another payment mechanism MCOs use to share risks with providers is withholding or risk pools. When an MCO uses withholding or risk pools a portion of provider reimbursements (usually 5–20 per cent) for patient services is placed in a pool. The pool is used to reimburse any claims that exceeded projections for a defined group of patients. The remaining funds after a specified period are distributed to the providers participating in the pool. The distribution of these pools depends in part or fully on the number of specialist referrals and diagnostic tests made (Rice, 2003; Randall, 1994; Saxton & Leaman, 1998). The widespread use of these compensation arrangements is shown by a 1999 national survey of MCOs conducted by Mathematica Policy Research Institute that found to pay primary care physicians approximately 42 per cent of MCOs used withholds or bonuses, 61 per cent used capitation, and 29 per cent used a combination of these two methods (equivalent percentage for specialists were 29 per cent, 13 per cent, and 6 per cent, respectively) (Rice, 2003). This capitation may mostly be at the group level since a Medical Group Management Association survey found only 3 per cent of capitated medical groups sub-capitate their individual providers (Stensen, 2000). Implications for Ethnic Disparities To the degree providers and health care organizations assume that minorities use more services and are more costly to treat and/or evidence shows this ethnic difference exists, goal-setting techniques create incentives to avoid minority clients. This is because if providers associate minority status with being an intensive service user they will wish to avoid minority clients so as not to be put at a financial disadvantage (or threatened with losses) (IOM, 2003; Rice, 2003; Snowden, 1998; Balsa et al., 2003; Tai-Seale, Freund, & Lo Sasso, 2001) and so that their profile does not move from
58
GREG A. GREENBERG
‘‘average’’ to ‘‘aberrant’’ (Anonymous, 1995; Davies, Washington, & Bindman, 2002; Elder, 2002; Perez, 2003). Risk adjustment may partially address this problem; however, little indication exists that risk adjustment occurs with regard to financial incentives (Knight, 1998; Kongstvedt, 2001; Simon, White, James, & Hernandez, forthcoming). If a MCO does risk adjust its provider profiles it is unlikely that they are adequately adjusted for service use risk factors associated with minority status because few MCOs have satisfactory risk adjustment measures, enough observations per provider, or enough providers to estimate reliable measures of appropriate service use (Hofer et al., 1999; Kerr et al., 1995). To summarize: Hypothesis 7. To the degree minorities are shown or assumed to be high need users providers will attempt to avoid them so as not to have a negative profile. Hypothesis 8. To the degree minorities are identified as especially costly to serve risk-based compensation will create incentives for providers and health care organizations to make fewer services available to minorities or to avoid these clients. An additional argument has been made with respect to minority providers: Hypothesis 9. Due to their higher minority patient load, profiling will lead to minority providers being profiled as high users, thus increasing the likelihood that they will have contracts terminated or denied (Elder, 2002; Mackenzie, Taylor, & Lavizzo-Mourey, 1999; Perez, 2003).
Client Coordination Although MCOs have implemented more complex practices to increase client involvement in the coordination of care, most common and almost universally used are cost-sharing practices, such as co-payments and deductibles (Glied, 2000). Cost sharing involves risk shifting to clients and requires clients to be more calculative. Additionally, the more these practices increase the personal cost of care the less health services an individual will use (Wallace, Enriquesz-Haas, & Markides, 1998). Such cost sharing may not just influence whether but when individuals receive care. Stated differently, an error term is being introduced for many low-income individuals in which care is obtained not when an individual needs it but when they can afford it. Rice (2003) has suggested that cost-sharing
U.S. Minorities’ Access to Health Care under MC
59
practices may disproportionately disadvantage minorities because of their average lower income and greater service needs. Cost sharing also discourages use of more price sensitive health services, such as preventive care, which may be particularly detrimental to minorities to the extent that they have worse health status and use preventive services less than whites (Davis, 1997b; Rice, 2003). To summarize: Hypothesis 10. Cost sharing will disproportionately hinder minorities’ access to care, particularly of price sensitive health care. Culture MCO managers may attempt to create a particular ethos or culture to influence provider behavior. For example, while informing and educating providers of organizational rules and goals MCO managers may try to influence the norms and values of providers through a variety of educational and informational techniques (e.g., introductory orientations, educational programs, seminars, telephone calls, newsletters and other printed materials, videos, websites, and site visits). Any setting in which providers work together and contract with (or are employed) by a few or one MCO over extended periods, such as group and staff model MCOs, offer even better opportunities for socialization. In these settings ‘‘culture’’ might also offer important support (or even substitute) for one or more of the MC practices discussed above. For example, in such settings ‘‘informal’’ guidelines can be used to establish or modify individual and group practice patterns or styles (Luft, 1999; Weiner & Lissovoy, 1993; White, 1999). No literature exists on how MCO managers attempt to influence the process of care in this manner affects ethnic differences in access to care; thus, no hypotheses were specified. However, to the degree that the outcomes of efforts by MCO managers to influence practice culture are similar to the forms of coordination already discussed, the ‘‘management of practice culture’’ should be discussed and examined. The potential consequences should not be underestimated given the importance of culture as a managerial tool in many technology and service intensive industries.
BOUNDARY SETTING The ability of an organization to differentiate members from nonmembers is a key way it achieves goals that for MCOs can be assumed to be maximizing profits, conserving resources, or effectively using resources. Our interest is in
60
GREG A. GREENBERG
how and whether MC boundary-setting practices used to achieve these goals may differentially affect minorities’ access to health care. As stated above, the focus here is on organizational boundaries defined by membership, i.e., providers who perform MCO activities (employees or contractors) and clients who benefit from MCO activities.
Clients MCOs are often accused of engaging in what is known as ‘‘cream-skimming’’ or risk segmentation, that is trying to select the healthiest patients and encouraging the sickest patients to go elsewhere, in order to maximize profits (Perez, 2003; Wallace et al., 1998; Currie & Fahr, 2005). Individuals enrolled in Medicare MCOs compared with those in traditional Medicare have in fact been found to have a lower prevalence of chronic conditions, disabilities, lower mortality, and a better self-reported health status (Knight, 1998; Wallace et al., 1998; Davis, 1997a). Several proxies for health status might be used to engage in this practice, such as age, income, and residential location. It has been hypothesized that MCOs will also wish to exclude minority patients (Davies et al., 2002; Schlesinger, 1987; Wallace et al., 1998), by such techniques as: designing benefit packages to select low risks (Kronick, 1999), increasing transaction costs for some applicants (Currie & Fahr, 2005), and failing to advertise in minority communities (Balsa et al., 2003; Perez, 2003; Rosenbaum, Serrano, Magar, & Stern, 1997). The last practice may be associated with what is known as ‘‘redlining.’’ Redlining is most commonly associated with housing. In the past it involved the actual practice of drawing a red line around poor and predominantly minority neighborhoods and then refusing to sell mortgages or home insurance in those neighborhoods (Perez, 2003). Several researchers suggest that because MCOs wish to avoid higher risk clients they may not offer services to areas with more minority clients, offer those areas fewer services at higher prices (Perez, 2003; Rosenbaum et al., 1997; Smith, 1999), or not locate their provider networks in particular communities (Anonymous, 1995; Rosenbaum et al., 1997; Smith, 1999). To summarize: Hypothesis 11. MCOs will use ethnicity as a proxy for the probable health care costs of a patient and use various boundary-setting techniques to avoid serving minority clients. Contracting with only certain employers might result in another form of redlining that could also differentially affect minorities, what might be called
U.S. Minorities’ Access to Health Care under MC
61
‘‘industrial redlining’’ (Zellers, McLauphlin, & Frick, 1992). Insurance firms often organize their marketing and set their premium levels by employer characteristics, such as size, age, type of industry, and employees’ age and gender. Premium rating in this manner is referred to as community rating by class. Over time premium rates may be further adjusted by claims experience (Knight, 1998; Blumberg & Nichols, 1996; Sturm, 2001). There has been reported industrial redlining of some industries, particularly those characterized as high risk and hazardous and the existence of higher premium rates for small businesses is well known (Frankford, 1999; Zellers et al., 1992). Although industrial redlining may result in employers or industries with more minority employees being underserved or being subject to higher premiums the literature has not addressed this issue; thus, no hypothesis was specified. Employees and Contractors A key difference between MC and fee-for-service (FFS) systems is that while indemnity insurers generally reimburse all providers, MCOs use a variety of techniques to decide who will provide care, allowing MCO managers significant influence over service quality and use. Some MCOs have increased this ability by including ‘without cause’ termination provisions in their provider contracts that enable them to end provider contracts without a particular reason (Knight, 1998). One technique by which MCOs decide who will provide care is by examining providers credentials. Credentialing may be based on a provider’s profile (Anonymous, 1995; Elder, 2002), board certification, privileges at a MCO network hospital, provider group size, and practice financial stability. Such credentialing may hinder minority providers from joining and maintaining membership in MCO provider panels since they are less likely to be board certified and also less likely to have privileges at major hospitals in MCO networks and instead work at smaller community hospitals (Mackenzie et al., 1999; Rosenbaum et al., 1997; Simon et al., forthcoming). They are also more likely to be in individual or small group practices rather than larger multi-disciplinary groups (Elder, 2002; Mackenzie et al., 1999). Their practices may also have lower levels of financial stability, due to their small size and the type of patients they serve (Simon et al., forthcoming). To summarize: Hypothesis 12. Criteria that determine which providers obtain MCO panel membership are more likely to exclude minority providers.
62
GREG A. GREENBERG
While credentialing practices that exclude minorities may not be due to the intentional actions by MCO managers, several researchers (Anonymous, 1995; Frankford, 1999; Mackenzie et al., 1999; Simon et al., forthcoming) have specifically suggested that: Hypothesis 13. MCOs will intentionally avoid employing or contracting with minority physicians so as to avoid minority patients. Minority providers who have difficulties getting MCO contracts will have a hard time surviving in areas with high levels of MCO penetration. Not only will it be difficult for them to attract new patients, they may also lose their better insured and healthier patients.
METHODS A systematic literature search was undertaken for articles on minority access to health care and MC in the United States. Using the terms ‘‘managed care and (race or ethnic)’’ the following databases were searched for articles published before March 2006: Proquest, JSTOR, EBSCO, Medline, Wilson Social Science Index, Lexis Nexis Health Journal section, and Econlit. All relevant literature cited in the obtained articles were also reviewed. To examine the validity of the hypotheses, this review only included quantitative empirical studies published in a peer-reviewed journal or that were an accepted dissertation. The study also had to allow for an evaluation of how the implementation of a specific MC practice affects ethnic differences in access. Therefore, to be included a study needed to: (1) compare FFS and MC systems in a way that allowed for a focus on a particular MC practice; (2) compare MCOs that did and did not employ a particular MC practice; or (3) measure a MC practice so that the outcome could not be attributed to other system characteristics (such as measuring preauthorization with the authorization decision and not with a service use measure). Results of these studies are presented below while Table 1 summarizes their primary characteristics.
EMPIRICALLY EVALUATING THE VALIDITY OF THE THEORY Empirical literature is available on few of the above hypotheses, those specified with respect to rules and programs as well as MCO
Source and Hypotheses
Published Empirical Studies on Managed Care and Minority Access to Services. Study Question
Methodology
Findings
63
Cross-sectional survey of a probability sample of Physician minority status and board certification are not related to MCO contract denial or office based primary care physicians in 13 termination. Physicians not in a solo practice large urban counties in California. Used an are less likely to have their contracts denied or analytic sample of 231 white, 56 black, 129 terminated (odds ratio=.3, 95% CI .2–.6). Asian, and 104 Latino physicians. Used Physicians on MCO panels have less minority multivariate logistic and linear regression clients. Board-certified physicians are more models in which all physicians practice and likely to have one or more MCO contracts demographic characteristics of interest are (odds ratio=2.6, 95% CI 1.2–5.6) included. Analysis was weighted to account for oversampling of nonwhite physicians and differences in response rates by county Survey data obtained from the Maryland Study Physician minority status is not related to MCO Elder (2002) The degree to which Hypotheses 12 and 13 contract denial or termination. A greater on Physicians Experience with Managed Care physician practice and percentage of Hispanic (b=.0049; p=.019) or Survey sample. Used an analytic sample of demographic black patients (b=.0036; p=.001 ) is 149 Hispanic, 413 white, 294 black, and 338 characteristics are significantly associated with MCO contract Asian physicians with managed care associated with MCO denial; Asian patients with contract experience active in Maryland. Physicians in contract denial or termination (b=.0042; p=.003) and Hispanic institutional and public health settings were termination patients with lower likelihood of contract not included or those who described their termination (b= .0065; p=.028). Board primary roles as teaching, research, and/or certification is significantly associated with a administrative. Used multivariate Probit lower likelihood of denial (b= .189; p=.003) models that included measures of provider but not termination characteristics, patient demographics, and the HMO competitive environment With higher MCO penetration (20% increase) State-level enrollment data for Medicaid Currie and Fahr The effects of MC the probability of Medicaid enrollment among managed care from 50 states (the National (2005) penetration of the Hypothesis 11 black children significantly declines (a 1.8% Health Insurance Survey supplements for Medicaid market on decline) (b= .088; p=.037) while no 1989, 1992, 1993, and 1994). The analytic minority children significant changes among Hispanics and sample consisted of 75,000 children. Used enrollment in Medicaid whites occur multivariate logistic regression models that
Bindman et al. (1998) The degree to which Hypotheses 12 and 13 physician practice and demographic characteristics are associated with MCO contract denial or termination
U.S. Minorities’ Access to Health Care under MC
Table 1.
64
Table 1. (Continued ) Source and Hypotheses
Haas et al. (2002) Hypothesis 2
Guwani et al. (2004–2005) Hypothesis 1
Methodology
Findings
included measures of the Medicaid income cutoff point to qualify for benefits, state unemployment rates, the growth of private MCOs, and demographic characteristics Ethnic differences in use Observational cohort using the nationally Ethnic differences that disadvantaged minorities representative 1996 Medical Expenditures of preventive health in whether an individual had a usual source of Panel Survey. The analytic sample consisted services in a MC plan care decrease under MC of 1,668 Hispanics, 309 Asians, 1,455 blacks, relative to a FFS plan and 8,657 whites. Used simple frequency tables Similar ethnic differences exist with respect to Uses data from a nationally representative Ethnic differences in use of a CDC guideline recommended probability sample of HIV-infected adults access in MC and FFS treatment for HIV/AIDs in health plans that plans receiving care in the United States obtained used gatekeepers and those that did not from the HIV Cost and Services Utilization Study. The analytic sample contained 388 whites and 488 blacks. Used multivariate regression models with sociodemographic and health status controls There exists no statistical association between RAND appropriateness and necessity criteria Ethnic differences in whether an ABS recommendation occurs and used to identify 680 Hispanic, 314 Black, and gatekeeper physician patient ethnicity 267 white post-angiography patients who recommendations for would benefit from ABS that were treated at 8 artery bypass surgery New York hospitals. Clinical, patient (ABS) and in obtaining telephone/mail survey, and Cardiac Surgery this surgery reporting system data analyzed with stepwise logistic regressions that included clinical and sociodemographic, and insurance status measures
GREG A. GREENBERG
Hannan et al. (1999) Hypothesis 1
Study Question
Similar ethnic differences exist with respect to having a usual source of care and a regular provider and in use of specialists in health plans that used gatekeepers and those that did not
Ethnic differences that disadvantage minorities in whether an individual had a usual source of care and a regular provider decrease under MC
Blacks are more likely to be denied authorization for care (odds ratio=1.52, 95% CI 1.18–1.94)
Minority status is not related to MCO contract denial or termination with the exception of Asian physicians whose contracts are terminated more often (b= .39; p=.04). Having a solo practice is associated with a greater difficulty of obtaining MCO contracts (b= 1.14; p=.00). having a contract terminated (b= .7; p=.00). and having fewer MCO clients (X2=141; p=.001)
65
Ethnic differences in Used data from the nationally representative access in MC and FFS Community Tracking Study Household plans Survey. The analytic sample consisted of 4,188 blacks, 3,379 Hispanics and 33,737 white nonelderly with insurance. Used multivariate regression models with sociodemographic and health status controls Leigh et al. (1999) Plan differences (MC vs Used data from the 1995–1996 KaiserCommonwealth Low-Income Survey. The FFS) in access and Hypothesis 2 analytic sample consisted of 925 blacks, 641 satisfaction among Hispanics, and 2,346 whites living in Florida, minorities and whites Tennessee, and Texas. Used frequency tables with results weighted to reflect population characteristics A retrospective cohort design using Lowe et al. (2001) Whether ethnic Hypothesis 1 computerized clinical data on 12,442 black differences exist in and 3,136 white patients at a single site. emergency department Multiple logistic regression in which triage service authorization score, age, gender, insurance type, and arrival time at emergency department were controlled for Used data from the Harris and Associates Mackenzie et al. The degree to which National Physicians Survey. The sample (1999) physician practice and excluded physicians who were government demographic Hypotheses 12 and 13 employees, hospital-based residents, or who characteristics are provided direct patient care part time. Some associated with barriers subgroups were oversampled to ensure in acquiring and reliable analysis. The analytic sample maintaining MCO consisted of 680 white, 68 Hispanic, 86 black, contracts and 126 Asian primary care and obstetrics/ gynecology physicians. Dependent measures were composite scales created using Pearson correlation and factor analysis. Used multivariate regressions in which weighting done to compensate for oversampling
U.S. Minorities’ Access to Health Care under MC
Hargraves et al. (2001) Hypotheses 1 and 2
66
Table 1. (Continued ) Source and Hypotheses
Study Question
Phillips et al. (2000) Hypothesis 2
Plan differences (MC vs FFS) in access among minorities and whites
Reschovsky (1999) Hypothesis 2
Plan differences (MC vs FFS) in access among minorities and whites
Methodology
Findings
Used the nationally representative 1996 Medical Ethnic differences that disadvantage minorities Expenditures Panel Survey. The analytic in whether an individual had a usual source of sample consisted of 1,866 Hispanics, 1,448 care decrease under MC blacks, 355 Asians, and 10,227 whites. Used a simple frequency table Used the nationally representative Community In comparison to FFS plans there exists greater Tracking Study Household Survey. The disadvantageous differences in MCOs analytic sample consisted of 35,874 between whites and Hispanics (but not blacks) individuals with private insurance (82.2% with regard to having a usual source of care white, 9.4% black, and 8.4% Hispanic). The paper provided percentage point differences between MC and non-MCO enrollees by ethnic group in having a usual source of care
Only study results relevant to this review are presented. Study did not explicitly address the question of whether ethnic differences in having a usual source of care would diminish or persist in a
GREG A. GREENBERG
MC system, lacked specification with regard to whether gatekeepers were used, and lacked a statistical test of this issue (and thus no controls were used).
U.S. Minorities’ Access to Health Care under MC
67
boundary-setting practices. As will be discussed below, with regard to the first topical area, the results were mixed, while for the second there were relatively consistent findings showing that MCO boundary-setting practices negatively affect minorities’ access to health services, albeit indirectly.
Coordination of Care Of the four studies that addressed the impact of UR and guidelines as specified by the first hypothesis, three studies found no impact and one found significant effects. Specifically, one study of gatekeeper provider recommendations for artery bypass surgery found no association between whether a recommendation occurred and patient ethnicity (Hannan et al., 1999) (see Table 1). Two other studies that compared health insurance plans with and without gatekeepers found that gatekeeping arrangements did not have a sizable effect on ethnic differences in the use of specialists (Hargraves et al., 2001) or the use of a CDC guideline recommended treatment for HIV/AIDs (Guwani & Weech-Maldonado, 2004–2005). The fourth study addressed another type of prospective UR, preauthorization, and found that blacks were 1.5 times more likely to be denied authorization for emergency department visits by MC utilization reviewers after adjusting for confounders (Lowe et al., 2001). Empirical studies were not found that addressed concurrent or retrospective UR. With respect to the second hypothesis, several studies found ethnic differences (that disadvantage minorities) in whether an individual has a usual source of care (Haas et al., 2002; Phillips et al., 2000; Leigh, Lillie-Blanton, Martinez, & Collins, 1999) and a regular provider (Leigh et al., 1999) are lower in MCOs, while one study found increased differences in MCOs between whites and Hispanics (but not blacks) with regard to having a usual source of care (Reschovsky, 1999). However, these studies have the following weaknesses: (1) they do not explicitly address the question of whether ethnic differences would ‘‘diminish or persist’’ in a MC system; (2) they lack specification regarding whether gatekeepers were used; and (3) do not use a statistical test (and thus do not use controls). A study that did not have these weaknesses found similar ethnic differences in having a usual source of care and a regular provider in health insurance plans that used gatekeepers and those that did not (Hargraves et al., 2001). To summarize, a large literature exists that contains many logical arguments for why programs in the form of UR should prove detrimental to minorities’ access to care. However, a limited number of relevant studies of
68
GREG A. GREENBERG
UR exist and those studies have mixed findings. Additionally, most of the studies that addressed a potentially positive aspect of UR (i.e., reduced ethnic differences in having a usual source of care and regular provider due to the use of gatekeepers) had several methodological weaknesses. Furthermore, there are no studies that address the third hypothesis (which emphasized the potentially positive outcome of UR increasing the standardization of care across patient groups). Thus, no firm conclusions could be reached regarding the impact of UR and more generally the increased use of rules and programs on ethnic differences in access to health services. Although the large literature on how other MC practices used to coordinate care (i.e., goal setting, client coordination, etc.) contribute to ethnic inequalities in health care access has logical coherence and face validity, the implications of this literature as spelled out in Hypothesis 4 through 10 have not been empirically examined. Therefore, because of the mixed findings with respect to rules and programs and a lack of empirical literature on the other types of MC practices that are used to coordinate care no conclusions could be reached with respect to MC practices used to coordinate care.
Boundary Setting Although few empirical studies exist and only one focused on clients, they generally show that MC boundary-setting practices hinder minorities’ access to care. The one available study (Currie & Fahr, 2005) that addressed boundary-setting practices that directly affect clients (i.e., Hypothesis 11) found that with higher MCO penetration the probability of Medicaid coverage among black children significantly declined while no significant changes occurred among Hispanic and white children. These results are consistent with Medicaid MCOs using ethnicity to engage in creamskimming with respect to blacks (but not Hispanics). Because this is only one study further confirmatory empirical research is needed to investigate whether MCOs are associated with increased cream-skimming along ethnic lines. Additionally, to the degree cream-skimming exists research is needed to investigate the manner by which it occurs. Regarding contractors, the findings of the existing empirical studies generally do not suggest that MCOs purposefully discriminate against minority physicians so as to avoid minority clients (i.e., Hypothesis 13). Specifically, several studies that used physician survey data found that the minority status of a physician is not associated with having an MCO
U.S. Minorities’ Access to Health Care under MC
69
contract denied or terminated when controlling for other factors (Bindman, Grumbach, Vranizan, Jaffe, & Osmond, 1998; Elder, 2002; Mackenzie et al., 1999). Asian physicians are a possible exception; their contracts were found by one study (Mackenzie et al., 1999) to be terminated more often (but not denied). However, this result was not confirmed by the two other studies. The empirical evidence provided more support for the suggestion that MC contracting practices indirectly disadvantage minority clients. Specifically, one study found that even when controlling for other factors providers with a greater percentage of minority patients are more likely to have had an MCO contract denied (Elder, 2002) and another study found that physicians on MCO panels have fewer minority clients (Bindman et al., 1998). Thus, although the studies do not support Hypothesis 13, evidence was found for another implication: MCOs contracting practices reduce the number of minority clients. MC contracting practices were also found to disadvantage minority providers because they are less likely to meet the criteria by which MCOs determine with whom they contract (i.e., Hypothesis 12). Specifically, several studies found a significant association between practice size (Bindman et al., 1998; Mackenzie et al., 1999) or board certification (Elder 2002) and either contract denial or termination. Additionally, although two studies found no significant relationship between board certification and contract denial or termination one found that board certified physicians saw greater numbers of MC patients (Mackenzie et al., 1999) and the other found that board certified physicians were much more likely to have a MC contract (Bindman et al., 1998). To conclude, the literature suggests that minorities may be directly disadvantaged because of MCO efforts to engage in cream-skimming or that they may be indirectly disadvantaged by MCO efforts to avoid minority physicians or due to physician credentialing. The findings of one study that examined the association of MC penetration with variation in Medicaid insurance levels across different ethnic groups suggest that MC boundarysetting practices directly disadvantage black patients (although not Hispanics). Additionally, empirical studies on MC contracting suggest that although MCOs do not purposefully avoid minority physicians, credentialing does disadvantage minority physicians, and physicians with more minority patients are at a disadvantage. Thus, although more research is needed, existing studies do have relatively consistent findings regarding the negative (but mostly indirect) effects of MCO boundary-setting practices on minority’s access to care.
70
GREG A. GREENBERG
DISCUSSION AND CONCLUSION This is the first paper to systematically review the literature and empirical evidence on the relationship between MC practices and ethnic differences in access to health services. The distinction was made here between MC practices used to define the boundaries of heath care organizations and those used to coordinate care. The later were further subdivided using Scott’s (2003) classification of the basic mechanisms by which organizations coordinate and control work. This typology facilitated the systematic discussion of MC practices and the application of concepts derived from organizational theory. It also allowed attention to be focused on the principal underlying each type of MC practice and on a basic disagreement in the literature. Specifically, it was shown that MC practices either: (1) stipulate and enforce how organizational goals are to be achieved (i.e., how work is to be done); (2) get providers and clients to internalize organizational goals or how they are achieved; or (3) screen out providers and clients who hinder the achievement of organizational goals. In the abstract this ‘‘bureaucratization’’ of health services delivery would appear to be neutral with respect to its affect on ethnic differences in access to health services. In fact, a significant number of writers discussed here make one of the following stronger arguments: (1) because MC practices standardize care and increase objectivity they reduce ethnic differences in access to health services or (2) the lack of effective bureaucratization contributes to ethnic inequalities (see Authority and Discretion subsection). However, shown here is how most writers point to many specific contextual factors, internal and external to the U.S. health care system that might cause MC practices to be implemented in ways that contribute to ethnic inequalities. Several weaknesses of the empirical literature prevent us from resolving this conflict in the literature and thus also knowing whether or not MC practices contribute to ethnic differences in access to care. First, relatively few empirical studies exist given the widespread use of MC practices as well as the salience and emphasis given to the issue of ethnic differences in health care access by policy makers, providers, and care recipients. Second, a fairly high proportion of the empirical studies were not reviewed because they do not compare MC and FFS systems or if they do, they do not adequately specify the characteristics of the systems being compared. In the former studies (i.e., case studies) it is impossible to determine the degree to which MC practices are responsible for any discovered ethnic inequalities, rather than other aspects of the U.S. health care system (such as hospital systems
U.S. Minorities’ Access to Health Care under MC
71
or practice groups) that exist in both FFS and MC systems. Ethnic differences in access documented in the latter studies cannot be attributed to a particular MC practice. In this regard it is surprising that only two studies (Guwani & Weech-Maldonado, 2004–2005; Hargraves et al., 2001) exist that use the promising approach of comparing health insurance plans based on whether they use a particular practice. Research in this area would also benefit from more ethnographic and comparative studies of different institutional settings. Such studies, few of which exist, would facilitate a deeper understanding of the processes associated with MC practices and the relevance of various contextual factors. Those empirical studies that were found to have an appropriate methodology only addressed four of the hypotheses specified above and the results lacked consistency for two of these hypotheses. Studies of whether use of a gatekeeper is associated with reductions in ethnic differences in having a usual source of care or regular provider were not consistent. However, the methodologically strongest papers found ethnic differences are similar in health insurance plans with and without gatekeepers (Guwani & Weech-Maldonado, 2004–2005; Hargraves et al., 2001). Also lacking consistency are the results of studies that address prospective review: three studies found that gatekeeping did not change ethnic differences in access to care and another found that preauthorization is associated with large ethnic differences. Studies of MC boundary-setting practices were more consistent, indicating mostly indirect but negative effects on minorities’ access to care. A potential criticism of this review is that it did not address several ways in which MC practices could be implemented that might advantage minorities. For example, provider performance monitoring may also occur with respect to such aspects of service delivery as patient satisfaction, wait time for appointments, and the use of preventive procedures, such as mammograms and immunization (Knight, 1998). One could also imagine financial payments tied to health promotion and disease screening. Additionally, creating financial incentives related to patient satisfaction may encourage providers to overcome communication barriers. Although MC practices can be implemented in ways that may be more beneficial to minorities, the focus here was on how MC practices appear to be most commonly implemented. It should also be kept in mind that regarding incentives tied to preventive care, providers may select who receives such care based on who they believe will offer the greatest payoff in terms of client outcomes. Such selection may be linked to client ethnicity (analogous to what occurs in some teaching environments).
72
GREG A. GREENBERG
Another related criticism that might be made is that several MC practices that might reduce ethnic disparities were not discussed. Examples include disease and case management, in which coordination of care is increased. This review focused on practices that have seen the most general use and are experienced by most care recipients. Additionally, very little literature exists on ethnic differences in access among clients who are more intensively ‘‘managed’’ in this manner and, perhaps more important, such a focus would ignore the possibility that minorities have less access to such intensive care management (Barrio et al., 2003). To conclude, some consistent empirical support exists for the argument that MC boundary-setting practices disadvantage minorities and much of the literature provides very coherent and logical arguments for why these and other MC practices disadvantage minorities. However, few studies exist of the MC practices used to coordinate care that have an appropriate methodology and the findings of such studies are inconsistent. Thus, it is far too early to reach any definitive conclusions regarding the effects on ethnic differences in access to health services of MC practices in general or of specific MC practices, particularly of those that are used to coordinate care. Because the growing costs of health care and increasing needs associated with an aging population will likely lead to increased use of MC practices to ration care, it is imperative that greater and more appropriate research efforts are focused on whether specific MC practices differently affect minorities’ access to heath care, and if so, what aspects of the MC practice are responsible, so that, if necessary, targeted policies can be developed.
NOTES 1. Only the term ‘‘ethnic’’ is used here, rather than ‘‘race and ethnic.’’ 2. The reason for the focus on this outcome is that minorities obtain a disproportionate share of their care from minority providers. For instance, although black physicians are only 4 per cent of the physician work force, they care for more than 20 per cent of black patients (Saha, Taggart, Komaromy, & Bindman, 2000). Minority providers also often have particular skills, such as language ability and cultural knowledge, and experience treating the special health care needs of minority groups (Anonymous, 1995; Elder, 2002; IOM, 2003; Mackenzie et al., 1999). Thus, if a MCO lacks minority providers, its minority members may find it more difficult to obtain services from an appropriate and conveniently located provider (Anonymous, 1995; Elder, 2002). A caveat to these benefits exists, though: efforts to encourage patient–provider matching on ethnicity would raise several troubling moral and legal issues; and studies have not confirmed the existence of benefits from ethnic matching for minority patients or that minority patients prefer minority providers when
U.S. Minorities’ Access to Health Care under MC
73
factors such as location are controlled for (Balsa et al., 2003; IOM, 2003; Stinson & Thurston, 2002). 3. A health insurance plan is any entity that provides insurance; that entity may or may not also coordinate service delivery (i.e., it may or may not be a MCO).
ACKNOWLEDGMENT Helpful comments on earlier drafts were received from Graham Cassano, Cressida Lui, and Theresa Scheid.
REFERENCES Anonymous. (1995). The impact of managed care on doctors who serve poor and minority patients. Harvard Law Review, 108(7), 1625–1642. Balsa, A., Seller, N., MgGuire, T., & Bloche, G. (2003). Clinical uncertainty and healthcare disparities. American Journal of Law and Medicine, 29, 203–219. Barrio, C., Yamada, A. M., Hough, R. L., Hawthorne, W., Garcia, P., & Jeste, D. V. (2003). Ethnic disparities in the use of public mental health case management services among patients with Schizophrenia. Psychiatric Services, 54(9), 1264–1270. Bindman, A. B., Grumbach, K., Vranizan, K., Jaffe, D., & Osmond, D. (1998). Selection and exclusion of primary care physicians by managed care organizations. Journal of the American Medical Association, 279(9), 675–679. Bloche, M. G. (2001). Race and discrimination in American medicine. Yale Journal of Health Policy, Law, and Ethics, 1, 95–131. Blumberg, L. J., & Nichols, L. M. (1996). First, do no harm: Developing health insurance market reform packages. Health Affairs, 17, 25–42. Chin, J. L. (2000). Culturally competent health care. U.S Department of Health and Human Services Public Health Reports, 115, 25–33. Currie, J., & Fahr, J. (2005). Medicaid managed care: Effects on children’s medicaid coverage and utilization. Journal of Public Economics, 89, 85–108. Davies, H. T., Washington, E. A., & Bindman, A. B. (2002). Health care report cards: Implications for vulnerable patient groups and the organizations providing them care. Journal of Health Politics, Policy, and Law, 27(3), 379–399. Davis, K. (1997a). Assuring Quality, Information, and Choice in Managed Care. New York: Commonwealth Fund. Davis, K. (1997b). Managed care, mental illness and African Americans: A prospective analysis of managed care policy in the United States. Smith College Studies in Social Work, 67(3), 624–641. Elder, K. T. (2002). Managed care’s impact on minority physicians and their patients. Ph.D. dissertation, Department of Public Policy, University of Maryland, Baltimore, MD. Fauman, M. (2002). Negotiating Managed Care: A Manual for Clinicians. Washington, DC: American Psychiatric Publishing. Fosset, J., & Thompson, F. (1999). Back-off not backlash in medicaid managed care. Journal of Health Politics, Policy and Law, 24(5), 1159–1172.
74
GREG A. GREENBERG
Frankford, D. (1999). Regulating managed care: Pulling the tails to wag the dogs. Journal of Health Politics, Policy and Law, 24(5), 1191–1200. Glied, S. (2000). Managed care. In: K. Arrow & M. D. Intriligator (Eds), Handbooks in economics (pp. 708–745). New York, NY: Elsevier. Greenberg, G. A., & Rosenheck, R. (2003). Change in mental health service delivery among blacks, whites, and hispanics in the department of veterans affairs. Administration and Policy in Mental Health, 31(1), 31–43. Grembowski, D. E., Cook, K. S., Patrick, D. L., & Roussel, A. E. (2002). Managed care and the U.S. health care system: A social exchange perspective. Social Science and Medicine, 54, 1167–1180. Guwani, J., & Weech-Maldonado, R. (2004–2005). Medicaid managed care and racial disparities in AIDS treatment. Health Care Financing Review, 26(2), 120–132. Haas, J., Phillips, K., Sonneborn, D., McCulloch, C., & Liang, S.-Y. (2002). Effects of managed care insurance on the use of preventive care for specific ethnic groups in the United States. Medical Care, 40(9), 743–751. Hannan, E., van Ryn, M., Burke, J. MS., Stone, D., Kumar, D., Arani, D., Pierce, W., Rafii, S., Sanborn, T., Sharma, S., Slater, J., & DeBuono, B. (1999). Access to coronary artery bypass surgery by race/ethnicity and gender among patients who are appropriate for surgery. Medical Care, 37(1), 68–77. Hargraves, J. L., Cunningham, P. J., & Hughes, R. G. (2001). Racial and ethnic differences in access to medical care in managed care plans. Health Services Research, 36(5), 853–868. Hofer, T. P., Hayward, R. A., Greenfield, S., Wagner, E. H., Kaplan, S. H., & Manning, W. G. (1999). The unreliability of individual physician report cards for assessing the costs and quality of care of a chronic disease. Journal of the American Medical Association, 281, 2098–2105. Institute of Medicine of the National Academies (IOM). (2003). Unequal treatment: Confronting racial and ethnic disparities in health care. In: B. D. Smedley, A. Y. Stith & A. Nelson (Eds), Washington, DC: The National Academy Press. Kerr, E. A., Mittman, B. S., Hays, R. D., Siu, A. L., Leake, B., & Brook, R. H. (1995). Managed care and capitation in California: How do physicians at risk control their own utilization. Annuals of Internal Medicine, 123, 500–504. Knight, W. (1998). Managed Care: What it is and How it Works. Gaithersburg, MD: Aspen. Kongstvedt, P. (2001). Compensation of primary care in managed health care plans. In: P. Kongstevedt (Ed.), The managed health care handbook (pp. 110–146). Gaithersburg, MD: Aspen. Kronick, R. (1999). Waiting for godot: Wishes and worries in managed care. Journal of Health Politics, Policy and Law, 24(5), 1099–1106. Kuttner, R. (1999). The American health care system: Employee-sponsored health coverage. New England Journal of Medicine, 340, 248–252. La Roche, M., & Turner, C. (2002). At the crossroads: Managed mental health care, the ethics code, and ethnic minorities. Cultural Diversity and Ethnic Minority Psychology, 8(3), 187–198. Leigh, W., Lillie-Blanton, M., Martinez, R. M., & Collins, K. (1999). Managed care in three states: Experiences of low-income African Americans and hispanics. Inquiry, 36, 318–331. Lowe, R., Do, S. C., Nasci, K., Gavin, L., Shaw, K., Zwanger, M., Zeccardi, J., Dalsey, W., Abbuhl, S., Feldman, H., & Berlin, J. (2001). Effect of ethnicity on denial of
U.S. Minorities’ Access to Health Care under MC
75
authorization for emergency department care by managed care gatekeepers. Academic Emergency Medicine, 8(3), 259–266. Luft, H. (1999). Why are physicians so upset about managed care? Journal of Health Politics, Policy and Law, 24(5), 957–972. Mackenzie, E., Taylor, L., & Lavizzo-Mourey, R. (1999). Experiences of ethnic minority primary care physicians and managed care: A national survey. The American Journal of Managed Care, 5(10), 1251–1264. Perez, T. (2003). The civil rights dimension of racial and ethnic disparities in health status. In: B. D. Smedley, A. Y. Stith & A. Nelson (Eds), Unequal treatment: Confronting racial and ethnic disparities in healthcare (pp. 626–663). Washington, DC: National Academy Press. Phillips, K., Mayer, M., & Aday, L. A. (2000). Barriers to care among racial/ethnic groups under managed care. Health Affairs, 19(4), 65–75. Randall, V. (1994). The impact of managed care organizations on ethnic Americans and underserved populations. Journal of Health Care for the Poor and Underserved, 5(10), 224–236. Randall, V., Giess, S., Boller, G., Tinkler, C., Bayless, S., Romero, A., Henry, S., & Whipple, C. (1996). ‘‘The market place and regulation: Section 115 medicaid waivers.’’ Medicaid managed care: Symposium on consumer protection in managed care. Seton Hall Law Review, 26, 1069–1142. Reschovsky, J. (1999). Do HMOs make a difference? Access to health care. Inquiry, 36, 390–399. Rice, T. (2003). The impact of cost containment efforts on racial and ethnic disparities in healthcare. In: B. D. Smedley, A. Y. Stith & A. Nelson (Eds), Unequal treatment: Confronting racial and ethnic disparities in healthcare (pp. 699–721). Washington, DC: National Academy Press. Rosenbaum, S. (2003). Racial and ethnic disparities in healthcare: Issues in the design, structure, and administration of federal healthcare financing programs. In: B. D. Smedley, A.Y. Stith & A. Nelson (Eds), Unequal treatment: Confronting racial and ethnic disparities in healthcare (pp. 664–698). Washington, DC: National Academy Press. Rosenbaum, S., Serrano, R., Magar, M., & Stern, G. (1997). Civil rights in a changing health care system. Health Affairs, 16(1), 90–105. Saha, S., Taggart, S., Komaromy, M., & Bindman, A. (2000). Do patients choose physicians of their own race? Health Affairs, 19(4), 76–83. Saxton, J., & Leaman, T. (1998). Managed Care Success. Gaithersburg, MD: Aspen. Schlesinger, M. (1987). Paying the price: Medical care, minorities and the newly competitive health care system. Milbank Quarterly, 65(2), 270–296. Schlesinger, M., & Gray, B. (1999). Institutional change and its consequences for the delivery of mental health services. In: A. Horowitz & T. Scheid (Eds), A Handbook for the study of mental health: Social contexts, theories, and system (pp. 427–448). Cambridge: University Press. Scott, W. R. (2003). Organizations: Rational, Natural, and Open Systems (5th ed.). Upper Saddle River, NJ: Prentice Hall. Simon, C., White, W., James, C., & Hernandez, L. (Forthcoming). African American physicians and managed care: What has been the impact? Smith, D. (1999). Health care divided: Race and healing a nation. Ann Arbor, MI: University of Michigan Press.
76
GREG A. GREENBERG
Snowden, L. (1998). Managed care and ethnic minority populations. Administration and Policy in Mental Health, 25(6), 581–592. Stensen, R. (2000). A physician’s guide to thriving in the new managed care environment. Wilsonville, OR: BookPartners. Stinson, M., & Thurston, N. (2002). Racial matching among African-American and hispanic physicians and patients. The Journal of Human Resources, 37(2), 410–428. Stoddard, J., Reed, M., & Hadley, J. (2003). Financial incentives and physicians perceptions of conflict of interest and ability to arrange medically necessary services. Journal of Ambulatory Care Management, 26(1), 39–50. Sturm, M. (2001). Underwriting and rating function by market. In: P. Kongstevedt (Ed.), The managed health care handbook (pp. 952–970). Gaithersburg, MD: Aspen. Tai-Seale, M., Freund, D., & Sasso, A. L. (2001). Racial disparities in service use among medicaid beneficiaries after mandatory enrolment in managed care: A difference-indifference approach. Inquiry, 38, 49–59. Wallace, S. P., Enriquesz-Haas, V., & Markides, K. (1998). The consequences of color blind health policy for older racial and ethnic minorities. Stanford Law and Policy Review, 9(2), 329–346. Weiner, J., & Lissovoy, G. (1993). Razing a tower of Babel: A taxonomy for managed care and health insurance plans. Journal of Health Politics, Policy and Law, 18(1), 75–112. White, J. (1999). Targets and systems of health care cost control. Journal of Health Politics, Policy and Law, 24(4), 653–696. Zellers, W. K., McLauphlin, C. G., & Frick, K. D. (1992). Small-business health insurance: Only the healthy need apply. Health Affairs, 11(1), 174–180.
KOREAN IMMIGRANTS AND HEALTH CARE ACCESS: IMPLICATIONS FOR THE UNINSURED AND UNDERINSURED Grace J. Yoo and Barbara W. Kim ABSTRACT As an ethnic group, Korean Americans have one of the highest uninsured rates in the U.S. (Brown et al., 2000). Through in-depth interviews (n = 14) and surveys (n = 268), this study found that one-third of the sample was uninsured. High premiums prevented the uninsured from purchasing health insurance. Although health insurance has been a strong predictor of health services utilization, this study also found that when examining the utilization of various health services by health insurance status, there were no major significant differences with the exception of Korean traditional health services. High deductibles prevented insured persons from utilizing health services.
Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 77–94 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00004-X
77
78
GRACE J. YOO AND BARBARA W. KIM
INTRODUCTION In 2005, over 46.6 million people were uninsured in the U.S. (U.S. Census Bureau, 2006). The uninsured in the United States go without needed regular health check-ups and seek care in extreme emergencies, and are as a result extremely disadvantaged in terms of health. Low socio-economic status, adverse health behaviors, and lack of health insurance have all been identified as social factors that impact health (Huang, Yu, & Ledsky, 2006; Franzini & Fernandez Esquer, 2006; Kirby, Taliaferro, & Zuvekas, 2006; Lantz et al., 2006; Farmer & Ferraro, 2005; Lillie-Blanton & Hoffman, 2005; Shone, Dick, Klein, Zwanziger, & Szilagyi, 2005; Hargraves & Hadley, 2003; Guendelman, Wyn, & Tsai, 2000). Compared to those that have health insurance, individuals who lack health insurance coverage have been shown to have higher risk-adjusted rates of decline in their overall health and physical functioning (Baker, Sudano, Albert, Borawski, & Dor, 2001) and higher risk-adjusted mortality (McWilliams, Zaslavsky, Meara, & Ayanian, 2004). Likewise, reduced access to care can have serious consequences for health outcomes via lack of preventive services use, delayed diagnosis of disease, and poor monitoring and control of chronic diseases (Institute of Medicine, 2002). Another important goal for the nation is to reduce the level of disparities in access to health care by race, ethnicity, and nativity in the United States. In 2003, immigrants were 26% of the nation’s uninsured; immigrants were more likely to be twice as likely uninsured than U.S. citizens, while newer immigrants also were less likely to be insured than older immigrants (Employee Benefits Research Institute, 2005a). In fact, the two largest immigrant groups that lack health insurance coverage in the United States are Latino (58%) and Asian (30%) non-citizens (Brown et al., 2000; Ku & Blaney, 2000). The higher rates of uninsurance in these groups are attributed to lower rates of employment-based insurance (Brown et al., 2000). This delay in seeking health care due to uninsurance has major health implications for communities. Delayed diagnoses of conditions such as cancer, heart disease, and diabetes can reduce effective treatment and management. Access barriers reduce the use of preventative services. In the last decade, Koreans have had one of the lowest rates of health insurance coverage among all racial and ethnic groups. The 1994 Current Population Survey found that 50% of Koreans ages 0–64 did not have health insurance. The 2001 California Health Interview Survey found that 21.7% of Korean children and 33.6% of non-elderly Korean adults ages 18–64 lacked health insurance among ethnic groups (Brown, Ponce,
Korean Immigrants and Health Care Access
79
Rice, & Lavarreda, 2002). The 2000 Korean American Health Survey of Koreans living in Los Angeles County found that 46% of the respondents did not have health insurance (Shin, Song, Kim, & Probst, 2005). The lack of health insurance coverage has had an impact on the health of the Korean American community as, in comparison, only 15.7% of the total U.S. population lacked insurance coverage in 2004 (U.S. Census, 2006). Factors such as income, education, citizenship, and access to job-based insurance explain racial and ethnic disparities in health insurance coverage and access to health care; Brown, Lavarreda, Rice, Kincheloe, and Gatchee (2005) also found that Korean immigrants do not carry health insurance for themselves and attributed this to the high rates of self-employment and working in small businesses that do not provide employmentbased insurance. Most Korean immigrant adults do not qualify for Medicaid, yet cannot afford private health insurance. Because many Korean immigrants do not have access to health insurance, this also means that these immigrants do not have a regular source of health care through which they can access cancer screenings and preventative services such as immunizations (Ryu, Young, & Park, 2001; Shin et al., 2005). Past research has shown that the lack of health insurance was the strongest predictor of health services utilization for Korean Americans (Ryu et al., 2001). At the same time, acculturation and income are factors in terms of health insurance coverage and health services utilization (Shin et al., 2005; Kim & Yoo, 2006). Research on Korean immigrants and their utilization of preventative health care such as breast and cervical screenings indicate that economic barriers (such as lack of health insurance), working long hours, language barriers, and length of residency in the U.S. affect utilization rates of preventative health practices (Juon, Choi, & Kim, 2000; Juon, Kim, Shankar, & Han, 2004; Lee, 2000; Kim et al., 1999; Maxwell, Bastani, & Warda, 2000).
KOREAN AMERICANS: DEMOGRAPHICS AND HISTORY The 2000 census counted over 1.2 million Korean Americans in the U.S.; the Korean population grew by about one-third between 1990 and 2000, and is the fifth largest Asian ethnic population (U.S. Census Bureau, 2001). Although Korean Americans celebrated the centennial anniversary of immigration in 2003, the census also indicated that 78% of the population is foreign-born. The Korean American population reflects a historical legacy
80
GRACE J. YOO AND BARBARA W. KIM
of exclusionary immigration and citizenship laws aimed at Asians as well as the significance of contemporary immigration policies and social, economic, and political conditions in South Korea and the U.S. Korean immigration to the United States can be divided into three different waves. Although few Koreans came to the U.S. before the turn of the 20th century, the first wave of Koreans, mostly men, migrated to the U.S. in 1903 as contract laborers who worked in the sugarcane fields in Hawaii and in agriculture along the West Coast. Over 7,000 Korean immigrants arrived before the Korean government ended emigration in 1905 in response to pressure from Japan. Although laborers were not allowed to emigrate, almost 1,000 Korean picture brides arrived between 1910 and 1924 to join – or in some cases, meet for the first time – their husbands under a provision of the Gentlemen’s Agreement between Japan and the U.S. With the passage of the Immigration Act of 1924, Koreans were barred from migrating to the United States (Hurh & Kim, 1980; Park, 1997). With the outbreak of the Korean War in 1950, the second wave of Korean immigrants, representing war brides, adoptees, and students, arrived. The largest wave of Korean immigrants arrived after the passage of the 1965 Immigration Act that abolished the racial quotas, which for decades had effectively excluded most Asians from immigrating to the U.S. The Immigration Act favored family reunification and the migration of specialized trained individuals in short supply for certain professional work such as health professionals, scientists, and engineers. Middle-class professionals from urban backgrounds, responding to economic, political, and cultural developments in South Korea and the U.S., largely comprised the third wave until the U.S.-amended immigration laws in the 1970s to restrict occupational preference admissions. Since then, Koreans have entered largely through family reunification preferences and represent a larger spectrum of education levels and occupational backgrounds more reflective of the general South Korean population (Park, 1997). A large percentage of those immigrating after 1965 have experienced downward mobility and language barriers in accessing work regardless of education levels and pre-migration occupations. In response, many Korean immigrants have steered toward self-employment and demonstrate one of the highest propensities among all racial/ethnic groups to enter into business, concentrating in retail and service businesses (Yu & Chang, 2003). In 2000, approximately 20% of Koreans were self-employed, the highest rate among racial/ethnic groups and double the rate of the total U.S. population (Korean American Coalition – Census Information Center & Center for Korean American and Korean Studies, 2002). Many Korean immigrants have
Korean Immigrants and Health Care Access
81
opted for self-employment because they could not find a job commensurate with their education and work experience in the primary labor market; this high self-employment rate may contribute to the fact that many Korean Americans lack job-based health insurance and have the lowest rates of health insurance coverage among all racial and ethnic groups in the U.S. (Brown et al., 2005). Ethnic entrepreneurship comes at a cost to the health and well-being of Korean American entrepreneurs, families, employees, and communities.
METHODOLOGY According to the 2000 Census, the Los Angeles-Riverside-Orange County and San Francisco-Oakland-San Jose metropolitan areas are respectively, home to the largest and the third largest Korean American populations in the U.S. (Yu & Chang, 2003). This study utilized a purposive sample of Korean immigrant small business owners age 18 and over residing in the greater Los Angeles and San Francisco Bay areas between 2005 and 2007, due to previous studies indicating the significance of ethnic entrepreneurship and access to health insurance. A total of 268 respondents completed survey questionnaires. Fourteen follow-up in-depth interviews, seven in the Los Angeles area and seven in the San Francisco Bay area, were conducted. Potential participants were recruited at their business locations in areas with a high concentration of Korean immigrant small businesses such as San Francisco’s Japantown and the Oakland’s Koreatown on Telegraph Avenue. They were also recruited through various Catholic and Presbyterian churches in the greater Los Angeles and San Francisco Bay areas. Questions in the self-administered survey included demographics (such as age, gender, education level, and length of residence in the U.S.), access to health care for respondents and their family members, utilization of health care services, and barriers and challenges to health care access. The survey questionnaires, in-depth interview schedule, and consent forms were written in English and translated into Korean and rechecked for consistency by bilingual research assistants. The collected survey data were cleaned, coded, and entered into a statistical program (SPSS). Descriptive frequencies were conducted in terms of the demographics and utilization of health insurance. Comparisons of categorical characteristics between Korean immigrant with variables of health insurance coverage and health service were done for statistical significance using w2 tests.
82
GRACE J. YOO AND BARBARA W. KIM
In-depth interview participants were asked about barriers and challenges with utilizing health insurance and retirement planning. Bilingual researchers conducted all interviews in Korean, which were tape-recorded with consent, translated, and transcribed into English. The transcripts were reviewed for content analysis and an open coding process to develop emerging categories from the data (Strauss & Corbin, 1998). All interviewees are identified by pseudonyms.
RESULTS Demographic Background The respondents’ ages ranged from 22 to 69 with a mean age of 49 years. While those who immigrated to the U.S. between 1991 and 2005 comprised 28.7% of the sample, 70.9% of the participants immigrated between 1965 and 1990 during the peak immigration decades of 1970s and 1980s that followed emigration and immigration reforms in South Korea and the U.S. The respondents were split evenly by gender and the majority (86.6%) were married while 13% were single (never-married, widowed, or divorced). About 90% had children. About 30% of the participants reported an annual family income below $44,999, 40% reported incomes between 45,000 and 89,999, and about 30% reported an annual family income above $90,000. The median household income of Korean Americans in the U.S. in 2003 was $40,183 (Asian and Pacific Islander American Health Forum, 2005). Additionally, 70% reported that their spouses also were employed. The results show that 71% had health insurance for themselves but no one indicated that they had disability, long-term care, or dental insurance.
Health Insurance Too Expensive Twenty-nine percent of our sample did not have health insurance. The uninsured participants most frequently cited affordability as the reason why they lacked health insurance for themselves and their spouses. Participants indicated that insurance payments were a ‘‘financial burden’’ and ‘‘too expensive’’ as an additional expenditure. Some respondents did not have health insurance for themselves and their spouses but their minor children were insured through programs such as Medicaid and Healthy Families, California’s State Children’s Health Insurance Program (SCHIP). Through
Korean Immigrants and Health Care Access
83
written responses or in-depth interviews, participants said that rather than paying expensive health insurance premiums every month, they opted to pay fee-for-services as needed, like Mr. Nam of San Francisco (personal communication, November 16, 2005): I am healthy, and there is no one in family that is seriously sick. Plus it costs so much. I feel no need for health insurance. Since we don’t have health insurance, if a major health issue came up, I would have to rely on my family for help.
Respondents cited their current good health and no perceived need for immediate or consistent medical care as a major reason for not having health insurance. Mrs. Choi, living in the Los Angeles area, saw paying for a doctor’s services as needed as a better financial alternative to paying high monthly insurance premiums (Mrs. Choi, personal communication, November 18, 2005). I have not had a major illness and that’s why I don’t have [health insurance]. Now that I’m older and I’ve been working so hard – I don’t have major illnesses but get exhausted so I have reasons to [see a doctor]. But the reason why I can’t get health insurance is because I would have to pay for it every month and rather y since I only see a doctor several times a year due to stress or colds, I think that it saves money to pay when I go rather than to get insurance.
Ultimately, not having health insurance impacts the financial state of uninsured Korean immigrants, and sometimes in the most catastrophic way by depleting savings and racking credit card debt. Ms. Suh, who lives in the Los Angeles area, had such an incident (Ms. Suh, personal communication, January 21, 2007): If it wasn’t so expensive I would definitely have health care. I try to live as healthy as possible by taking supplements and having a good diet so that I don’t have to visit the doctor as much. There was an incident 7 years ago when my youngest son got sick when he was a month old and was hospitalized for 4 days. Those 4 days cost $43,000 y The bill was put onto our credit card.
Ms. Suh eventually negotiated a lower repay amount with the hospital and paid off the medical bills, but her family continues to be uninsured. Mrs. Park of Los Angeles said that she could not afford health insurance premiums but their family’s income makes them ineligible for Medi-Cal or Healthy Families. She shared what happened when her child had an accident (Mrs. Park, personal communication, December 1, 2005): y last year, my youngest broke her arm. There was also a hairline fracture. That took about $6000, $7000 for same-day surgery, cast, etc. So when you have one incident like that, it’s really difficult y . I think [the physician] saw her for about 20 minutes and put a
84
GRACE J. YOO AND BARBARA W. KIM cloth sling on it – a bandage? And then we were charged $670 ... I was so upset. This was on top of a $200 deposit that they made us pay, just to see someone, even if they didn’t have anyone who could take care of it.
Although upset by the cost and the process of her uninsured child’s emergency treatment, Mrs. Park could not address these matters with the clinic staff, due to her limited English proficiency. Like Ms. Suh, she used a credit card to pay her child’s medical bill. The uninsured respondents who cited affordability as a barrier shared that they tried to take care of their health through exercise, diet, and taking traditional Korean herbal medicine. Despite their efforts, Ms. Suh and Mrs. Park’s examples illustrate the precarious financial situations of working, self-employed families in the U.S. Health Insurance Used Only for Emergency Purposes Respondents cited affordability most frequently as the reason why they chose their particular insurance plan (39.2%). Table 1 shows that for the respondents, the mean yearly deductible was $7,173 but the minimum ranged from $100 to $16,800. The variation in the yearly deductible of health insurance demonstrates that respondents were purchasing a variety of health insurance packages, including preferred provider organizations (PPOs), health management organizations (HMOs), and other insurance plans. Those with higher yearly deductibles often had lower monthly costs, while those with higher monthly premiums had lower yearly deductibles. Higher deductibles also meant that respondents had large families, with each family member having a deductible to meet before their health insurance would cover health costs. Monthly costs for insurance plans ranged from $16 to $1,400 with the mean being $381 per month. Co-payments for office visits ranged from $5 to $273. The limitations of these numbers do not account those with pre-existing conditions which may result in higher monthly payments and deductibles.
Table 1.
Mean Out-of-Pocket Payments among the Insured.
Mean Costs Mean yearly deductible (n=31) Mean co-payment for office visit (n=43) Monthly payment for health insurance (n=56)
$7,173.77 $40.42 $381
Korean Immigrants and Health Care Access
85
Because many of the respondents wanted to maintain a low monthly payment, often their yearly deductibles were quite high. This meant that when they utilized health care, it was often not covered by their insurance company. Several respondents discussed the difficulty of obtaining health insurance that provided 100% coverage and having expensive deductibles that prevented them from seeing health care providers on a regular basis. Rather, they treated their health care plans as coverage to be used for only emergency purposes. For example, Mrs. Lim explained (Mrs. Lim, personal communication, November 17, 2005): Sometimes when I have a headache, I feel like I should get an X-ray and really check things out. But then I feel it’s such a small thing and I know it will cost me hundreds of dollars. So I just ignore it and just go in when I think it’s an emergency.
As a result of the high premiums and deductibles, even those who had insurance discussed their reluctance to seeking health care and incurring out-of-pocket expenses, accessing health services only for what they perceived as an emergency or a serious illness. Mrs. Choi had to pay $3,000 out-of-pocket for a breast biopsy to test a suspicious lump and said, ‘‘It felt like I didn’t have health insurance’’ (Mrs. Choi, personal communication, December 11, 2006). For respondents with high deductibles, health care visits and procedures were costs they incurred and paid rather than those covered by monthly premiums and co-payments. All the respondents noted that the key obstacle for health insurance and ultimately health care access was the cost. Mrs. Chung, who also noted that ‘‘finding and making the time to go is difficult’’ while running a business, discussed how cheaper health care coverage often meant higher deductibles and co-payments (Mrs. Chung, personal communication, January 19, 2007): The problem is money. If the coverage is cheaper, then it would be easier. In our situation, because the deductible is high, any check-ups $100 and below we pay with cash (out-of-pocket) y But if I need to be hospitalized or have a serious illness, then I’ll use the health insurance coverage because I need to.
Although this study found that Korean immigrant entrepreneurs continue to have a high rate of uninsurance, having health insurance did not translate into an increased use of health services. When examining the utilization of various health services by different health insurance status, there were no major significant differences with the exception of Korean traditional health services. There were significant differences in the utilization of Korean traditional health services between the insured and uninsured. The uninsured were more likely to have seen a Korean traditional medicine
86
GRACE J. YOO AND BARBARA W. KIM
practitioner (88%) in the last year compared to those who were insured (66%). This difference (66% compared to 88%) was statistically significant ( p-value=0.014). Moreover, when examining health care services, more Korean immigrants sought out Korean traditional health services than conventional health services. Respondents also utilized traditional healing practices at a high rate, with 69% of participants reporting that they or someone in their family had visited a traditional doctor, most frequently for acupuncture treatments and herbal medicine. Mr. Moon, who is uninsured, stated his reasons for use of Korean traditional healers (Mr. Moon, personal communication, November 22, 2005): I had pain in my shoulders so I went to a Korean acupuncturist. And I also took han-yak [Korean herbal medicine] because I believe that it works to heal the body more naturally and better than prescribed medicine.
Participants generally utilized traditional Korean practitioners for preventive care (check-ups and health supplements) and treatments for ongoing symptoms (e.g., allergies and backaches). They also commented that the costs to see a Korean traditional medicine practitioner were more affordable than seeing a medical provider and that costs ranged from $10 for herbal remedies to $50 for an acupuncture session. When exploring use of emergency-related health services between the insured and uninsured, there were no significant differences (Table 2). Thirteen percent of the uninsured had utilized emergency-related services, while 16% of the insured had utilized such services. At the same time, 38% of those with health insurance utilized adult preventative health services compared to 29% of those without health insurance. This difference (38% compared to 29%) was not statistically significant ( p-value=0.436). When examining diagnostic type of health services, 35% of those with health insurance utilized diagnosis and treatment services for illness and injuries compared to 33% of those without health insurance. This difference (35% compared to 33%) was not statistically significant ( p-value=0.863). When examining child preventative health services, 6% of those with health insurance utilized such services as compared to 13% without health insurance. This difference (13% compared to 6%) was not statistically significant ( p-value=0.381). In terms of family planning and pregnancyrelated services, 6% of those with health insurance utilized family planning and pregnancy-related services compared to 8% of those without health insurance. This difference (6% compared to 8%) was again not statistically significant ( p-value=0.634).
Korean Immigrants and Health Care Access
Table 2.
87
Respondent’s Family Utilization of Health Services by Health Insurance Status. Percentage of Utilization
p-Value
Herbalist/acupuncturist Without health insurance With health insurance
88 70
0.014
Emergency health services Without health insurance With health insurance
13 16
0.768
Adult preventative health services Without health insurance With health insurance
29 38
0.436
Diagnostic/treatment Without health insurance With health insurance
33 35
0.863
Child preventative health services Without health insurance With health insurance
13 6
0.381
6 8
0.634
Family’s Use of Health Services (n=249)
Family planning/pregnancy-related services Without health insurance With health insurance
Even when respondents had health insurance, this did not mean they had access to other insurances that covered other health-related expenses. None of the participants indicated that they had other types of health-type insurances including dental, vision, disability, and long-term care insurance. For example, one respondent who has health insurance but no dental insurance incurred a $10,000 bill for dental implants.
DISCUSSION Making ends meet for immigrant families becomes a barrier to not only acquiring health insurance, but also in buying quality, accessible health insurance. Twenty-nine percent of immigrants in the sample were uninsured, while 71% of this sample had health insurance coverage. Even with obtaining a modest amount of health insurance, financial difficulties and
88
GRACE J. YOO AND BARBARA W. KIM
barriers still pose a challenge to accessing health care services. Past research has indicated that uninsured Korean immigrants were unable to access needed health services (Ryu et al., 2001). However, the high costs associated with accessing health insurance did not translate into the receipt of care among the insured in this study; there were no significant differences in different types of health care utilization between the insured and uninsured. As a result, Korean immigrants with health insurance could be classified as underinsured – those with inadequate health insurance coverage and having medical bill problems (Pryor, Cohen, & Prottas, 2007). According to Schoen, Doty, Collins, and Holmgren (2005), 12 percent of those insured in the U.S. were underinsured and were significantly more likely to forgo needed health care due to costs compared to individuals with adequate insurance coverage. As a result, the underinsured, similar to the uninsured, are unable to seek proper care and adhere to follow-up care and prescribed medication as recommended by their physicians. This study illustrates that even when Korean immigrants purchase private health insurance, costs to use such health services are a barrier to needed health services. Pryor et al. (2007) found that those who purchase private insurance often chose out of necessity to choose plans with lower monthly premiums; compared to those with employer-sponsored health plans, those with private health plans not only experienced higher premiums, but also higher deductibles, and levels of cost-sharing. Citing financial burdens, they opted for buying a cheaper health insurance plan that entailed lower monthly costs. However, the disadvantage to such a benefit has meant a higher yearly deductible and a higher co-payment per visit. For example, one respondent mentioned that her family would need to pay $4,000 out of their own pocket before their health insurance would pay for any medical costs, and as a result she delayed seeking out different preventative cancer screenings. Research by Davis, Doty, and Ho (2005) found that highdeductible plans can increase financial barriers to obtaining necessary care without protection against financial hardships, as individuals with a high deductible have significantly greater difficulty accessing care due to cost compared to those with a lower or no deductible. Thirty-eight percent of adults with deductibles of $1,000 or more reported problems with health care access, not filling prescriptions, not accessing specialist care, and forgoing recommended tests and needed care. Owing to the financial costs of conventional health care and cultural reliance on traditional Korean medicine and practices, Korean immigrants in this study have sought out traditional medicine for preventive care and treatments. The results indicate that uninsured Korean immigrants were
Korean Immigrants and Health Care Access
89
more likely to seek out traditional health care providers than those that were insured, especially with the comparatively affordable costs of herbal medicine and acupuncture treatments. Previous research has shown that among Asian immigrants traditional health care is still widely practiced, but that those with insurance were more likely to utilize traditional Asian medicine (Ma, 1999; Pourat, Lubben, Wallace, & Moon, 1999; Jenkins, Le, McPhee, Stewart, & Ha, 1996). The respondents did not cite cost as a barrier to accessing traditional Asian medicine. The costs of health care access are barriers for both the uninsured and insured, but immigration status can also act as a barrier in seeking affordable public health insurance programs. Several uninsured respondents in this study had minor children who were covered through Healthy Families, which provides health, dental, and vision coverage with no or low co-payments to children. However, according to the Asian Pacific Islander American Health Forum (2005), newer, uninsured respondents expressed concern that their immigration status also impacted their ability to access publicly funded health care programs. Most legally admitted immigrants who arrived after 1996 are not eligible for programs such as Medicaid and SCHIP until they have resided in the U.S. for five years, while some immigrants who would be eligible for these programs fear that these benefits may negatively impact their process of applying for U.S. citizenship. As a result, non-U.S. citizens, regardless of immigration status, are less likely to utilize such insurance programs.
CONCLUSION Studies have shown that non-elderly people with low income, limited English proficiency, foreign birth, lack of citizenship, and high rates of self-employment are more likely to be uninsured (Brown et al., 2005; Employee Benefits Research Institute, 2005b). Moreover, previous research has consistently stated that Korean immigrants represent a high percentage of the uninsured and that lacking health insurance is a barrier to seeking and obtaining health services. Twenty-nine percent of our participants did not have health insurance – a figure much lower than previously reported for this population. As Korean immigrants have become more aware of the importance of the need for health insurance in the U.S., more people in the last decade have been purchasing private, non-group health insurance. Yet even when they purchase health insurance, Korean immigrants often find
90
GRACE J. YOO AND BARBARA W. KIM
themselves underinsured and still face difficulties in accessing needed health services. Those who purchase private non-group health insurance have higher co-payments, deductibles, and premiums compared to those with employersponsored health plans (Pryor et al., 2007). Korean immigrants, who have high rates of self-employment and have no access to employer-sponsored health plans, obtain private, non-group health insurance plans. Research on the barriers and challenges of Korean immigrants obtaining and utilizing a quality, affordable, non-group health insurance plan has largely been unexplored. The findings of this study show that Korean immigrant entrepreneurs continue to exhibit high rates of uninsurance and even those with private health insurance were limited in their health care coverage. Their choices were often limited to a plan with lower monthly premiums but higher co-payments and higher deductibles. The results in this study show that the mean household deductibles were over $7,000. This means that families must spend thousands of dollars out-of-pocket before their health insurance covers health expenses. Our findings also indicate that as a result of these expenses, having health insurance did not automatically translate into utilizing health care services. Respondents did not utilize their coverage for more preventative health screenings and expected that they would use it in cases of emergencies and serious or catastrophic illnesses. According to Pryor et al. (2007), health insurance plans are not comprehensive, meaning that they fail to cover medically necessary and effective services and treatments (such as prescription drugs) as well as preventive care and disease management. Such plans must also be affordable and accessible even for persons with preexisting conditions and health risk factors, through state and federally regulated standards that oversee and implement comprehensive and affordable coverage and benefits. Affected by historical immigration and naturalization exclusion policies, the contemporary Korean American population is a largely foreign-born group. Scholars have noted that due to linguistic and cultural barriers that they face in the primary labor market, remarkable numbers of post-1965 Korean immigrants have turned to self-employment for livelihood. But this tendency to work as small business owners, has resulted in significant trends toward being uninsured or underinsured that threatens the health and financial security of Korean immigrant families. Newer, uninsured, and lowincome Korean immigrants, who are unable to purchase health insurance, face hurdles in accessing health care but so do middle-income Korean Americans who are caught in the escalating costs and confusing policies of
Korean Immigrants and Health Care Access
91
existing health care coverage plans. Health insurance and immigrants’ access to adequate, comprehensive coverage is a major policy issue. New immigrants face a five-year ban on access to health and other programs as a result of the 1996 Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA). Although native-born Americans made up 73.9% of the uninsured population in 2003, immigrants accounted for an 86% growth in the uninsured population between 1998 and 2003 (Employee Benefits Research Institute, 2005a). The involvement of immigrant voices in the policy debates around the escalating and potentially devastating costs, financial and otherwise, of contemporary health care access in the U.S. is essential. They are often not voiced due to language barriers but illustrate the lack of health insurance coverage options available to low- and middle-income individuals and families who opt for no or inadequate coverage in response to unaffordable premiums, co-payments, and/or deductibles. Mr. Song of Los Angeles (Mr. Song, personal communication, December 20, 2006) summarizes the difficulties and the health care crisis in today’s society: I work over 80 hours a week. I am considered middle-income. We, middle-income immigrants, who are working extremely hard for our families should have some benefits. We should have health insurance. We pay lots of taxes, but we do not see any benefits. We are stuck. We are not able to go to the doctor when we need to. When we get health insurance, we end up paying for everything. It’s totally unfair.
ACKNOWLEDGMENTS The authors thank the assistance of Yongkyoo Jason Chang and the assistance of the following students: Kathie Ahn, Mina Chang, Ellie Hong, Suk Hong, Sin (Nicole) Kang, Brian Kim, Ji Hee Kim, Minnie Kim, Shin-hyoung Kim, Jin Lee, Yun-chu Lee, Chong Lim, Gabriel Oh, and Jenny Suh. They also thank all the study participants and the Korean American Economic Research Council of the Korean American Economic Development Center for providing a Community Research Grant to conduct this research project.
REFERENCES Asian and Pacific Islander American Health Forum. (2005). Diverse Communities, Diverse Experiences: The Status of Asian Americans & Pacific Islanders in the U.S.: A Review of
92
GRACE J. YOO AND BARBARA W. KIM
Six Socioeconomic Indicators and their Impact on Health. San Francisco: Asian and Pacific Islander American Health Forum. Retrieved on February, 2007, from http:// www.apiahf.org/resources/pdf/Diverse%20Communities%20Diverse%20Experiences.pdf Baker, D. W., Sudano, J. J., Albert, J. M., Borawski, E. A., & Dor, A. (2001). Lack of health insurance and decline in overall health in late middle age. New England Journal of Medicine, 345(15), 1106–1112. Brown, E. R., Lavarreda, S. A., Rice, T., Kincheloe, J. R., & Gatchee, M. S. (2005). The state of health insurance in California: Findings from the 2003 California health interview survey. Retrieved on June 20, 2006, from http://www.healthpolicy.ucla.edu/pubs/files/ SHIC03_RT_0 81505.pdf Brown, E. R., Ojeda, V., Wyn, R., & Levan, R. (2000). Racial and ethnic disparities in access to health insurance and health care. Retrieved on October 10, 2005, from http:// healthpolicy.ucla.edu/pubs/files/Racialand EthnicDisparitiesReport.pdf Brown, E. R., Ponce N., Rice T., & Lavarreda S. A. (2002). The state of health insurance in California: Findings from the 2001 California health interview survey. Retrieved on June 20, 2006, from http://healthpolicy.ucla.edu/pubs/files/shic062002.pdf Davis, K., Doty, M. E., & Ho, A. (2005). How high is too high?: Implications of high-deductible health plans. Boston: The Commonwealth Fun. Retrieved on January 8, 2007, from http://www.cmwf.org/usr_doc/816_Davis_how_high_is_too_high_impl_HDHPs.pdf Employee Benefits Research Institute. (2005a). Immigrants make up a growing share of U.S. population without health insurance, study finds. Retrieved on March 30, 2006, from http://www.ebri.org/publications/prel/index.cfm?fa=prelDisp&content_id=3527 Employee Benefits Research Institute. (2005b). Sources of health insurance and characteristics of the uninsured: Analysis of the March 2005 Current Population Survey. EBRI Issue Brief #287. Retrieved on March 30, 2006, from http://www.ebri.org/pdf/briefspdf/ EBRI_IB_11-20051.pdf Farmer, M. M., & Ferraro, K. F. (2005). Are racial disparities in health conditional on socioeconomic status? Social Science Medicine, 60(1), 191–204. Franzini, L., & Fernandez Esquer, M. E. (2006). The association of subjective social status and health in low-income Mexican-origin individuals in Texas. Social Science Medicine, 63(3), 788–804. Guendelman, S., Wyn, R., & Tsai, Y. W. (2000). Children of working low-income families in California: Does parental work benefit children’s insurance status, access, and utilization of primary health care? Health Services Research, 35(2), 417–441. Hargraves, J. L., & Hadley, J. (2003). The contribution of insurance coverage and community resources to reducing racial/ethnic disparities in access to care. Health Services Research, 38(3), 809–829. Huang, Z. J., Yu, S. M., & Ledsky, R. (2006). Health status and health service access and use among children in U.S. immigrant families. American Journal of Public Health, 96(4), 634–640. Hurh, W. M., & Kim, K. C. (1980). Korean immigrants in America: A structural analysis of ethnic confinement and adhesive adaptation. Macomb: Department of Sociology and Anthropology, Western Illinois University. Institute of Medicine. (2002). Care without Coverage: Too Little, Too Late. Washington, DC: National Academy Press. Jenkins, C. N., Le, T., McPhee, S. J., Stewart, S., & Ha, N. T. (1996). Health care access and preventive care among Vietnamese immigrants: Do traditional beliefs and practices pose barriers? Social Science and Medicine, 43(7), 1049–1056.
Korean Immigrants and Health Care Access
93
Juon, H. S., Choi, Y., & Kim, M. T. (2000). Cancer screening behaviors among KoreanAmerican women. Cancer Detection and Prevention, 24(6), 589–601. Juon, H. S., Kim, M., Shankar, S., & Han, W. (2004). Predictors of adherence to screening mammography among Korean American women. American Journal of Preventive Medicine, 39(3), 474–481. Kim, B., & Yoo, G. (2006). Saving for health and the future: Korean immigrant entrepreneurs. Paper presentation at the National Association of Ethnic Studies, April, San Francisco, CA. Kim, K., Yu, E. S., Chen, E. H., Kim, J., Kaufman, M., & Purkiss, J. (1999). Cervical cancer screening knowledge and practices among Korean-American women. Cancer Nursing, 22(4), 297–302. Kirby, J. B., Taliaferro, G., & Zuvekas, S. H. (2006). Explaining racial and ethnic disparities in health care. Medical Care, 44(5 Suppl.), I64–I72. Korean American Coalition – Census Information Center & Center for Korean American and Korean Studies, California State University, Los Angeles. (2002). Korean employment, poverty, and naturalization in the United States: Tables and highlights. Los Angeles: KAC – CIC and CKKS, CSULA. Ku, L., & Blaney, S. (2000). Health coverage for legal immigrant children: New census data highlight importance of restoring Medicaid and SCHIP coverage. Washington, DC: Center on Budget and Policy Priorities. Retrieved on August 14, 2006, from http:// www.cbpp.org/10-4-00health.pdf Lantz, P. M., Mujahid, M., Schwartz, K., Janz, N. K., Fagerlin, A., Salem, B., Liu, L., Deapen, D., & Katz, S. J. (2006). The influence of race, ethnicity, and individual socioeconomic factors on breast cancer stage at diagnosis. American Journal of Public Health, 12, 2173–2186. Lee, M. C. (2000). Knowledge, barriers, and motivators related to cervical cancer screening among Korean-American women: A focus group approach. Cancer Nursing, 23(3), 168–175. Lillie-Blanton, M., & Hoffman, C. (2005). The role of health insurance coverage in reducing racial/ethnic disparities in health care. Health Affairs, 24(2), 398–408. Ma, G. X. (1999). Between two worlds: The use of traditional and western health services by Chinese immigrants. Journal of Community Health, 24(6), 421–437. Maxwell, A. E., Bastani, R., & Warda, U. S. (2000). Demographic predictors of cancer screening among Filipino and Korean immigrants in the United States. American Journal of Preventive Medicine, 18(1), 62–68. McWilliams, J. M., Zaslavsky, A. M., Meara, E., & Ayanian, J. Z. (2004). Health insurance coverage and mortality among the near-elderly. Health Affairs, 23(4), 223–233. Park, K. (1997). The Korean American dream: Immigrants and small businesses in New York City. Ithaca, NY: Cornell University Press. Pourat, N., Lubben, J., Wallace, S. P., & Moon, A. (1999). Predictors of use of traditional Korean healers among elderly Koreans in Los Angeles. The Gerontologist, 39(6), 711–719. Pryor, C., Cohen, A., & Prottas, J. (2007). The illusion of coverage: How health insurance fails people when they get sick. Boston: The Access Project. Retrieved on April 8, 2007, from http://www.accessproject.org/adobe/the_illusion_of_coverage.pdf Ryu, H., Young, W. B., & Park, C. (2001). Korean American health insurance and health services utilization. Research in Nursing & Health, 24(6), 494–505.
94
GRACE J. YOO AND BARBARA W. KIM
Schoen, C., Doty, M. M., Collins, S. R., & Holmgren, A. L. (2005). Insured but not protected: How many adults are underinsured? Health Affairs, 115, e697–e705. Shin, H., Song, H., Kim, J., & Probst, J. (2005). Insurance, acculturation, and health service utilization among Korean-Americans. Journal of Immigrant Health, 7(2), 65–74. Shone, L. P., Dick, A. W., Klein, J. D., Zwanziger, J., & Szilagyi, P. G. (2005). Reduction in racial and ethnic disparities after enrollment in the State Children’s Health Insurance Program. Pediatrics, 6, e697–e705. Retrieved on October 8, 2006, from http://pediatrics.aappublications.org/cgi/content/full/115/6/e697?maxtoshow=&HITS= 10&hits=10&RESULTFORMAT=&fulltext=Reduction+in+racial+and+ethnic+ disparities+after+e&andorexactfulltext=and&searchid=1&FIRSTINDEX=0&sortspec=relevance&resourcetype=HWCIT Strauss, A., & Corbin, J. (1998). Basics of qualitative research: Grounded theory procedures and techniques. Newbury Park, CA: Sage. U.S. Census Bureau. (2001). Census 2000 demographic profile. Retrieved on October 1, 2005, from http://www.census.gov/prod/cen2000/dp1/2kh00.pdf U.S. Census Bureau. (2006). Income, poverty, and health insurance coverage in the United States: 2005. Washington, DC: U.S. Government Printing Office. Retrieved on November 10, 2006, from http://www.census.gov/prod/2006 pubs/p60-231.pdf Yu, E., & Chang, E. (2003). Entrepreneurs par excellence. In: E. Lai, & D. Arguelles (Eds), The new face of Asian Pacific America: Numbers, diversity and change in the 21st century (pp. 57–66). San Francisco: UCLA Asian American Studies Center and AsianWeek.
THE SOCIAL UNDERPINNINGS OF TRUST IN PHYSICIANS Marc A. Musick and Meredith G. F. Worthen ABSTRACT The past few years have seen the emergence of a research literature dedicated to defining and understanding trust in physicians. Much of this research, however, focuses on a narrow set of explanations for the generation of physician trust. The purpose of this chapter is to expand on research by introducing new ideas to the study of physician trust. Employing data from the 1998 General Social Survey, the chapter shows that social resources, vulnerability in finances and in perceptions about the end of life, and exposure to unstable environments all are fairly consistent predictors of physician trust.
As the size of the older adult population in the United States continues to grow, so too will the need for health care and its attendant costs. There seems to be no easy solution for derailing the health care implications of this demographic certainty. Yet, researchers and clinicians should and are looking for ways to help alleviate the likely enormous burdens our society will face. One possible route to preventing these enormous costs is the most obvious one: keep people healthy as long as possible. Research has shown that a variety of controllable factors, such as diet, exercise, and smoking Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 97–123 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00005-1
97
98
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
contribute to the health and well-being of the population. It also seems to be the case that the ways that individuals interact with the health care system itself can have an effect on health or at least on the maintenance of health regimens. Individuals can and do interact with the health care system at multiple levels. They take on multiple roles, such as patients or family members of those with an illness. They come in contact with multiple types of players in the health care setting, such as nurses, doctors, and pharmacists. And they interact with the system via economic transactions, whether costs are borne out of pocket or through insurance coverage. For many researchers, however, the fundamental relationship that exists in the health care system is that between the physician and the patient. As a number of authors have argued, qualities of this relationship are important for maintaining health through adherence to treatment regimens and the willingness to seek preventative or palliative care when needed (Hall, Dugan, Zheng, & Mishra, 2001; Leisen & Hyman, 2001). The purpose of this chapter is to examine one critical aspect of the patient–physician relationship: trust. Although a variety of studies have discussed and examined the issue of trust of physicians (e.g., Mechanic, 1998), we still know little of what predicts trust outside of the clinical setting. Yet, knowledge of the social factors that generate trust may be crucial for understanding inequalities in health care access and outcomes. Abundant research has documented connections between socioeconomic status and health showing that, in general, persons with lower levels of education or income are more likely to engage in unhealthy behaviors and are more likely to be in worse health overall (Lantz et al., 1998). It is possible that socioeconomic differences in adherence to treatment regimens or attitudes about care are linked to the trust of physicians, or the possible lack of trust of physicians by certain segments of the population. As such, to fully understand why it is that persons of certain status levels have difficulty interacting with the health services sector, we must first understand the quality of their interaction with physicians and their attitudes towards physicians in general. In the first part of this chapter we will review the literature on patient– physician trust and note the limitations that exist in that literature. Next we go beyond previous literature by suggesting a variety of social factors that are conducive for building trust in physicians. We test our assertions using data on physician trust from the General Social Survey. We conclude with a discussion of the limitations of our own research and suggestions for future research.
The Social Underpinnings of Trust in Physicians
99
LITERATURE REVIEW AND THEORY Previous Research on Physician Trust Numerous authors have noted the recent rise in academic research on the patient–physician trust relationship. These authors have argued that the apparent erosion of trust, most notably due to the rise of managed care and health maintenance organization (HMO)-based medicine (Emanuel, 1993; Mechanic & Schelsinger, 1996; Gray, 1997), has been the primary motivator of this research. Yet, as Pearson and Raeke (2000) note, little empirical research has examined other motivators of trust of physicians. They note, ‘‘y the paucity of empirical research on trust has provided little data pointing to clear correlates of patient-physician trust’’ (p. 511). In another recent work, Thom, Ribisl, Stewart, Luke, and The Stanford Trust Study Physicians (1999) note that at the time of conducting their study, they had found only one published work that contained a quantified measure of patient trust. In short, although there has been an apparent increase in the level of interest regarding patient–physician trust, empirical research on the issue has lagged far behind.
Definitions of Physician Trust Before reviewing the small amount of empirical research on patient– physician trust, it will be useful to state more clearly what we mean by trust. Various authors have provided definitions of trust in this context, and though most definitions differ slightly, most contain two central elements. First, as Mechanic (1996) noted, trust involves some notion that the other entity which is to be trusted will act in a manner that is in accord with the needs and desires of the ego. He further argued that trust exists at two levels. The first, interpersonal trust, involved relationships between individuals wherein trust is a part of the relationship. In this form, trust is built over time through the perception that actions are in accord with expectations. The second form, social trust, focused more on common understandings about how institutions should behave according to normative standards. Trust in this fashion is given when those individuals and institutions do tend to behave in the accepted ways. The second major facet of trust was well-explained by Hall et al. (2001). These authors noted that trust must involve vulnerability. That is, trust is only possible when some ego is in a position of vulnerability related to some
100
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
other entity. Where no vulnerability exists, trust is not possible. Somewhat paradoxically, the authors further explained that greater levels of vulnerability should lead to greater trust due to the heightened need created by that vulnerability. All persons are vulnerable to health problems, yet not everyone necessarily relies on the health care system. Though, for the purposes of this chapter, we assume that our respondents, who are adults living in the United States, will rely primarily on the health care system for health problems, and as such, will have to make decisions about trust of physicians and the health care system in general. Given that our definition assumes a minimum level of vulnerability, the definition we use in this chapter focuses more on the behavior of other entities and whether it is in accord with expectations and accepted standards. Moreover, we employ Mechanic’s notion of thinking about trust as existing at both interpersonal and social levels. It might be the case that individuals have great trust in their own physicians but less in the system or physicians in general. It should be noted that these authors and others (Mechanic, 1998; Hall et al., 2001) have also argued that trust in physicians is multi-dimensional. That is, trust exists not only at two levels but in a variety of different ways. In a later work, Hall, Camacho, Dugan, and Balkrishnan (2002a) listed these areas as: global trust, competence, fidelity, honesty, and confidentiality. This list is similar to that proposed by Mechanic (1998) but differs due to the inclusion of an element of control on the part of physicians to effectively manage care. As Hall et al. (2002a) noted, these dimensions arose from qualitative research on physician trust and seem in accord with the ways that individuals perceive and interact with physicians. Yet in their work and others, analyses on indices meant to tap these different dimensions have revealed a unidimensional factor structure. In short, while this multidimensionality may exist in the minds of respondents, it does not appear in statistical analyses. Consequently, in our own work we focus more on the interpersonal/social distinction rather than the dimensions discussed by these authors.
What Predicts Trust? As noted, little is known about the factors that predict trust of physicians. This lack of knowledge appears to stem from two sources. First, a real consensus determining the best ways to measure physician trust has not been achieved. Various authors (Hall et al., 2002a, 2002b; Anderson & Dedrick, 1990;
The Social Underpinnings of Trust in Physicians
101
Leisen & Hyman, 2001) have proposed measures of physician trust, but none of them have been adopted as an accepted standard. Second, as Pearson and Raeke (2000) have argued, there is simply too little empirical data and research available to determine what factors are conducive to physician trust. Generally, extant research on the topic focuses its efforts on three contexts: (1) clinical/doctor-specific; (2) insurance availability, choice, and forms of payment; and (3) other factors, usually sociodemographic. Clinical Research in the clinical setting has primarily focused on the relationship between physicians and patients. For example, some studies have examined the length of the patient–physician relationship and have found that longer relationships are associated with more trust (Kao, Green, Davis, Koplan, & Cleary, 1998a; Kao, Green, Zaslavsky, Koplan, & Cleary, 1998b; Thom et al., 1999). Another study (Thom, 2000) employed a physician training intervention to determine if levels of patient trust could be increased based on that training. The author found that the intervention had no effect on trust and further noted that little other research exists on the subject. These findings revealed that physician skill training has no effect on trust outcomes. Other researchers (Meit, Williams, Mencken, & Yasek, 1997) considered whether having patients wear a gown in the clinical setting effected trust; they found that gowning status had no effect on trust levels. Insurance, Choice, and Payment Types A variety of studies have examined factors related to health care plans and insurance. In general, research has shown that a greater choice of physicians or the ability to select physicians is associated with more trust (Kao et al., 1998a; Thom et al., 1999). Not surprisingly, those with health insurance were more trusting than those without (Doescher, Saver, Franks, & Fiscella, 2000). But the type of insurance appears to have a minor effect on trust as well: private non-HMO patients and those with Medicare tend to be more trusting than individuals with other types of insurance (Doescher et al., 2000; Reschovsky, Kemper, & Tu, 2000). Other Factors Of these other factors, the one that has received a great deal of attention is race. Two studies (Doescher et al., 2000; Boulware, Cooper, Ratner, LaVeist, & Powe, 2003) focusing on race found that non-Hispanic Blacks tended to report lower levels of physician trust compared to non-Hispanic Whites. These studies argued that due to a history of unfair treatment by the
102
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
health care system, most notably the Tuskegee Syphilis Study, Blacks should be less trusting of the system and doctors in general. A second factor considered by these authors and others was age: studies show that in general, older adults tended to be more trusting than younger ones (Thom et al., 1999; Doescher et al., 2000). In terms of education and income, studies either found that higher levels of each were associated with more trust or that there was no effect of these on trust (Kao et al., 1998b; Thom et al., 1999; Doescher et al., 2000). In terms of health status, one study (Doescher et al., 2000) showed that healthy respondents report the highest levels of trust while another showed no relationship between health and trust (Kao et al., 1998a). Kao et al. (1998a) showed that certain personality types and beliefs also had no effect on trust. Other than the factors noted above, the research literature revealed little about the factors that predict trust. However, one recent study deserves special mention because it used part of the data used in this study. In this previous study, Pescosolido, Tuch, and Martin (2001) used eight questions on physician trust to develop two indices of general trust in physicians. Their results indicated that for positively worded questions about physician trust, significant predictors are patient gender, race, age, income, education, marital status, and urbanization. They also found that self-rated health and the availability of insurance were significant predictors of both of their measures of trust.
Factors Contributing to Physician Trust Based on this review of the literature, it appears that we know little about what generates physician trust due to the scarcity of studies on the topic. Of those few studies that do exist, most examine only a small set of predictors and often overlap a great deal with one another. This chapter employs many of those same predictors to help solidify the literature in those areas. For example, like others, we employ an array of sociodemographic factors such as gender, race, age, and socioeconomic status. Like Pescosolido and her colleagues (2001) we also employ measures of marriage and urbanicity. Other factors in our model include self-rated health, insurance coverage, the ability to choose a physician, and confidence in medicine in general– factors that overlap with other studies in the literature. However, we also include a variety of other factors that have not been used in previous studies on this topic. Following is a brief listing of these additional factors and a discussion of our expectations regarding the behaviors of these factors.
The Social Underpinnings of Trust in Physicians
103
Social Resources Previous research has shown that social resources, such as social integration and support, are conducive for better health and well-being (House, Landis, & Umberson, 1988). Here we consider the possibility that access to different forms of social resources is beneficial for creating trust in physicians. Generally, the maintenance of successful social relationships requires trust. Although the trust that is built within relationships is specific to those relationships, the process of building that trust can lead to skills or attitudes that encourage trust in other settings (Stolle, 2001). Consequently, we should expect that those who are more socially integrated should be more trustful in general, including more trustful of physicians. Social resources can serve a second function in the generation of physician trust. Research has shown that social support can act as a buffer to the deleterious effects of stress on health (House et al., 1988). Support behaves in this fashion because it provides instrumental and emotional goods that are useful for overcoming the consequences associated with stressful events. Further, social support has also been evident in studies that show ‘‘having social ties to other individuals or groups results in better psychological and physical health’’ (Antonucci, Fuhrer, & Dartigues, 1997; Berkman, 1995; Blazer, 1982). As Hall and his colleagues (2001) have argued, an essential component of trust is vulnerability. Although social resources can never eradicate the vulnerability associated with the health care system, social resources can give individuals more confidence in those settings and can provide the resources, whether they be instrumental, cognitive, or emotional, needed to better navigate the health care system. In sum, it is reasonable to expect that people who are more socially integrated and have more access to social resources will be the most trusting of physicians.
Unstable Environments On the other hand, there are certain forms of social interaction that are likely detrimental for trust in general, including trust of physicians (Thiede, 2005). There are contexts in society in which trust is not necessary, and in some cases is harmful, for successful interaction. For example, in situations where people seem willing to take advantage of one another or where motives are easy to hide and difficult to discern, trust is not beneficial and may even be harmful. Likewise, frequent interactions in contexts that are not built on stable relationships are likely to create attitudes about others
104
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
that carry over from those settings and mold attitudes about others and society in general. In short, frequent interaction in contexts where stable long-term relationships are not possible is likely to breed distrust in general, including distrust of physicians. Knowledge Intelligence or the ability to employ knowledge in a specific setting or interaction appears important for the generation of trust in general (Yamagishi, 2001). Greater intelligence and social intelligence in particular both allow individuals to correctly interpret situational cues and evidence. These more accurate appraisals in turn bolster trust because they provide the sense that if some other entity were to act in a way that goes against the needs or desires of the ego, that behavior would be recognized as untrustworthy. Likewise, those with little situational knowledge or intelligence are less able to discern important clues and as a result, make reasonable appraisals of trustworthiness of alters. When in a vulnerable situation, those low on knowledge and intelligence will be unwilling to provide as much trust due to the extra vulnerability it imposes. Therefore, we expect that those with more intelligence and knowledge, especially of the health care system, should be more trusting of physicians.
Vulnerability Even though vulnerability is a necessary condition for trust, it does not necessarily follow that those who are more vulnerable will be more trusting. Rather, it is possible that higher levels of objective and perceived vulnerability should be related to less trust of physicians. This relationship should be so for two reasons. First, persons who feel more vulnerable will also feel that they have more to lose by placing trust in others within that setting. Even though they may be forced into relying on other entities, due to health concerns in this instance, those who have more perceived vulnerability will need to be more vigilant and more protective, thus less trusting, to prevent their perceived fears from becoming reality. Second, those who are vulnerable, especially in financial terms, have little to fall back on if trust is broached. Knowing this to be the case, those who are more vulnerable will be able to extend less trust as doing so heightens the harm if the trust was misplaced.
The Social Underpinnings of Trust in Physicians
105
METHODS Data Data for this study came from the General Social Survey (GSS) distributed by Inter-university Consortium for Political and Social Research (Davis, Smith, & Marsden, 2003). The survey, which has been collected on a yearly basis from 1972 to 1994 (with the exception of years 1979, 1981, and 1992), has been used in numerous previous studies and has been shown to be both reliable and representative of the adult population of the United States. The GSS is an independently drawn sample of persons eighteen years or older who are English speaking and live in the United States in noninstitutionalized settings. We use data from 1998 which contains an extensive array of questions on trust in physicians. Although the total sample size for that year was 2,832, our analyses use a smaller subset of respondents who were asked the physician trust questions. Missing cases for predictors were imputed using the mean value for ordinal and continuous variables and the mode for dichotomous ones; however, no imputation was made for the physician trust variables.
MEASUREMENT Trust in Physicians The GSS contains 20 items measuring trust in physicians. To our knowledge, only the Pescosolido et al. (2001) article has used these variables in a study predicting trust of physicians. In that research, the authors used a small subset of the variables and created two indices based on positive and negative wordings of the items. Since we conceptualized trust in physicians as existing both at the interpersonal and societal level, we chose to incorporate all the physician trust items rather than just those focused on by Pescosolido et al. (2001). However, like those authors, we also felt that negative and positive wording differences did exist within the list of items. Given that no research could be found that had examined how the items should be split into indices, we completed psychometric analyses of the items to determine the optimum index configuration. In the first step of this process we performed principal component factor analyses with oblique rotation to determine the optimal grouping. We found that four indices resulted based on positive/negative wordings and on
106
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
Table 1.
Confirmatory Factor Analysis of Physician Trust Items.
Personal physician-positive (PER-POS) My doctor is willing to refer me to a specialist when needed. I doubt that my doctor really cares about me as a person. I trust my doctor’s judgments about my medical care. I feel my doctor does not do everything she/he should for my medical care. I trust my doctor to put my medical needs above all other considerations when treating my medical problems. My doctor is a real expert when taking care of medical problems like mine. I trust my doctor to tell me if a mistake was made about my treatment. Physicians in general-positive (GEN-POS) Doctors always do their best to keep their patient from worrying. Doctors are very careful to check everything when examining their patients. Doctors always treat their patients with respect. Doctors never recommend surgery unless there is no other way to solve the problem. Doctors always avoid unnecessary patient expenses. Personal physician-negative (PER-NEG) I worry that my doctor is being prevented from telling me the full range of options for my treatment. I worry that I will be denied the treatment or services I need. I worry that my doctor will put cost considerations above the care I need. Physicians in general-negative (GEN-NEG) Doctors are not as thorough as they should be. Sometimes doctors take unnecessary risks in treating their patients. Doctors cause people to worry a lot because they do not explain medical problems to patients. The medical problems I have had in the past are ignored when I seek care for a new medical problem. w2/d.f. Root mean square error of approximation (RMSEA) Adjusted goodness of fit index (AGFI)
Mean
b
R2
3.95
.77
.43
3.62
.89
.58
3.85 3.52
1.00 .99
.73 .71
3.72
.98
.71
3.57
.96
.67
3.43
.91
.61
3.25
.65
.33
2.88
1.00
.77
3.19 3.08
.87 .64
.58 .32
2.61
.55
.24
3.43
.99
.74
3.44
.96
.71
3.42
1.00
.76
2.78 3.06
1.00 .76
.59 .34
3.00
.95
.54
3.53
.90
.48
425.94/146 .041 .98
The Social Underpinnings of Trust in Physicians
107
whether questions focused on a particular doctor or doctors in general. Because factor analysis of this sort is limited, we then went on to test that configuration in confirmatory factor analysis (CFA) using the LISREL computer package. The results of the CFA are shown in Table 1 along with the wordings of all the items and their mean scores. The analyses show that the physician trust items do indeed fall out along four indices as indicated above. Looking at the R2 scores shown in the table, it is readily apparent that the indices for personal physicians are better at explaining their underlying items than the indices for physicians in general. Because the RMSEA is below .05 and the AGFI is over .95, it is also clear that the model fits the data well. The indices that are used in the remainder of the analyses were created by taking the mean score for the items making up a particular grouping. Regardless of the name employed for each index, all are coded such that higher scores indicate more trust.
Predictors Social Resources We employ two measures of social resources. The first, service attendance, indicates how often respondents attend religious services. Responses range from (1) never to (9) several times a week. The second measure, social interaction, is an index of three items that indicate how often respondents spend an evening with (a) neighbors, (b) friends outside of the neighborhood, and (c) relatives. Higher scores on the index indicate more frequent interaction. Unstable Environments As we argued above, frequent social interaction in contexts where stable relationships are not possible or not sought might lead to more distrust. One setting that seems ripe for these types of atmospheres includes bars, clubs, and taverns. Consequently, our measure of interaction, carousing, is an indicator of how often respondents go to a bar, club, or tavern. Responses range from (1) never to (7) almost every day. Knowledge Here we include two items: one indicating knowledge in general and the other indicating knowledge relating to health and health care. First, verbal ability, is measured by having respondents identify synonyms for ten words
108
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
and then adding up the number of correct responses. The second, knowledge of illness, is a dichotomous variable coded (1) for respondents who have had a life-threatening illness or who have had a close friend or relative who have had a life-threatening or terminal illness. Respondents who had no interaction with illness in these ways were coded zero. Vulnerability We use three measures of vulnerability, two of which are in the context of the health setting concerning end of life (EOL). The first, EOL pain, is based on a question that asks respondents whether they believe doctors will be able to control their pain at the end of life. Responses range from (1) strongly to (5) strongly disagree that doctors will be able to control their pain. The second measure, EOL vulnerability (a=.64), is a three-item index reflecting the vulnerability that respondents feel about their end of life experiences. For each item respondents were asked whether they strongly agreed, agreed, disagreed, or strongly disagreed. The items are: (a) I worry about the economic burden that a terminal illness might cause my family; (b) I worry that if I run out of money or health insurance I will get second class health care; and (c) I worry about the emotional burden that my family might face making decisions for me at the end of life. The index was coded so that higher scores reflect a greater sense of vulnerability. Our third measure of vulnerability, finances, indicates how satisfied respondents are with their financial situations. Responses range from (1) not satisfied at all to (3) pretty well satisfied. Self-Rated Health Respondents were asked to rate their health. Responses range from (1) poor to (4) excellent. Health Care Flexibility This indicator is a typology composed of two dichotomous measures. The first of these asks respondents if they must use a book or directory to choose a doctor. The second asks whether respondents can choose any doctor they want. The variables representing the typology are coded as follows. No care flexibility indicates people who said that they had to use a book or directory and could not pick any doctor they wanted. Some care flexibility reflects people who do not have to use a directory or who can choose any doctor but not both. Most care flexibility refers to respondents who do not have to use a directory and can choose any doctor they want. All three of these variables are coded 0–1 with 1 indicating the category mentioned. The reference
The Social Underpinnings of Trust in Physicians
109
category for all three indicators are those respondents who were not asked the questions, i.e., those without health insurance. Put another way, to be asked those two questions, respondents had to have health insurance; consequently, the omitted category for all three variables is respondents without health insurance. Confidence in Medicine Respondents were asked how much confidence they have in medicine. Responses range from (1) hardly any to (3) a great deal.
Generalized Trust It is possible that trust in physicians is simply an outgrowth of generalized trust and has no special features unto itself. Consequently, to more accurately gauge the effects of the items listed above, trust in a general sense should be controlled. We use two indicators to measure generalized trust. First, trusting, is an indicator of how trusting respondents perceive themselves to be. Responses range from (1) very distrustful to (4) very trusting. The second indicator, others trustworthy (a=.71) is an index composed of two items: (a) whether people are fair and (b) whether people are helpful. The index is scored so that higher values indicate more trust in others to behave in fair and helpful ways.
Sociodemographics Finally, we include a number of sociodemographic predictors that overlap with those used in previous studies. These include gender (0=male, 1=female), race (0=White, 1=Non-white), age (years), education (years completed), income (levels ranging from [1] under $1,000 to [23] $110,000 or over), missing income (0=reported income, 1=did not report income), working full-time (0=not working full-time, 1=working full-time), marital status (0=not married, 1=married), and city size (categories ranging from [1] open country to [10] large central city). A variable for missing income was included due to the large number of missing cases on the income variable. As noted previously, we imputed mean values for all missing cases on predictor variables. To ensure that this imputation did not affect the estimate for income, we included the missing data variable to cancel the potential biasing effect of the imputation.
110
Table 2.
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
Descriptive Statistics of and Correlations between Physician Trust Indices. N
GEN-POS GEN-NEG PER-POS PER-NEG
1373 1374 1364 1362
Range
1–5 1–5 1–5 1–5
Correlations
Mean
3.01 3.08 3.66 3.42
GEN-POS
GEN-NEG
PER-POS
1.00 .47 .42 .28
– 1.00 .48 .51
– – 1.00 .54
All correlations are significant at po.001.
RESULTS Table 2 shows the univariate statistics of the four trust indices along with the correlations between them. The literature in this area suggests that people tend to be more trusting of their own doctors than they are of doctors in general. The means in Table 2 support this contention. For example, the mean for physicians in general-positive wording (GEN-POS) was 3.01 compared to 3.66 for personal physicians-positive wording (PER-POS). All of the correlations between the trust indices are positive, strong, and significant. Of the correlations, the two furthest apart conceptually, physicians in general-positive wording and personal physician-negative wording have the lowest correlation of .28. Table 3 shows the univariate statistics for all the predictor variables along with the zero-order associations between the predictors and each of the trust indices. These estimates reveal some findings that are in accord with our expectations and others that are not. However, rather than provide them for discussion, we have done so to provide some baseline comparison for the next table of adjusted estimates. Table 4 reports unstandardized coefficients for the regression of the four physician trust indices on the predictors shown in Table 3. It is important to note that we have excluded three variables from this table: confidence in medicine, social interaction, and carousing. We have excluded those variables due to their location in the survey. That is, the GSS does not ask all questions of all respondents. Instead, it employs a design that rotates questions and asks some of one portion of respondents but not for others. Due to this design, the indicated variables were not asked of all respondents in these analyses. Consequently, we have included them in different models as shown in the next table.
The Social Underpinnings of Trust in Physicians
Table 3.
Descriptive Statistics of Predictors and Zero-Order Associations with Physician Trust Indices. Range
Zero-Order Associationsa
Mean GEN-POS
Sociodemographics Female Non-white Age Education Income Missing income Working full-time Married City size Health-related Self-rated health No care flexibility Some care flexibility Most care flexibility Resources Service attendance Verbal ability Knowledge of illness Vulnerability EOL vulnerability EOL pain Finances Generalized trust Trusting Others trustworthy Reduced sample factors Confidence in medicine Social interaction Carousing a
111
GEN-NEG
PER-POS
PER-NEG
.58 .21 45.47 13.27 15.42 .11 .55 .48 6.97
.08 .15 .00 .03 .02 .05 .19 .01 .01+
.03+ .06 .00 .04 .01 .08 .05 .06 .02
.05 .08 .01 .00 .00+ .00 .11 .09 .01
.02 .06 .01 .02 .02 .02 .09 .12 .01
1–4 0–1 0–1 0–1
3.05 .23 .32 .31
.03 .10 .08 .16
.10 .14 .21 .31
.04 .07 .17 .33
.09 .17 .33 .65
1–9 0–10 0–1
4.66 6.12 .65
.02 .07 .15
.02 .05 .03
.04 .00 .03
.03 .04 .00
1–5 1–5 1–3
3.45 2.52 2.07
.10 .13 .07
.14 .11 .14
.10 .11 .12
.25 .11 .24
1–4 0–1
3.44 .35
.13 .07
.00 .21
.12 .11
.10 .25
1–3 1–7 1–7
2.34 3.98 2.37
.16 .03+ .05
.22 .00 .03
.17 .04 .05
.20 .02 .05
0–1 0–1 18–89 0–20 1–23 0–1 0–1 0–1 1–10
Unstandardized OLS regression coefficients are reported. All estimates are unadjusted.
po.05; po.01; po.001.
For each outcome in Table 4, we have included two models. The first model includes all variables except generalized trust while the second includes those as well. The purpose of this arrangement was to determine whether generalized trust mediates the effects of the other factors. By setting up the models in this step-wise fashion, we can determine whether they do indeed act as mediators.
112
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
Table 4.
Estimated Net Effects of Health and Other Predictors on Physician Trusta. GEN-POS
GEN-NEG
PER-POS
PER-NEG
Model Model Model Model Model Model Model Model 1 2 1 2 1 2 1 2 Sociodemographics Female Non-white Age Education Income Missing income Working full-time Married City size Health-related Self-rated health No care flexibility Some care flexibility Most care flexibility Resources Service attendance Verbal ability Knowledge of illness Vulnerability EOL vulnerability EOL pain Finances Generalized trust Trusting Others trustworthy Intercept Adjusted R2 a
.11 .09+ .00+ .01 .02 .01 .10 .05 .02
.11 .10 .00 .01 .02 .01 .10 .05 .21
.03 .02 .00 .02 .00 .12+ .07 .04 .02
.03 .02 .00 .02 .00 .11+ .07 .04 .26
.02 .02 .00 .00 .00 .07 .04 .04 .00
.02 .01 .00 .00 .00 .06 .05 .04 .00
.03 .12+ .00 .01 .00 .11 .03 .06 .00
.03 .12 .00+ .00 .00 .11 .03 .06 .00
.02 .13 .12 .15
.02 .13 .13 .15
.06 .07 .14 .21
.06 .06 .14 .20
.05 .03 .12 .21
.05 .04 .13 .21
.08 .07 .22 .44
.08 .07 .23 .44
.00 .03 .06
.02 .01 .01
.02 .00 .01
.00 .02+ .03
.00 .02+ .03
.01 .01 .00 .05 .04 .03 .11 .11 .06
.04 .04 .10 .10 .06 .06 .19 .19 .01 .11 .08 .07 .07 .07 .05 .05 .06 .05 .07 .07 .05 .04 .12 .12 – – 3.68 .12
.08 – .06 – 3.42 2.64 .13 .09
.02 – .12 – 2.71 3.56 .09 .09
.08 – .02 – 3.27 3.07 .10 .15
.06+ .08 2.87 .15
Unstandardized OLS regression coefficients are reported.
po.05; po.01; po.001.
As suggested in previous research, the sociodemographic predictors have varying effects on trust. For GEN-POS, we found that women were less trusting, non-whites were more trusting, people with higher levels of income were less trusting, full-time workers were less trusting, and people who lived in larger cities were more trusting. Age, education, and city size exerted
The Social Underpinnings of Trust in Physicians
113
significant effects in other models; otherwise the sociodemographic variables contributed little to explaining trust. The health-related factors appeared to have a more pervasive effect on trust. For three of the outcomes, self-rated health exerted a positive and significant effect on trust. More importantly, flexibility of health care options had significant effects on all trust outcomes. In three of those cases, the more flexibility respondents had, the more trusting they were of physicians. Recall that the reference category for these three items is composed of respondents without insurance. Consequently, the estimates of no care flexibility for GEN-NEG, PER-POS, and PER-NEG indicate that having insurance but no flexibility in the choice of doctors is no better than having no insurance in terms of generating trust. In contrast, having insurance and a great deal of flexibility tends to greatly improve levels of trust. The only measure of social resources used in these models, service attendance, shows mixed results. For one outcome, PER-POS, higher levels of service attendance are associated with more trust, but for all other outcomes, attendance has no effect. Likewise, the knowledge variables show mixed results. For both, higher levels are indicative of lower levels of trust as measured by GEN-POS. However, verbal ability, like education, is a positive predictor of trust in the GEN-NEG model. In contrast, to the mixed findings for social resources and knowledge, we find that the vulnerability indicators have the expected effects for all outcomes. That is, for the EOL vulnerability variables, a greater sense of vulnerability leads to lower levels of trust for all types. The finances variable shows that people who are more satisfied with their finances, and thus less vulnerable, report higher levels of trust along all dimensions. Of all of the factors included in the model, this set, along with health care flexibility, are those most consistently related to physician trust. Finally, as was the case for many of the other variables in the model, the estimates for generalized trust reveal mixed findings. In several cases, higher levels of trust were associated with more trust in physicians. However, in more cases these trust items had a marginal or no effect on the trust outcomes. Further, comparing the two models for each outcome, we find no evidence that these general trust items serve as mediators of the other predictors. Rather, their effects, where they exist, appear independent of the other predictors. In Table 5 we report the results for the three items we were unable to include in Table 4. Note that although only the variables in question are shown, these estimates are adjusted for all of the variables shown in Table 4. In terms of confidence in medicine, we find that respondents who have a
114
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
Table 5.
Estimated Net Effects of Reduced Sample Predictors on Physician Trusta. GEN-POS
Confidence in medicine Social interaction Carousing Intercept Adjusted R2 N
GEN-NEG
PER-POS
PER-NEG
Model 1
Model 2
Model 1
Model 2
Model 1
Model 2
Model 1
Model 2
.12
–
.19
–
.15
–
.17
–
– – 3.31 .15 929
.03 – .03 – 3.26 2.24 .11 .14 916 931
.01 .02 2.73 .07 915
– – 3.07 .12 923
.05 – .04 – 3.10 2.57 .10 .17 910 922
.00 .03 2.92 .12 909
a
Unstandardized OLS regression coefficients are reported. Estimates are adjusted for all the variables shown in Table 3. po.05; po.01; po.001.
general confidence in medicine also report higher trust for all four outcomes. In contrast, the social interaction variable shows mixed results. That is, in most cases social interaction has no effect on trust except for the PER-POS outcome. In this instance, higher levels of social interaction are associated with greater trust. The results for the carousing variable are also mixed but tend to be in the direction expected. That is, for GEN-POS and PER-POS, respondents who went to bars, clubs, and taverns more frequently reported less trust.
DISCUSSION Summary: Predictions and Results The physician–patient relationship is an important facet of the health care system and is deserving of careful consideration. Physicians occupy a unique status in the physician–patient relationship because they often have power over much of the information on which patients rely as well as other aspects of the care and decision-making process (Barber, 1983). Due to this power over important information, trust levels of physicians are issues of concern. As a result of a distinctive role in a highly trusted
The Social Underpinnings of Trust in Physicians
115
profession, the general public is wary to trust physicians despite the fact that many reports reveal competence of physicians as higher than most other professions (Barber, 1983). The number of malpractice suits is reaching a level some deem as ‘‘epidemic’’ which constantly calls into account the trustworthiness of physicians (Barber, 1983; Craige, 2004). More recently, an increase in malpractice claims has appeared along with a ‘‘rapid escalation in medico-legal activities to redress medical errors, negligence, risk, and harm’’ (Marjoribanks, Good, Lawthers, & Peterson, 1996, p. 164). Trust is also a significant feature of the physician–patient interaction since studies show those who are distrusting are not likely to seek health care (Nickerson, Helms, & Terrell, 1994). Thus, trust is an important facet of physician–patient relationship and is deserving of more examination. Indeed, an assessment of trust in the medical setting may also help us understand inequalities in health care access, quality, and outcomes. Our research has documented that numerous factors related to socioeconomic status, such as verbal ability, health status, and vulnerability in finances, are predictive of levels of physician trust. Given that trust is important for establishing effective doctor–patient communication and for ensuring that treatment regimens are met, understanding that these socioeconomic-related factors lead to trust provides some insight into why those inequalities exist. Thus, any attempts at reducing inequalities in access and quality must address issues of trust and the factors that lead to distrust of physicians. Our findings provided other insights into predictors of trusting relationships. We expected that some factors leading to trusting relationships might include health status, health care flexibility, vulnerability, confidence in medicine, and unstable environments. We found some support for our expectations which we will discuss in the remainder of this section.
Health Status Persons who rate themselves with a high health status were more trusting than those who rated themselves with lower health status levels. Our findings were consistent with another study showing that high levels of trust are often indicative of good health (Barefoot et al., 1998). A similar relationship may have appeared in our research but due to the crosssectional nature of the data we were unable to discern it.
116
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
Flexibility of Health Care Options Our study further showed that greater levels of flexibility in choosing health care options contributed to a greater amount of trust. Similarly, Kao et al. (1998a) found that patients who were able to choose their physicians reported more physician trust. The choice to remain with the same physician also played a role in generation of trust levels in a study comparing continuity of care in the United Kingdom and United States. The results of this study revealed that trust in physicians is often present in those patients who see the same physician regularly (Mainous III, Baker, Love, Gray, & Gill, 2001). In the studies mentioned, as well as in our study, the trust of one’s physician seems to be positively related to the flexibility of health care options.
Vulnerability We examined three types of vulnerability as predictors of trust: vulnerability in finances, end of life vulnerability, and end of life pain. Those who reported satisfaction in their finances reported higher levels of trust along all dimensions. We chose to inspect this variable because we perceived vulnerability as a necessity in the creation of trust (Hall et al., 2001). Second, we focused on end of life vulnerability and found that an increased level of vulnerability that respondents felt about their end of life experiences decreased trust. Third, we examined end of life pain vulnerability and found that persons who felt that doctors might not be able to control the pain at the end of their life were also more likely to distrust. Hall et al. (2001) explained that greater levels of vulnerability should lead to greater trust due to the heightened need created by that vulnerability. Our findings, however, were exactly the opposite. With such contradictory findings, further research in this area is needed.
Knowledge, Confidence, Unstable Environments Other predictors of trusting relationships might include general knowledge, knowledge of illness, general confidence in medicine, and exposure to unstable environments. Our findings showed that general knowledge, in terms of verbal ability and education level, appeared to be a positive predictor of trust. The dimensions of this aspect of trust could be a result of
The Social Underpinnings of Trust in Physicians
117
physician–patient status consistency, that is to say that more educated persons are more similar to physicians in education level and feel more able to trust physicians while uneducated persons view physicians as untrustworthy persons due to their higher status. A study attempting to understand the development of physician attributes revealed that these status inconsistencies are often neutralized (Beagan, 2000). The author found that medical students did not view social class and education levels as significant in the practice of medicine (Beagan, 2000). The knowledge of illness variable produced variable results. Reasons for these variable results will be discussed further in the following sections. In terms of confidence in medicine, we found that respondents who had a general confidence in medicine also reported higher trust. Blendon and Benson (2001) examined Americans’ views on health policy and in a sense, their general confidence in medicine. They found that most Americans are not pleased with the health care system (Blendon & Benson, 2001). From this study and our findings, it might follow that the majority of Americans are not trusting since they are not pleased with the health care system. We found, however, that a significant amount of people are trusting. With these conflicting results, this area of research deserves more attention. For exposure to unstable environments, respondents who went to bars, clubs, and taverns more frequently reported less trust. It may be that persons in regular contact with environments that do not commonly house trusting relationships, such as bars, clubs, and taverns, do not feel trust is part of their daily repertoire and therefore feel less comfortable trusting persons. Little research on this area has been conducted and more investigation would provide a further understanding of our results.
Generalized Trust Our generalized trust findings were variable. Most instances revealed a minor or insignificant effect on trust levels. Persons who were more likely to be trusting in general were not necessarily more likely to trust physicians. Put another way, trust of one’s physician might be unrelated to generalized trust since our results revealed such variable findings. Little research has been conducted examining the unique components of trust in physicians when compared to trust in other entities. It has been suggested through Thom’s (2000) work that certain training methods unique to the medical staff might induce trust. This does imply that trusting relationships in the medical field require something qualitatively different that trusting
118
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
relationships with other entities. Overall it appears that the trust between physician and patient is distinct from generalized trust.
Other Factors (Race) We expected other predictors of trusting relationships to include race and age. Focusing on race, Doescher et al. (2000) and Boulware et al. (2003) determined that non-Hispanic Blacks reported lower levels of physician trust when compared to non-Hispanic Whites. Merrill and Allen (2003) also found that Hispanics were the least satisfied with physicians and overall health care when compared to Whites and Blacks. In contrast to Doescher et al. (2000) and Boulware et al. (2003), we found that non-whites were most trusting. Age had an effect on trust levels in the research of Thom et al. (1999) and Doescher et al. (2000). These authors found that older adults were often more trusting than younger adults. Our findings revealed little information about age and trust since our results were somewhat variable. Overall, we found that sociodemographic predictors had variable effects on trust while Merrill and Allen (2003) found that sociodemographic factors were related to patient satisfaction with physicians. With such mixed results, sociodemographics and trust deserved more attention.
Limitations This study, although supporting many of our expectations, faces several limitations. First, our measure of illness knowledge may be inadequate. The illness knowledge measure indicated respondents who have had a lifethreatening illness or who have had a close friend or relative who have had a life-threatening or terminal illness. It is possible that persons involved in the medical field that did not have an illness and were not close to any person with an illness could have been coded as having little knowledge of illness. Yet respondents involved in the medical field do have knowledge of illness which creates error in the measure. Second, due to the cross-sectional nature of the data, we were unable to test for reciprocal or reverse causation of the data. Third, the omnibus nature of the GSS did not allow us to look at other possible measurements of certain variables. For example, although there are many more measures of health, the GSS provides only self-rated health as a measure of health. Fourth, the split-ballot design also created some problems. In split-ballot design, some respondents are asked certain
The Social Underpinnings of Trust in Physicians
119
questions while other respondents are not asked these questions. As a result of this design, it is not possible to estimate the effects of all our measures at the same time and thus test and overall model predicting physician trust.
Future Research Further research in trust would be beneficial to understanding the components of the patient–physician relationship. A study on the adverse effects of distrust might increase knowledge about the enhancement of communication between physician and patient. One such examination focusing of the adverse effects of distrust in one’s physician found negative health outcomes appear when attitudes of mistrust, doubt, and cynicism are present (Smith & Gallo, 2001). Additional studies involving women and older people showed that distrust and suspiciousness negatively impact health status (Adams, 1994; Barefoot et al., 1987; Barefoot, Larsen, Lieth, & Schroll, 1995). Pennebaker (1990) found that a disclosure of significant pessimistic events created decreases in number of hospital visitations. A study taking a closer look at the decision to seek health care revealed that ‘‘people who are more trusting of health care providers are more likely to seek out medical care and remain under care once treatment is initiated’’ (Freburger, Callahan, Currey, & Anderson, 2003, p. 56). This study demonstrated the positive aspects that can result from a trusting relationship between physician and patient. Further studies aimed at increasing trust along these lines would create an outlet for understanding the physician–patient relationship. Another study also found that lower levels of trust often decrease the amount of health care sought (Nickerson et al., 1994). Trust relationships deserve further investigation since they often promote the search for medical care and the desire to remain under medical supervision. These studies demonstrate that distrust creates adverse effects in the physician–patient relationship while trust fosters and environment where patients seek out health care. In addition, a further look into a possible connection between specific illnesses and trust levels might be another outlet for research. One study examined physician trust in patients with rheumatic disease. The findings of this study showed that ‘‘patients with less trust in their rheumatologists were more likely to make independent decisions about their care’’ (Freburger et al., 2003, p. 55). Another study performed on highly distrustful persons found those respondents to have higher levels of cardiovascular arousal (Christensen & Smith, 1993). These findings create a much-needed area for future research. It may be the case that people with specific diseases create
120
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
trust. A future research design might examine a specific disease such as high obesity and diabetes and the trust levels individuals with such diseases possess. With the obesity rates reaching an epidemic rate in the United States, the search to find trust within patients with specific illnesses is a pressing issue (Simpson, 2003). Another outlet for future research would be in possible methods to increase trust. The implementation of methods that might encourage trust might create an environment that promotes trust between patient and physician. One study attempted to harness the components of trusting relationships through creating a ‘‘training’’ session for physicians in hopes to foster trust (Thom, 2000). Thom’s study ‘‘did not find any effect from one-day physician training intervention on physician-patient trust’’ (Thom, 2000, p. 251). Further research along similar lines, however, might yield some significant results. A possible research design might be an experiment where one group of doctors attends a seminar where patients describe to them the things that make them trust and distrust physicians. The other group would not attend the seminar and the trust levels of the two groups of physicians’ patients would be compared at the outset of the seminar. A further look into physician trust ‘‘training’’ methods deserves attention since the connection between trust and patients is so vital.
REFERENCES Adams, S. H. (1994). Role of hostility in women’s health during midlife: A longitudinal study. Health Psychology, 13, 488–495. Anderson, L. A., & Dedrick, R. F. (1990). Development of the trust in physician scale: A measure to assess interpersonal trust in patient–physician relationships. Psychological Reports, 67, 1091–1100. Antonucci, T. C., Fuhrer, R., & Dartigues, J. F. (1997). Social relations and depressive symptomatology in a sample of community-dwelling French older adults. Psychology of Aging, 12, 189–195. Barber, B. (1983). The logic and limits of trust. New Brunswick, NJ: Rutgers University Press. Barefoot, J. C., Larsen, S., Lieth, L., & Schroll, M. (1995). Hostility, incidence of acute myocardial infarction, and mortality in a sample of older Danish men and women. American Journal of Epidemiology, 142, 477–484. Barefoot, J. C., Maynard, K. E., Beckham, J. C., Brummett, B. H., Hooker, K., & Siegler, I. C. (1998). Trust, health, and longevity. Journal of Behavioral Medicine, 21, 517–526. Barefoot, J. C., Siegler, I. C., Nowlin, J. B., Peterson, B. L., Haney, T. L., & Williams, R. B. (1987). Suspiciousness, health, and mortality: A follow-up study of 500 older adults. Psychosomatic Medicine, 49, 450–457.
The Social Underpinnings of Trust in Physicians
121
Beagan, B. L. (2000). Neutralizing differences: Producing neutral doctors for (almost) neutral patients. Social Science & Medicine, 51, 1253–1265. Berkman, L. F. (1995). The role of social relations in health promotion. Psychosomatic Medicine, 57, 245–255. Blazer, G. (1982). Social support and mortality in an elderly community population. American Journal of Epidemiology, 115, 684–694. Blendon, R. J., & Benson, J. M. (2001). Americans’ views on health policy: A fifty-year historical perspective. Health Affairs, 20, 33–46. Boulware, L. E., Cooper, L., Ratner, L., LaViest, T., & Powe, N. (2003). Race and trust in the health care system. Public Health Reports, 118, 358–365. Christensen, A. J., & Smith, T. (1993). Cynical hostility and cardiovascular response during selfdisclosure. Psychosomatic Medicine, 58, 150–155. Craige, B. (2004). The medical malpractice ‘‘crisis’’: Myth and reality. The North Carolina State Bar Journal (Summer), 8–10. Davis, J. A., Smith, T. W., & Marsden, P. V. (2003). General social surveys, 1972–2002: (Cumulative file) (Computer file), 2nd ICPSR version. Chicago, IL: National Opinion Research Center (producer), 2003. Storrs, CT: Roper Center for Public Opinion Research, University of Connecticut/Ann Arbor, MI: Inter-university Consortium for Political and Social Research (distributors). Doescher, M. P., Saver, B. G., Franks, P., & Fiscella, K. (2000). Racial and ethnic disparities in perceptions of physician style and trust. Archives of Family Medicine, 9, 1156–1163. Emanuel, E. J. (1993). Managed competition and the patient–physician relationship. New England Journal of Medicine, 329, 879–882. Freburger, J. K., Callahan, L. F., Currey, S., & Anderson, L. (2003). Use of the trust in physician scale in patients with rheumatic disease: Psychometric properties and correlates of trust in the rheumatologist. Arthritis & Rheumatism (Arthritis Care & Research), 49, 51–58. Gray, B. H. (1997). Trust and trustworthy care in the managed care era. Health Affairs, 16, 34–49. Hall, M. A., Camacho, F., Dugan, E., & Balkrishnan, R. (2002a). Trust in the medical profession: Conceptual and measurement issues. Health Services Research, 37, 1419–1439. Hall, M. A., Dugan, E., Zheng, B., & Mishra, A. (2001). Trust in physicians and medical institutions: What is it, can it be measured, and does it matter? The Milbank Quarterly, 79, 613–640. Hall, M. A., Zheng, B., Dugan, E., Camacho, F., Kidd, K., Mishra, A., & Balkrishnan, R. (2002b). Measuring patients’ trust in their primary care providers. Medical Care Research and Review, 59, 293–318. House, J. S., Landis, K., & Umberson, D. (1988). Social relationships and health. Science, 241, 540–545. Kao, A. C., Green, D., Davis, N., Koplan, J., & Cleary, P. (1998a). Patients’ trust in their physicians: Effects of choice, continuity, and payment method. Journal of General Internal Medicine, 13, 681–686. Kao, A. C., Green, D. C., Zaslavsky, A., Koplan, J., & Cleary, P. (1998b). The relationship between method of physician payment and patient trust. Journal of the American Medical Association, 280, 1708–1714. Lantz, P., House, J., Lepkowski, J., Williams, D., Mero, R., & Chen, J. (1998). Socioeconomic factors, health behaviors, and mortality: Results from a nationally representative
122
MARC A. MUSICK AND MEREDITH G. F. WORTHEN
prospective study of US adults. Journal of the American Medical Association, 279(21), 1703–1708. Leisen, B., & Hyman, M. R. (2001). An improved scale for assessing patients’ trust in their physician. Health Marketing Quarterly, 19, 23–42. Mainous, A., III., Baker, R., Love, M., Gray, D., & Gill, J. (2001). Continuity of care and trust in one’s physician: Evidence from primary care in the United States and the United Kingdom. Family Medicine, 33, 22–27. Marjoribanks, T., Good, M., Lawthers, A., & Peterson, L. (1996). Physicians’ discourses on malpractice and the meaning of medical malpractice. Journal of Health & Social Behavior, 37, 163–178. Mechanic, D. (1996). Changing medical organization and the erosion of trust. Milbank Quarterly, 74, 171–189. Mechanic, D. (1998). The functions and limitations of trust in the provision of medical care. Journal of Health Politics, Policy, and Law, 23, 661–686. Mechanic, D., & Schelsinger, M. (1996). The impact of managed care on patients’ trust in medical care and their physicians. Journal of the American Medical Association, 275, 1693–1697. Meit, S. S., Williams, D., Mencken, F., & Yasek, V. (1997). Gowning: Effects on patient satisfaction. Journal of Family Practice, 45, 397–401. Merrill, R. M., & Allen, E. (2003). Racial and ethnic disparities in satisfaction with doctors and health providers in the United States. Ethnicity and Disease, 13, 492–498. Nickerson, K. J., Helms, J. E., & Terrell, F. (1994). Cultural mistrust, opinions about mental illness, and black students’ attitudes toward seeking psychological help from white counselors. Journal of Counseling Psychology, 41, 378–385. Pearson, S. D., & Raeke, L. (2000). Patients’ trust in physicians: Many theories, few measures, and little data. Journal of General Internal Medicine, 15, 509–513. Pennebaker, J. W. (1990). Opening up: The healing power of confiding in others. New York: Morrow. Pescosolido, B., Tuch, S., & Martin, J. (2001). The profession of medicine and the public: Examining Americans’ changing confidence in physician authority from the beginning of the ‘health care crisis’ to the era of health care reform. Journal of Health and Social Behavior, 42, 1–16. Reschovsky, J. D., Kemper, P., & Tu, H. (2000). Does type of health insurance affect health care use and assessments of care among the privately insured? Health Services Journal, 35, 219–237. Simpson, I. (2003). The disappearance of the family meal and the epidemic of obesity. Annual Meetings of the Southern Sociological Society. Smith, T. W., & Gallo, L. (2001). Personality traits as risk factors for physical illness. In: A. Baum, T. Revenson & J. Singer (Eds), Handbook of health psychology (pp. 139–174). Mahwah, NJ: Lawrence Erlbaum. Stolle, D. (2001). Getting to trust: An analysis of the importance of institutions, families, personal experiences and group membership. In: P. Dekker & E. M. Uslaner (Eds), Politics in everyday life: Social capital and participation (pp. 118–132). New York: Routledge. Thiede, M. (2005). Information and access to health care: Is there a role for trust? Social Science & Medicine, 61, 1452–1462.
The Social Underpinnings of Trust in Physicians
123
Thom, D. H. (2000). Training physicians to increase patient trust. Journal of Evaluation in Clinical Practice, 6, 245–253. Thom, D. H., Ribisl, K., Stewart, A., Luke, D., & The Stanford Trust Study Physicians. (1999). Further validation and reliability testing of the trust in physician scale. Medical Care, 37, 510–517. Yamagishi, T. (2001). Trust as a form of social intelligence. In: K. Cook (Ed.), Trust in society (pp. 121–147). New York: Russell Sage Foundation.
DISPARITIES IN MEDICAID SUPPORT FOR AND THE QUALITY OF NURSING FACILITY LONG-TERM CARE ACROSS THE STATES Jean Giles-Sims and Charles Lockhart ABSTRACT The ‘‘baby-boom’’ generation is poised for retirement. Yet the American states exhibit sharp inequalities in the public support they provide for nursing facility long-term care for the elderly, a form of health care that few Americans can afford to purchase privately. Further, remarkable disparities exist, both within and among states, in the quality of nursing facility care. We describe cross-state variation in Medicaid support for and the quality of nursing facility care, offer regression models that provisionally explain the sources of these inequalities, comment on the social implications of these disparities and recommend a solution.
Public policy support of long-term care for the elderly varies strikingly across the states. In 2001 for instance, Iowa had over four times more Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 125–148 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00006-3
125
126
JEAN GILES-SIMS AND CHARLES LOCKHART
Medicaid-certified nursing facility beds per 1,000 state residents 65 or older (91) than Nevada (22) (Gregory & Gibson, 2002). The proportion of North Dakota’s overall Medicaid expenditure that was devoted to nursing facility care (34%) was over three times Utah’s (9%) (U.S. Department of Health and Human Services (USDH&HS, 2003). Rhode Island’s Medicaid expenditure for nursing facility care per state resident 65 or older ($2,716) was virtually four times greater than Florida’s ($689) (USDH&HS, 2003). And, New York’s Medicaid nursing facility reimbursement rate (per resident day) ($143) was nearly twice that of Arkansas’ ($74) (Centers for Medicare and Medicaid (CMS) data courtesy of Janet Freeze). Additionally, these last two examples of cross-state variation have been corrected for differences in states’ cost of living (Berry, Fording, & Hanson, 2000, updated via ICPSR). Over the last decade, the federal government has repeatedly transferred responsibilities for joint federal-state public assistance programs such as Medicaid to state governments (Pear, 2005a, 2005b). These transfers maintain, possibly strengthen, long-standing constraints in some states on care options for the elderly that are readily available in others. Differences such as these are important for several reasons. First, they affect the degree to which elderly citizens are able to access appropriate forms of care. For instance, a state that certifies comparatively few Medicaid nursing facility beds per 1,000 elderly state residents is apt to leave more vulnerable elderly persons without appropriate care than a state that certifies more. These cross-state inequalities in Medicaid supported nursing facility care for the elderly offer a stark contrast with the uniformity of support for acute medical care provided by the federal Medicare program. Second, these inequalities represent a public problem since Medicaid is the only or primary payer of the long-term care expenses for two-thirds of our nation’s nursing facility population (Harrington, Carrillo, & Mercado-Scott, 2005). Although most recent expansions of Medicaid (including the closely related State Child Health Insurance Program) have involved pregnant women and children, Medicaid devotes more resources to the acute health and long-term care of the elderly (31%) than to acute medical and other care for children in low-income households (23%) (USDH&HS, 2000). Third, the state of residence has an impact on the quality of nursing facility care in addition to financial support and access issues. As we show below, sharp inequalities in public resources in different states contribute importantly to remarkable disparities in the quality of care elderly Americans experience in nursing facilities. While quality of care varies
Disparities in Medicaid Support
127
from facility to facility within states, state-level data on quality of care reveal surprising disparities in the quality of care among states. Overall, these cross-state inequalities suggest that the United States is poorly prepared to serve the long-term care needs of a rapidly growing elderly population. Several factors contribute to Medicaid paying for a large portion of nursing facility long-term care. Most persons lack the resources to pay for even a few months of extremely expensive nursing facility care out-ofpocket. Additionally, private long-term care insurance continues to be more expensive than many persons can afford and is often sold in limited amounts or for restricted contingencies that constrain the help it provides with regard to the varying circumstances that elderly policyholders actually encounter (Matthews, 2001). Moreover, while Medicare pays substantial portions of the acute medical care expenses for most of the elderly, it generally covers nursing facility costs only for short spells of skilled nursing care following hospitalization (currently about 12% of nursing facility residents). Consequently, except for the truly wealthy, most elderly persons whose chronic cognitive or physical conditions land them in nursing facilities either are initially or soon become dependent on the aspect of state-level public social policy that Medicaid-supported long-term care for the elderly represents. These factors combine to make Medicaid support of nursing facility longterm care for the elderly an unusual application of state-level public assistance policy. Generally, these programs serve limited portions of the socially disadvantaged. However, when Medicaid supports long-term care for the elderly, the recipients represent a broader cross-section of the American population. Most recipients have spent their lives as property owners, employees (even employers) and taxpayers, so their families resemble those of public officials more closely than do those of other public assistance beneficiaries. Thus states may respond to the public assistance needs of the elderly in distinctive ways. The liberal-conservative conflict over public support for these citizens may be muted as conservative officials empathize with recipients from families relatively similar to their own (Bailey & Rom, 2004). Further, state legislators and administrators as well as the general public favor the elderly as recipients of public assistance benefits over the impoverished non-elderly households that are the numerically predominant public assistance beneficiaries (Cook & Barrett, 1992). So we may find that different patterns of factors explain the cross-state disparities in Medicaid support of nursing facility long-term care for the elderly than account for cross-state differences in the generosity of other state-level public social programs.
128
JEAN GILES-SIMS AND CHARLES LOCKHART
PREVIOUS WORK Researchers working in three complementary, partially overlapping, yet distinctive areas of inquiry contribute relevant background to this study of cross-state variation in public social program resource support and quality of care. One approach focuses on using broad state characteristics to explain cross-state differences among public social programs. It offers the strength of explicit, rival explanatory models resting on variables that have long been central to social science explanations. Typically, the independent variables of this literature are employed to explain variation in the relative generosity of resources allocated. Major explanatory variables include alternative measures of state predisposition such as the ideologies of and competition among political elites (e.g., Barrilleaux, Holbrook, & Langer, 2002) and state political culture (e.g., Elazar, 1966); state capacities such as socioeconomic conditions (e.g., Sharkansky & Hofferbert, 1969) and capabilities of state-level officials (e.g., Schneider, 1993) or combinations of variables in these and other related categories (e.g., Grogan, 1999). A second approach examines patterns of policy diffusion among the American states. Initially this approach was largely separated from the state characteristics literature discussed above and emphasized diffusion via geographic proximity (e.g., Gray, 1973; Walker, 1969, 1973). Although this approach has increasingly drawn many of its variables from the state characteristics literature, its distinctive focus involves the timing of states’ decisions to adopt programs or program elements. These decisions are clearly influenced externally by the actions of other states and the national government as well as internally by municipal innovations (Shipan & Volden, 2004). Overall, this literature evaluates a hypothesis of cross-state policy convergence. Recent work shows that numerous state characteristics beyond geographic proximity foster inter-state policy diffusion, including professional associations (e.g., Balla, 2001); policy entrepreneurship (e.g., Mintrom, 1997); shared ideology (e.g., Grossback, Nicholson-Crotty, & Peterson, 2004); similar state circumstances (e.g., Case, Hines, & Rosen, 1993) and the relative success of innovations (e.g., Volden, 2006). Emulating the policy innovations of others often coexists with policy evolution through the addition of new policy increments in adoptive states (e.g., Hays, 1996). Moreover, patterns of legislative, judicial and administrative diffusion appear to differ (Berry, 1994; Canon & Baum, 1981). Further, some patterns of policy diffusion depend on states’ varying capacities to take advantage of national programs, such as different degrees
Disparities in Medicaid Support
129
of Medicaid cost-sharing associated with states’ federal medical assistance percentages (Welch & Thompson, 1980). A third approach is adopted by health policy researchers in applied fields rather than traditional social science disciplines. This approach is distinguished by the character of its independent and dependent variables. These studies typically highlight health policy design characteristics as independent variables. Similarly, the dependent variables are apt to represent a range of quite specific policy outcomes. Many contributors to this approach examine facets of Medicaid’s generally more controversial role of supporting acute medical care for low-income but non-elderly beneficiaries (e.g., Cromwell, Hurdle, & Schurman, 1987; Kronebusch, 1997). Other researchers concentrate on Medicaid’s support of long-term care, often through contracts with publicor private-sector institutions. Much contract work (e.g., Harrington et al., 2005; Manard, 2002) is descriptive, providing a rich data source. Still other work dealing with Medicaid’s support of long-term care helps to explain disparities in program utilization or expenditure outcomes (e.g., Grabowski, Angelelli, & Mor, 2004; Kane, Kane, Ladd, & Nielsen Veazie, 1998; Swan, Harrington, & Pickard, 2001). These studies often find that policy design has more explanatory power than do the independent variables which dominate the first two literatures discussed above.
RESEARCH DESIGN AND DATA We draw on all three of these approaches in describing and offering provisional explanations for cross-state disparities in Medicaid support for and the quality of nursing facility long-term care for the elderly. Our descriptive analyses show how the states differ on three scales: state total Medicaid nursing facility resource support, state per-unit Medicaid nursing facility resource support and the quality of state nursing facility processes. We then provide a series of three regression models directed toward explaining cross-state variation on these scales in turn. In the third regression model, we include the dependent variables in the first two regression models (state total and per-unit Medicaid nursing facility resource support, respectively) among the independent variables to explain cross-state disparities in the quality of state nursing facility processes. We standardize (z-scores) all variables prior to analysis. Our dependent variables are lagged two years behind our independent variables – 2001 and 1999, respectively. The dependent variables focus exclusively on aspects
130
JEAN GILES-SIMS AND CHARLES LOCKHART
of Medicaid’s support of nursing facility long-term care for the elderly both because it absorbs the majority (nearly 70% nationwide) of Medicaid financial resources devoted to long-term care (Gibson, Gregory, Houser, & Fox-Grage, 2004) and because vastly more extensive cross-state data are available for nursing facility long-term care than for home and communitybased (HCB) care.
Dependent Variables We use two measures of state Medicaid nursing facility resource support, ‘‘total’’ and ‘‘per-service-unit’’ (hereafter, just per-unit). We adopt two indices for a couple of reasons. First, while important measures of state Medicaid nursing facility resource support exist in each category, we find that as total resources rise, per-unit resources do not rise as rapidly and may even fall. Thus, reasonable indices for total and per-unit resource support correlate poorly (r=.22 in this study). So an analysis sensitive to the distinction between these two categories appears appropriate. Second, it is likely that the factors influencing the two categories have important distinctive aspects. For instance, total resource support is usually the result of relatively open public-sector decisions (e.g., the number of Medicaidcertified nursing facility beds authorized by law). These decisions are largely the result of the constellation of political forces in a state which are not limited to but generally include liberal-conservative competition.1 In contrast, minimally how per-unit resources (e.g., the state Medicaid nursing facility reimbursement per resident day) are actually employed, lies largely within the resource allocation procedures of private service-provider bureaucracies. (See Tables 1 and 2 for a summary of variable data and for variable descriptions, sources and dates, respectively.) Our first dependent variable (variable 1) is a measure of state total Medicaid nursing facility resource support in that we count all the resources in particular categories that are applied statewide. For purposes of crossstate comparison some of these resources are still calculated in per-unit terms, but the units are broad state characteristics (e.g., a state’s 65 or older population). Variable 1 takes the form of a three-item scale (a=.79). Each state’s score on this scale is derived by adding the z-scores for its (1) number of Medicaid-certified nursing facility beds per 1,000 state residents 65 or older; (2) percentage of Medicaid expenditures devoted to nursing facilities and (3) Medicaid nursing facility expenditures per 1,000 state residents 65 or older.
Disparities in Medicaid Support
Table 1.
131
Summary of Variables.
Variables (All State-Level) Dependent variables 1. Three-item state total Medicaid nursing facility resource support scale (added z-scores). Also an independent variable in Table 5, regression Model 3 2. Three-item state per-unit Medicaid nursing facility resource support scale (added z-scores). Also an independent variable in Table 5, regression Model 3 3. Four-item quality of state nursing facility processes scale (added z-scores) Independent variables 4. State characteristics/policy diffusion: contiguous neighboring states’ average Medicaid nursing facility expenditures/ 1,000 65+ state residents 5. State characteristics/policy diffusion: state federal medical assistance percentage 6. Health policy: percentage of state nursing facility residents with Medicaid as their primary payor 7. Health policy: percentage of state nursing facilities operated on a forprofit basis Control variables 8. State average resident acuity summary score (labor intensiveness of a state’s nursing facility residents) 9. Percentage of state Medicaid long-term care expenditures devoted to home and community-based (HCB) care a
Mean
S.D.
Low Value
High Value
.29a
2.59
6.18
7.74
.33a
2.18
4.36
4.82
.32a
3.19
9.63
5.85
955.40
344.37
266
2,156
60.72
8.51
50.00
76.78
66.36
7.18
50.0
82.1
62.15
14.79
12.6
81.4
98.29
11.84
69.8
123.6
26.95
12.68
.0
61.1
Means are not equal to zero because the 2001 data are a subset of a longitudinal set.
We use a multi-item scale for this variable and also for our other two dependent variables because the level of performance of individual states over a range of resource support and quality of care criteria varies a great deal. While our scales still provide incomplete operationalization of state nursing facility resource support and quality of care, the operationalization
132
JEAN GILES-SIMS AND CHARLES LOCKHART
Table 2.
Variable Descriptions, Sources and Dates.
Variables (All State-Level) Dependent variables 1. Three-item state total Medicaid nursing facility resource support scale (a=.79). [Added z-scores for the number of state Medicaid-certified nursing facility beds/1,000 65+ state residents; the percentage of state Medicaid expenditures devoted to nursing facilities and state Medicaid nursing facility expenditures/1,000 65+ state residents.] Also an independent variable in Table 5, regression Model 3. 2. Three-item state per-unit Medicaid nursing facility resource support scale (a=.79). [Added z-scores for state Medicaid nursing facility reimbursement per resident day; state Medicaid nursing facility expenditures/state Medicaid nursing facility beneficiary and state average total nurse hours/nursing facility resident day.] Also an independent variable in Table 5, regression Model 3 3. Four-item quality of state nursing facility processes scale (a=.78). [Added z-scores for the percentage of nursing facilities without deficiencies with respect to preserving resident dignity, food sanitation, limiting residents’ accidents and housekeeping.] Independent variables 4. State characteristics/policy diffusion: contiguous neighboring states’ average Medicaid nursing facility expenditures/1,000 65+ state residents 5. State characteristics/policy diffusion: state federal medical assistance percentage 6. Health policy: percentage of state nursing facility residents with Medicaid as their primary payor
Sources
Dates
Authors’ scale from: Gregory and Gibson (2002, R22), USDH&HS (2003, T110), and Gregory and Gibson (2002, R2)
2001a
Authors’ scale from: data provided by Janet Freeze, CMS, USDH&HS (2003, T111), and Harrington et al. (2005, T26)
2001a
Authors’ scale from: Harrington, Carrillo, Wellin, and Shemirani (2002, T35)
2001
USDH&HS (2001, p. T110) and U.S. Census Bureau (2000, T24)
1999
Federal Register (1997)
1999
Harrington et al. (2005, T6)
1999
Disparities in Medicaid Support
133
Table 2. (Continued ) Variables (All State-Level) 7. Health policy: percentage of state nursing facilities operated on a for-profit basis Control variables 8. State average resident acuity summary score (the average labor intensiveness of a state’s nursing facility residents) 9. Percentage of state Medicaid longterm care expenditures devoted to HCB care
Sources
Dates
Harrington et al. (2005, T7)
1999
Harrington et al. (2005, T13)
1999
USDH&HS, CMS 64 Quarterly Report (1999)
1999
a The date is 1999 when this variable is used as an independent variable in Table 5, regression Model 3.
they do provide is more extensive and adequate than that of the individual scale elements. Various diagnostic statistical tests that we have performed support the appropriateness of relying on these multi-variable scales rather than on individual variables.2 Each scale’s elements are selected on the basis of prominence in one or more of the literatures introduced above, scale coherence, scale distinctiveness (from each of our other two dependent variables) and face validity. Variable 2, our state per-unit Medicaid nursing facility resource support scale (a=.79) also has three elements: (1) the Medicaid nursing facility reimbursement rate/resident day; (2) Medicaid nursing facility expenditures/ state nursing facility resident and (3) total facility nurse hours/resident day. Variables 1 and 2 are the dependent variables in our first and second regression models below (Table 5). In the third regression model, they are used as independent variables to assess their influence on our quality of state nursing facility processes scale. This process quality scale (variable 3) is our third dependent variable. It is drawn from CMS OSCAR (Online Survey Certification and Reporting) system data on nursing facility deficiencies. Its four elements (a=.78) are the percentages of a state’s nursing facilities without deficiencies with respect to preserving resident dignity, food sanitation, residents experiencing accidents and facility housekeeping. These four criteria cover a broad range of nursing facility responsibilities.
134
JEAN GILES-SIMS AND CHARLES LOCKHART
Independent Variables Initially, we sought to include independent variables representing the ‘‘predispositions’’ and the ‘‘capacities’’ aspects of the ‘‘state characteristics’’ literature. For predispositions, we chose Berry, Ringquist, Fording, and Hanson’s (1998, updated via ICPSR) index of state (citizenry) political liberalism. For capacities, we chose state per capita personal income. In our initial analyses, each of these variables produced modest effects with respect to our state total and per-unit Medicaid nursing facility resource support scales, especially the former. When we removed variable 4 (discussed next – the average Medicaid nursing facility expenditures/1,000 65+ state residents of a state’s contiguous neighbors) from the models, state liberalism and income acquired some of the influence previously projected by variable 4. Conversely, when we replaced variable 4 but deleted state liberalism and income, average Medicaid nursing facility expenditures/1,000 65+ state residents (variable 4) acquired most of the effects that liberalism and income had exerted in the previous models. Each of these two manipulations of the models left the effects of the other independent variables largely unchanged.3 Since we are concerned to limit the number of independent and control variables to those which our 48 cases can handle,4 we decided that it was preferable to drop state liberalism and income from explicit inclusion in our regression models and to let them ‘‘speak’’ through or, perhaps better, have their ‘‘shadow’’ cast by the effects projected by the average Medicaid nursing facility expenditures/1,000 65+ state residents of a state’s contiguous neighbors. So our first independent variable is inspired by the geographic proximity aspect of the ‘‘policy diffusion’’ literature. Yet we are not testing a policy diffusion hypothesis or employing a time-series analysis. That is, we are not examining relations between the timing of program adoption and other variables. But we are looking for the sort of regional distinctiveness found by Erikson, Wright, and McIver (1993), Lockhart and Giles-Sims (2005) and others, and we use geographic proximity to examine the possibility of regional effects. But we think that the effects we find arise more from factors such as shared ideology or similar state circumstances (e.g., state income levels) than from geographic proximity per se. Specifically, we construct variable 4 by averaging the scores for all of a state’s contiguous neighbors with respect to these neighboring states’ Medicaid nursing facility expenditures/1,000 state residents 65 or older. This statistic is one element of our state total nursing facility resource adequacy scale. Cromwell et al. (1987) – see also Kane et al. (1998) – argue that this
Disparities in Medicaid Support
135
type of measure offers the best single index of a state’s commitment to public social program support by expressing the depth of program coverage across the entire potential recipient population. Variable 5, a state’s federal medical assistance percentage, taps explicitly into one of the contemporary foci of the policy diffusion literature: similar state circumstances. A state’s federal medical assistance percentage is based (non-linearly) on state income levels. Effectively, states above a certain income level have a 50% cap on federal cost-sharing of their Medicaid expenditures. As state incomes fall, federal cost-sharing rises (to a high of 76.82% for Mississippi). So, states with relatively poor populations are encouraged to spend more than they otherwise might on the medical care of particularly vulnerable persons. We draw our next two independent variables from the ‘‘health policy’’ literature. Variable 6 is the percentage of a state’s nursing facility residents who have Medicaid as their only or primary (generally by a wide margin) payor of their nursing facility expenses. This measure varies widely across the states, and others have found evidence that, as this percentage rises, quality of care declines (e.g., Walshe & Harrington, 2002). Variable 7 is the percentage of a state’s nursing facilities that are operated on a for-profit basis. This measure also varies sharply across the states, and – similarly to the situation with variable 6 above – as this percentage increases, quality of care falls (e.g., Harrington, Mullan, & Carrillo, 2004; Harrington, Zimmerman, Karon, Robinson, & Beutel, 2000).
Expectations for the Independent Variables We anticipate that variable 4, the average Medicaid nursing facility expenditures/1,000 65+ state residents of a state’s contiguous neighbors, will be positively and statistically significantly related to our total and perunit state Medicaid nursing facility resource support scales. We recognize that this is an ‘‘umbrella’’ variable which is projecting effects arising from variables not present in our analysis such as state ideology and income levels. We expect that the low-income effect of a state’s federal medical assistance percentage will predominate over the eliciting-more-spending effect in the instance of nursing facility long-term care. Thus we think that this variable will be inversely, though not necessarily strongly, related to our total and per-unit state Medicaid nursing facility resource support scales. We anticipate that, with the exception of the relation between the percentage of state nursing facility residents who have Medicaid as their
136
JEAN GILES-SIMS AND CHARLES LOCKHART
primary payor and our state total Medicaid nursing facility resource support scale (which will be positive and relatively strong), both of our ‘‘health policy’’ measures will be inversely related to all three of our dependent variables.
Control Variables We employ two control variables in this study. The first of these (variable 8) is a state’s average ‘‘resident acuity summary score.’’ This is a measure of the ‘‘difficulty,’’ or, perhaps better, the labor intensiveness of dealing with the persons who comprise the state’s nursing facility resident population. We expect that this variable will be positively and strongly associated with our total and per-unit state Medicaid nursing facility resource support scales and inversely associated with our state quality of nursing facility processes scale. Our other control variable (variable 9) is the percentage of a state’s Medicaid long-term care expenditures that is devoted to HCB care. Since the U.S. Supreme Court’s Olmstead v. L.C. ruling in 1999, states have made greater efforts to employ alternatives to nursing facility care (e.g., adult daycare, assisted living, etc.). They have been motivated both by the pressures this court decision created for offering alternatives to nursing facility care and by the desire to reduce public-sector expenses. While HCB options bring some new applicants for care ‘‘out of the woodwork,’’ so to speak, they are generally less expensive per person served than the nursing facility alternative. Also, Medicaid-supported HCB options are generally limited in that they are undertaken through waivers that allow states to constrain the geographic scope, overall numbers and/or the precise nature of the services provided. Nonetheless, Medicaid-supported alternatives to institutional care have expanded considerably since 1999. As a consequence of this shift toward HCB alternatives, both current state nursing facility expenditures (increased costs per recipient) and HCB expenditures (increased numbers of recipients) are typically rising (Kitchener, Ng, Miller, & Harrington, 2005). We use this variable to control for variation in the degree to which states have shifted long-term care for the elderly from nursing facilities to HCB venues.
RESULTS Table 3 shows how states rank on our three dependent variables. With respect to state total Medicaid nursing facility resource support, the top 15 ranks are comprised of nine Upper Midwestern states (Illinois, Indiana,
Disparities in Medicaid Support
137
Table 3. State Rankings and Scores on the State Nursing Facility ‘‘Total’’ and ‘‘Per-Unit’’ Resource Support Scales and the Quality of State Nursing Facility Processes Scale. Rank
State
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
RI CT ND MA MN OH IL NE IA NY MO PA WI SD IN NH LA OK KS NJ MT AL TN KY AR WY MS DE GA TX MD VT WV MI ME CO NC WA ID VA
1. Total Resourcesa 7.738 5.224 4.825 4.121 3.768 3.641 2.680 2.576 2.353 2.352 2.192 2.126 1.988 1.588 1.576 1.405 1.218 1.028 .838 .630 .583 .154 .140 .093 .003 .141 .144 .216 .597 .620 .702 .746 .819 .830 .988 1.102 1.306 1.542 1.565 1.626
State
DE RI NY CT PA ME MA OH ID AL MD CA WA ND NJ VT MI WV KY NC NM MT MN WY SC NH AZ FL NV CO MS OR WI WA NE UT IL TN MO OK
2. Per-Unit Resourcesa 4.820 3.918 3.762 3.133 2.352 2.264 2.205 1.925 1.921 1.914 1.816 1.416 1.128 1.111 1.028 .820 .769 .289 .041 .208 .222 .552 .591 .592 .885 .957 1.028 1.038 1.049 1.123 1.139 1.260 1.335 1.341 2.132 2.209 2.230 2.250 2.289 2.404
State
VT MA PA MD VA ND RI MT NH OR NM MN MO NE CT NJ IN NV WI UT CO NY KS IA LA WV OH TX SD TN NC IL SC OK ID DE FL AL KY WY
3. Quality of Processesb 5.846 4.519 4.238 3.548 3.548 3.427 3.362 3.210 2.940 2.928 2.672 2.619 2.444 2.091 1.697 1.056 .999 .601 .497 .333 .304 .298 .003 .152 .564 .680 .708 1.119 1.136 1.182 1.208 1.463 1.634 1.718 1.850 2.032 2.076 2.172 2.322 3.389
138
JEAN GILES-SIMS AND CHARLES LOCKHART
Table 3. (Continued ) Rank
State
41 42 43 44 45 46 47 48
FL CA SC NM UT NV OR AZ
a
1. Total Resourcesa 2.122 2.422 2.758 2.979 3.249 3.858 3.939 6.184
State
GA IN SD KS TX IA AR LA
2. Per-Unit Resourcesa 2.491 2.639 2.744 3.002 3.146 3.601 3.794 4.361
State
GA ME WA MS MI AR CA AZ
3. Quality of Processesb 3.526 3.954 4.643 4.987 5.153 5.176 5.884 9.634
Added z-scores for three variables rounded to three decimal places. Added z-scores for four variables rounded to three decimal places.
b
Iowa, Minnesota, Nebraska, North Dakota, Ohio, South Dakota and Wisconsin), five Northeastern states (Connecticut, Massachusetts, New York, Pennsylvania and Rhode Island) and Missouri. These results are striking. Each of these multi-state groups is comprised of contiguous states, and Missouri abuts the former group although it cannot be described as Upper Midwestern. The bottom 15 ranks on this scale include nine Western states (Arizona, California, Colorado, Idaho, Nevada, New Mexico, Oregon, Utah and Washington), four Southeastern states (Florida, North Carolina, South Carolina and Virginia), along with Maine and Michigan. The larger multi-state group represents contiguous states, and three of the four Southeastern states are contiguous. However, Maine and Michigan are representative of the geographic regions at the top of this ranking. On state per-unit Medicaid nursing facility resource support, the top 15 places are held by seven Northeastern states (Connecticut, Maine, Massachusetts, New Jersey, New York, Pennsylvania and Rhode Island), three Western states (California, Idaho and Washington), two Upper Midwestern states (North Dakota and Ohio), two Middle Atlantic states (Delaware and Maryland) and Alabama. Six of the states in the Northeastern group are contiguous, and the two Middle Atlantic states abut them. In the bottom 15 places in this ranking we find seven Southern – predominantly South Central – states (Arkansas, Georgia, Louisiana, Missouri, Oklahoma, Tennessee and Texas), six Midwestern states (Illinois, Indiana, Iowa, Kansas, Nebraska and South Dakota) and two Western states (Utah and Washington). The six South Central states are contiguous, and they abut the six contiguous Midwestern states.
Disparities in Medicaid Support
139
With regard to the quality of state nursing facility processes, in the top 15 ranks of this index we find six Northeastern states (Connecticut, Massachusetts, New Hampshire, Pennsylvania, Rhode Island and Vermont – five of the six contiguous), three Upper Midwestern states (Minnesota, Nebraska and North Dakota), three Western states (Montana, New Mexico and Oregon), two Middle Atlantic states (Maryland and Virginia) and Missouri. The bottom 15 ranks are filled by six Southern – predominantly Southeastern – states (Alabama, Arkansas, Florida, Georgia, Kentucky and Mississippi – five of the six contiguous), five Western states (Arizona, California, Idaho, Washington and Wyoming), with Delaware, Maine, Michigan and Oklahoma as geographic outliers. Regional groupings become less prominent as we move from total to perunit resource support and on to quality of nursing facility processes. Nonetheless, overall, Northeastern and Upper Midwestern states tend to rank more highly on all three of our dependent variable scales than do South Central, Southeastern or Western states. We create a regression model for each of our three dependent variables. These models employ either of two combinations of four independent variables, drawn from six possibilities, and two control variables. Models 1 and 2, directed at explaining cross-state variation in total and per-unit state Medicaid nursing facility resource support, respectively, draw on the four independent and the two control variables introduced above. Model 3, aimed at explaining cross-state variation in the quality of state nursing facility processes, drops the two ‘‘state characteristics/policy diffusion’’ variables, which are more appropriate for explaining state public resources than private nursing facility processes. Instead, this model uses the two independent variables discussed as representing the ‘‘health policy’’ literature above, adds the state total and per-unit Medicaid nursing facility resource support scales as independent variables and relies on the same two control variables. The rationale for adding the two state Medicaid nursing facility resource support scales comes from Donabenian’s (2003) wellknown tripartite conception of health care quality: structure influencing process; process, in turn, influencing outcome. Table 4 shows correlations among the independent and control variables employed in these three models. With two exceptions, instances of high correlations among these variables occur either between independent variables that are not used in the same model or between independent and control variables. In any case these instances do not prompt much disruption in the effects that the individual variables in question exert on the dependent variables.
140
Table 4.
Bivariate Correlations among Eight Independent and Control Variables.
2. Per-unit resource support 4. Neighboring states 5. Medical assistance % 6. Medicaid as primary payor 7. % nursing facilities for-profit 8. Resident acuity score 9. % Medicaid long-term care $ to HCB Significant ato.05. Significant ato.01. Significant ato.001.
2. Per-Unit Resource Support
4. Neighboring States
5. Medical Assistance %
6. Medicaid as Primary Payor
7. % Nursing Facilities ForProfit
8. Resident Acuity Score
.22 .74 .23 .04
.36 .38 .04
.31 .06
.16
.31
.14
.12
.07
.55
.37 .06
.30 .06
.17 .02
.08 .05
.52 .28
.26 .08
.09
JEAN GILES-SIMS AND CHARLES LOCKHART
1. Total Resource Support
Disparities in Medicaid Support
141
Model 1 of Table 5 produces effects that are largely but not entirely consistent with the expectations that we expressed above. The neighboring states’ policy variable reveals a positive and statistically significant effect on state total Medicaid nursing facility resource support. The effect of a state’s federal medical assistance percentage is negative and not strong. The proportion of a state’s nursing facility residents who have Medicaid as their primary payor is, unsurprisingly given the composition of this dependent variable, positive and relatively strong. In contrast, the percentage of a state’s nursing facilities operated on a for-profit basis has an inverse and statistically significant effect. So far, so good, but the strong effect of a state’s average resident acuity summary score is surprising for its direction, negative rather than positive. The percentage of a state’s Medicaid nursing facility expenditures that are devoted to HCB care has an expected inverse, though slight, effect. Model 2 produces results pretty close to our expectations for state perunit Medicaid nursing facility resource support. Once again, the neighboring states’ policy variable has a positive and statistically significant effect. A state’s federal medical assistance percentage has an inverse and somewhat stronger effect than in Model 1. Both the ‘‘health policy’’ variables have negative effects, as we expected, although neither effect is strong. A state’s average resident acuity summary score has the expected positive and statistically significant effect in Model 2, suggesting that the more laborintensive a state’s nursing facility population, the more resources are expended per resident or resident day. The percentage of a state’s Medicaid expenditures devoted to HCB care has an unexpected positive but modest effect. Each of the ‘‘health policy’’ variables in Model 3 (the percentage of state nursing facility residents with Medicaid as their primary payor and the percentage of state nursing facilities operated on a for-profit basis) has, as expected, inverse relations with state quality of nursing facility processes, but their influence is more modest than we anticipated. State total Medicaid nursing facility resource support has an expected positive and statistically significant effect on this dependent variable, but state per-unit Medicaid nursing facility resource support has virtually no effect at all, a result that we find surprising. A state’s average resident acuity summary score produces a modest positive effect when we would have anticipated a negative influence, and a similar situation exists for our second control variable (the percentage of state Medicaid long-term care expenditures devoted to HCB care) which exerts an unexpected positive and statistically significant effect.
Independent Variables
6. % of state nursing facility residents with Medicaid as their primary payor 7. % of state nursing facilities operated on a for-profit basis 1. State total Medicaid nursing facility resource support scale 2. State per-unit Medicaid nursing facility resource support scale 8. State average resident acuity summary score (labor intensiveness of a state’s nursing facility residents) 9. % of state Medicaid long-term care expenditures devoted to HCB care Adjusted R2 Significant ato.05. Significant ato.01. Significant ato.001.
Standard errors in parentheses.
Model 3: Beta Coefficients for: Dependent Variable #3: Quality of State Nursing Facility Processes Scale
Model 1: Beta Coefficients for: Dependent Variable #1: State Total Medicaid Nursing Facility Resource Support Scale
Model 2: Beta Coefficients for: Dependent Variable #2: State per-Unit Medicaid Nursing Facility Resource Support Scale
1.77 (.30) .25 (.26) .62 (.34) .80 (.30)
.85 (.36) .48 (.31) .13 (.40) .34 (.36)
.84 (.29)
.89 (.34)
.57 (.58) .34 (.56) .60 (.20) .04 (.21) .40 (.57)
.14 (.24)
.10 (.29)
.95 (.42)
.62
.25
.27
JEAN GILES-SIMS AND CHARLES LOCKHART
4. Neighboring states’ average Medicaid nursing facility expenditures/1,000 65+ state residents 5. State federal medical assistance percentage
142
Table 5. Regression Models Involving Six Independent Variables and Two Control Variables for State Nursing Facility ‘‘Total’’ and ‘‘Per-Unit’’ Resource Support and Quality of State Nursing Facility Processes.
Disparities in Medicaid Support
143
DISCUSSION First, in terms of describing cross-state variation in state Medicaid nursing facility resource support and quality of care, we find striking inequalities and disparities. Several New England and Upper Midwestern states are more than four standard deviations (on a three-variable scale) higher than a group of Western states on state total Medicaid nursing facility resource support. Several Northeastern states are similarly above a group of Midwestern/South Central states on state per-unit Medicaid nursing facility resource support. And, several Northeastern states are more than eight standard deviations (on a four-variable scale) above a collection of Western and South Central states and Michigan on the quality of state nursing facility processes. These degrees of variation translate into sharp inequalities of access (e.g., to nursing facility beds) and in financing the care (e.g., nurse hours per resident day) of the residents who occupy existing beds. An unsurprising result is that marked disparities occur in the quality of care nursing facility residents experience across the American states. Further, there are clear regional patterns to these inequalities and disparities. With a few exceptions, the Northeast and the Upper Midwest regions have far deeper resources and provide vastly better quality of care than do the Western and Southeastern/South Central regions. In a society in which many aspects of life, spurred by the pervasive influence of the mass media and the growing reach of national chains for nearly all products and services, are increasingly homogenized this Balkanization with respect to support for and the quality of nursing facility long-term care for the elderly is surprising. In a society where citizens are proclaimed to share due process and other egalitarian public-sector benefits, these inequalities and disparities among services for some of society’s most vulnerable members are discouraging. With respect to explaining these variations, we follow Donabedian’s (2003) suggestion of structure influencing process and begin with the former. Our results reveal that selective ‘‘state characteristics’’ variables (e.g., state liberalism and income level) contribute less to explaining state total and perunit Medicaid nursing facility resource support than they generally offer explanations of the resources devoted to other state-level public social programs. Nonetheless, state characteristics, explicitly in the form of aspects of the Medicaid nursing facility policies of a state’s contiguous neighbors and implicitly and less prominently in the form of a state’s ideology and income level, exert a positive and statistically significant effect on a state’s total Medicaid nursing facility resource support. These characteristics shown in Table 5 in the umbrella variable of the average Medicaid nursing
144
JEAN GILES-SIMS AND CHARLES LOCKHART
facility expenditures/1,000 state residents age 65 or older of a state’s contiguous neighbors which, as we have explained above, appears to project much of the more modest influence of a state’s liberalism and income level as well. These characteristics are also reflected in the groups of geographically contiguous states in the top and bottom 15 ranks of Table 3. The other two statistically significant influences (each inverse) on state total Medicaid nursing facility resource support are provided by the percentage of state nursing facilities operated on a for-profit basis and a state’s average resident acuity summary score. The first of these is in keeping with the general efficacy of health policy variables to explain, statistically at least, matters such as program expenditures. Our provisional explanation for the surprising inverse relationship between state total Medicaid nursing facility resource support and the average labor intensiveness of a state’s nursing facility resident population is that as states become more stringent about allowing persons to access Medicaid-supported nursing facility care, the care required by those who do gain access is significantly greater. State characteristics, again via aspects of the Medicaid nursing facility policies of a state’s contiguous neighbors which appears to project the more limited influence of a state’s ideology and income, exert a positive and statistically significant influence on a state’s per-unit Medicaid nursing facility resource support. The other statistically significant factor is the average labor intensiveness of a state’s nursing facility population. This time the relationship is positive. As the labor intensiveness of a state’s nursing facility population escalates, the expenditures and nurse hours per resident day as well as the expenditures per resident also increase. We think that the shift of this variable’s effect from negative to positive in its influence on state total as opposed to per-unit Medicaid nursing facility resource support illustrates the distinctive decision-making venues in which these two dependent variables arise, distinctiveness that we discussed in our section introducing the dependent variables above. Unsurprisingly, the strongest influence on the quality of state nursing facility processes is the level of state Medicaid nursing facility resources, at least what we have termed total resource support which has a positive, statistically significant effect on the quality of state nursing facility processes. The other statistically significant influence (also positive) comes from the percentage of state Medicaid long-term care expenditures that are devoted to HCB care. The positive character of this relationship is surprising, and our best provisional explanation for it to date is that, as the percentage of a state’s Medicaid long-term care recipients in HCB care grows, the number of nursing facility residents falls sufficiently to facilitate better care among them.
Disparities in Medicaid Support
145
So, thinking backward across the explanatory path that we have traced, the primary factor explaining the quality of a state’s nursing facility processes is the level of a state’s total Medicaid nursing facility resource support. It is reasonable to imagine that, were states which currently exhibit low total Medicaid nursing facility resource support levels to upgrade their scores on this scale, improvements in the quality of these states’ nursing facility processes would be likely to follow. Accordingly, the disparities in care that residents of various states currently experience would likely diminish by virtue of states now offering relatively low-quality care moving toward more encouraging levels. Further, the primary factor explaining the level of a state’s total Medicaid nursing facility resource support appears to be an only partially specified set of state characteristics that display themselves in a set of regional groupings. It is unlikely that states in the West and Southeastern/South Central regions could, on their own, quickly come to resemble those of the Northeast and Upper Midwest with respect to predisposition, capacity and thus resource support even if they wished to do so. So upgrading resource support for what is likely to be, as the baby-boom generation continues to age, an era of increased demand for nursing facility care for the elderly is apt to have to be introduced into the former regions from the outside. The federal government currently does supplement the resources of these generally low-income regions through the federal medical assistance percentage. But our data suggest (Table 5) that the levels at which it currently does so are insufficient, at least with regard to spending for long-term care, to offset inadequate indigenous state predispositions, capacities and resource support. In order to reduce sharp inequalities of resource support and disparities in quality of care, the federal government would have to do more with regard to supporting long-term care in these states.
NOTES 1. Certainly interest group influences, some of which are difficult to categorize as liberal or conservative, are involved in these decisions as well. Clive Thomas was kind enough to provide us with the latest round of data (Thomas & Hrebenar, 2004) on the relative influence of various interest groups across the American states. Unfortunately, nursing facility interests are specifically identified in too few states to create sufficient variance on this criterion among the vast majority of states. 2. In terms of diagnostics we have engaged in item-scale correlations and multimethod, multi-trait convergent-discriminant analysis (Campbell & Fiske, 1959) as well as principle component analysis with measures for related concepts. In addition,
146
JEAN GILES-SIMS AND CHARLES LOCKHART
we have compared regressions involving the other variables in this study with the full scale and each of its individual component elements. Similar diagnostics have been employed with respect to variables 2 and 3 for which we have also constructed multiitem scales. 3. The average Medicaid nursing facility expenditures/1,000 65+ state residents of a state’s contiguous neighbors correlates highly with state liberalism (r=.48) and income (r=.49), and these latter two variables are also highly correlated (r=.53), all significant at r.001. 4. Standards as to how many independent variables 48 cases can support vary from one authority to another. We acknowledge that we are pressing the limit to employ four independent and two control variables in each of our regression models.
ACKNOWLEDGMENTS We thank Bayliss Camp and Joanne Green for computer assistance; Estelle Alexander, Lori Anderson, Mary Conte, Janet Freeze, Deborah Kidd, Brenda C. Spillman, Clive S. Thomas, Craig Volden and Tish Williams for data; Howard Leichter, Mark Schlesinger and Tim Tilton for comments on previous versions of this chapter and Keon Montgomery for work on an initial version.
REFERENCES Bailey, M. A., & Rom, M. C. (2004). A wider race? Interstate competition across health and welfare programs. Journal of Politics, 66(2), 326–347. Balla, S. J. (2001). Interstate professional associations and the diffusion of policy innovations. American Politics Research, 29(3), 221–245. Barrilleaux, C., Holbrook, T., & Langer, L. (2002). Electoral competition, legislative balance, and American state welfare policy. American Journal of Political Science, 46(2), 415–427. Berry, F. S. (1994). Sizing up state policy innovation research. Policy Studies Journal, 22(3), 442–456. Berry, W. D., Fording, R. C., & Hanson, R. L. (2000). An annual cost of living index for the American states, 1960–1995. Journal of Politics, 62(2), 550–567. Berry, W. D., Ringquist, E., Fording, R. C., & Hanson, R. L. (1998). Measuring citizen and government ideology in the American states, 1960–1993. American Journal of Political Science, 42(1), 327–348. Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multimethod-multi-trait matrix. Psychological Bulletin, 65(1), 81–105. Canon, B. C., & Baum, L. (1981). Patterns of adoption of tort law innovations: An application of diffusion-theory to judicial doctrines. American Political Science Review, 75(4), 1975–1987. Case, A. C., Hines, J. R., Jr., & Rosen, H. S. (1993). Budget spillovers and fiscal policy interdependence: Evidence from the states. Journal of Public Economics, 53(3), 285–307.
Disparities in Medicaid Support
147
Cook, F. L., & Barrett, E. J. (1992). Support for the American welfare state: The views of congress and the public. New York: Columbia University Press. Cromwell, J., Hurdle, S., & Schurman, R. (1987). Defederalizing Medicaid: Fair to the poor, fair to taxpayers? Journal of Health Politics, Policy and Law, 12(1), 1–34. Donabedian, A. (2003). An introduction to quality assurance in health care. New York: Oxford University Press. Elazar, D. J. (1966). American federalism: A view from the states. New York: Thomas Y. Crowell. Erikson, R. S., Wright, G. C., & McIver, J. P. (1993). Statehouse democracy: Public opinion and policy in the American states. New York: Cambridge University Press. Federal Register. (2006). Available at: http://gpoaccess.gov/fr/index.html. September 23. Gibson, M. J., Gregory, S. R., Houser, A. N., & Fox-Grage, W. (2004). Across the states: Profiles of long-term care. Washington, DC: AARP Public Policy Institute. Grabowski, D. C., Angelelli, J. J., & Mor, V. (2004). Medicaid payment and risk-adjusted nursing home quality measures. Health Affairs, 23(5), 243–252. Gray, V. (1973). Innovation in the states: A diffusion study. American Political Science Review, 67(4), 1174–1185. Gregory, S. R., & Gibson, M. J. (2002). Across the states: Profiles of long-term care. Washington, DC: AARP Public Policy Institute. Grogan, C. M. (1999). The influence of federal mandates on state Medicaid and AFDC decision-making. Publius, 29(3), 1–30. Grossback, L. J., Nicholson-Crotty, S., & Peterson, D. A. (2004). Ideology and learning in policy diffusion. American Politics Research, 32(5), 521–545. Harrington, C., Carrillo, H., & Mercado-Scott, C. (2005). Nursing facilities, staffing, residents, and facility deficiencies, 1998 through 2004. Report prepared for the Service Employees International Union. Department of Social and Behavioral Sciences. University of California, San Francisco. Harrington, C., Carrillo, H., Wellin, V., & Shemirani, B. B. (2002). Nursing facilities, staffing, residents, and facility deficiencies, 1995 through 2001. Report prepared for the Service Employees International Union. Department of Social and Behavioral Sciences. University of California, San Francisco. Harrington, C., Mullan, J. T., & Carrillo, H. (2004). State nursing home enforcement systems. Journal of Health Politics, Policy and Law, 29(1), 43–73. Harrington, C., Zimmerman, D., Karon, S. L., Robinson, J., & Beutel, P. (2000). Nursing home staffing and its relationship to deficiencies. Journal of Gerontology: Social Sciences, 55B, S278–S287. Hays, S. P. (1996). Influences on reinvention during the diffusion of innovations. Political Research Quarterly, 49(3), 631–650. Kane, R. L., Kane, R. A., Ladd, R. C., & Nielsen Veazie, W. (1998). Variation in state spending for long-term care: Factors associated with more balanced systems. Journal of Health Politics, Policy and Law, 23(2), 363–390. Kitchener, M., Ng, T., Miller, N., & Harrington, C. (2005). Medicaid home and communitybased services: National program trends. Health Affairs, 24(1), 206–212. Kronebusch, K. (1997). Medicaid and the politics of groups, recipients, providers and policy makers. Journal of Health Politics, Policy and Law, 22(3), 839–878. Lockhart, C., & Giles-Sims, J. (2005). Variation in elderly friendliness across the U.S. states: Operationalization, rankings and selected consequences. Publius, 35(3), 425–447.
148
JEAN GILES-SIMS AND CHARLES LOCKHART
Manard, B. (2002). Nursing home quality indicators: Their uses and limitations. Washington, DC: AARP Policy Research Institute. Matthews, J. L. (2001). Beat the nursing home trap: A consumer’s guide to assisted living and long-term care (3rd ed.). Berkeley, CA: Nolo. Mintrom, M. (1997). Policy entrepreneurs and the diffusion of innovation. American Journal of Political Science, 41(3), 738–770. Pear, R. (2005a). U.S. gives Florida a sweeping right to curb Medicaid. New York Times (October 20), pp. A1, A20. Pear, R. (2005b). Congress weighs big cuts to Medicaid and Medicare. New York Times (October 30), p. A17. Schneider, S. K. (1993). Examining the relationship between public policies: AFDC and Medicaid. Public Administration Review, 53(4), 368–380. Sharkansky, I., & Hofferbert, R. I. (1969). Dimensions of state politics, economics and public policy. American Political Science Review, 63(3), 867–879. Shipan, C. R., & Volden, C. (2004). The diffusion of state-level anti-smoking policies in a federal system. Paper presented at the annual meeting of the State Politics and Policy Conference, Akron, OH. Swan, J. H., Harrington, C., & Pickard, R. B. (2001). Explaining state Medicaid nursing home reimbursement. Journal of Health and Social Policy, 12(4), 53–73. Thomas, C. S., & Hrebenar, R. J. (2004). Interest groups in the states. In: V. Gray & R. L. Hanson (Eds), Politics in the American states: A comparative analysis (8th ed., pp. 100–128). Washington, DC: CQ Press. U.S. Census Bureau. (2000). Statistical abstract of the United States: The nation’s data book. (on-line version). Washington, DC: U.S. Department of Commerce. U.S. Department of Health and Human Services (USDH&HS). (1999). Centers for Medicare and Medicaid. 64 Quarterly Expense Report. Available on line at www.cms.hhs.gov; then Research, Statistics, Data and Systems; then Computer Data and Systems; then Medicaid Budget and Expenditure System (MBES). U.S. Department of Health and Human Services. (2000). A profile of Medicaid: Chart book 2000. Washington, DC: DH&HS. U.S. Department of Health and Human Services. (2001). Health care financing review. Medicare and Medicaid Statistical Supplement. Available on line at www.cms.hhs; then Research, Statistics, Data and Systems; then Statistics, Trends and Reports. U.S. Department of Health and Human Services. (2003). Health care financing review. Medicare and Medicaid Statistical Supplement. Volden, C. (2006). States as policy laboratories: Emulating success in the children’s health insurance program. American Journal of Political Science, 50(2), 294–312. Walker, J. L. (1969). The diffusion of innovations among the American states. American Political Science Review, 63(3), 880–899. Walker, J. L. (1973). Problems in research on diffusion of policy innovations. American Political Science Review, 67(4), 1186–1191. Walshe, D., & Harrington, C. (2002). Regulation of nursing facilities in the United States: An analysis of resources and performance of state survey agencies. The Gerontologist, 42(4), 475–487. Welch, S., & Thompson, K. (1980). The impact of federal incentives on state policy innovation. American Political Science Review, 74(4), 715–726.
PROFESSIONAL POLITICS AND THE CHALLENGE OF ANESTHESIA AVAILABILITY IN RURAL HOSPITALS Deborah Sullivan and Leah Rohlfsen ABSTRACT Rural areas that are struggling to recruit and retain qualified health practitioners are caught in the crossfire of turf battles between allied health practitioners and physician groups. The most intensely political of these inter-occupational turf battles is between anesthesiologists (MDAs) and certified registered nurse anesthetists (CRNAs), who are the sole providers of anesthesia in two-thirds of rural hospitals (American Association of Nurse Anesthetists (AANA), 2007a, 2007b). The ability of many rural hospitals to provide anesthesia services is dependent on CRNAs. This study uses data collected from CRNAs in Iowa and Arizona in 2005 to focus on the impact of the turf battle on the professional interactions and opinions of CRNAs. Arizona and Iowa were chosen for this study because not only do the policies in these states contrast with each other, but the contexts in which CRNAs practice are also dissimilar. The majority of Arizona’s CRNAs work in urban areas in close proximity with MDAs. Most CRNAs in Arizona report that their workplace Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 149–167 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00007-5
149
150
DEBORAH SULLIVAN AND LEAH ROHLFSEN
interactions with MDAs have suffered as a result of the turf battle, despite the lack of any action to opt out of the federal Medicare requirement of physician supervision of CRNAs. While most CRNAs in Iowa perceive that job opportunities and the quality and cost of health care have improved as a result of opting out of the federal supervision requirement of CRNAs, the impact on their social interactions in the workplace depends on location and the structural context of their work. Most CRNAs in Iowa’s urban areas continue to work in a structural context of de facto supervision by MDAs. As a result, only a minority report that their professional interactions in the workplace have improved. The outcome for Iowa’s rural CRNAs is decidedly different. The majority function as independent practitioners and have experienced an improvement in their social interactions in the workplace and greater economic reward. These occupational privileges should improve the ability of Iowa’s rural hospitals to recruit and retain CRNAs and, as a consequence, surgical services in rural areas.
INTRODUCTION The Census Bureau defines rural areas as open country with settlements of fewer than 2,500 residents. After a century of net out-migration of young adults, the proportion of residents living in such environments has declined from 54 percent in 1910 to 21 percent today (U.S. Census Bureau, 1912, 2007). Nevertheless, the number of people in rural America, 59 million, is equivalent to the population of Italy. Despite their large number, rural residents have far more problems with access to health care because of more limited availability than their urban counterparts. The older age distribution of the population left behind and the growth of retirement communities in rural areas exacerbate the problems. All other things being equal, an older population has a higher prevalence of chronic disease. However, all other things are not equal because, compared to urban populations, rural populations have a higher prevalence of additional risk factors for health problems including lower levels of education and income as well as higher levels of health risk behaviors such as smoking, obesity, and physical inactivity (Tompkins, James, Sam, & Elizabeth, 2005). Despite more health care needs, rural residents are less likely than urban residents to have private health insurance (Kaiser Commission on Medicaid and the Uninsured, 2003). Low levels of private health insurance coverage,
Professional Politics and the Challenge of Anesthesia Availability
151
combined with low population density and low incomes, challenge the ability of many rural areas to maintain sufficient human and technological resources to meet the health care needs of residents. The more distant a rural area is from a large urban area, the less likely residents are to have access to a full range of health care services, including surgery which depends, not only on the availability of surgeons, but also on the availability of qualified providers of anesthesia – the focus of this paper. The 2,009 nonfederal, acute care community hospitals in rural America (American Hospital Association, 2007) are the heart of health care services in these areas. In some places, the local hospital is the only source of health care in the community. It is a fragile system. Many small towns experienced the closure of their local hospitals in the 1980s when faced with a federal agenda to contain the escalating costs of Medicare. Tight finances do not help rural hospitals with their perennial struggle to recruit and retain the personnel necessary to meet their communities’ needs. A shortage of qualified practitioners is one of the most common obstacles to health care in rural areas (Ricketts, Karen, & Randy, 1999, p. 7). While urban areas have approximately 225 physicians per 100,000 residents, the ratio in rural areas is only 119 (Larson & Norris, 2003, p. 27). The national shortage of nurses is also most severe in rural areas. Much of the disparity in the availability of physicians is due to the high opportunity costs assumed by specialist physicians working in rural areas where the number of potential patients is smaller and a greater portion of them are uninsured or covered only by Medicare, part A. A recent study reports that the availability of some physician specialists such as internists has increased, however the availability of other specialists such as anesthesiologists continues to decline in rural areas (Rosenthal, James, Fox, Wysong, & FitzPatrick, 1997). The Chair of the American Society of Anesthesiologists’ Committee on Rural Access to Anesthesia Care estimates that less than 5 percent of the country’s active anesthesiologists practice in rural areas (Schweitzer, 2006). By comparison, the corresponding rate for surgeons is 12.5 percent. The difference underscores the insufficiency of available anesthesiologists in rural areas. Physicians and allied health practitioners are reluctant to locate in rural areas for many reasons, ranging from lower potential incomes and concerns about professional isolation to limited health facilities and a lack of employment and educational opportunities for other family members. The federal government has established many incentive programs such as the Medical Incentive Payments, Rural Outreach Grant Program, Health Careers Opportunity Program, Health Resources and Services
152
DEBORAH SULLIVAN AND LEAH ROHLFSEN
Administration’s Centers of Excellence Program, as well as Title VII and VIII health professional training funding, and scholarship and loan repayment programs to encourage physicians and other health practitioners to locate in rural areas. Success has been limited and rural hospitals continue to close. Another strategy of the federal government has been to adopt policies to encourage the training and expanded use of nurse practitioners and physician assistants in rural areas. About 20 percent of nurse practitioners and 18 percent of physician assistants now practice in rural areas (Hart, Lishner, & Rosenblatt, 2003, p. 13). Nurse practitioners and physician assistants have become a critical part of the rural health workforce. Nurse practitioners, for example, are involved with 37 percent of rural hospital outpatient visits, compared to only 5 percent of these visits in urban areas (Anderson & Hampton, 1999). An increasing number of research studies indicate that the quality of nurse practitioner and physician assistant care is comparable to that of a generalist physician in areas where there is an overlap in their scope of practice (e.g. Lenz, Mundinger, Kane, Hopkins, & Lin, 2004; Mundinger et al., 2000; Rosenblatt et al., 1997). A similar conclusion was reached in a study of surgical mortality by type of anesthesia provider (Pine, Holt, & Lou, 2003). Hospital administrators and others have used such studies to support the expanded recruitment of these mid-level health practitioners to meet the needs of rural areas. Rather than embracing the expanded use of qualified mid-level practitioners who have been trained and licensed to perform a wide array of medical services, physician groups have mounted an increasingly united wall of resistance. They have contested all research that finds no difference in the outcomes of physicians and allied health practitioners, and they have lobbied hospital administrators, regulatory agencies, and political leaders to keep these practitioners in a subordinated position, supervised by physicians. The risks of anesthesia are particularly difficult to quantify because they are now so low. The Centers for Medicare and Medicaid Services (2001a) concludes that due to ‘‘[a]dvances in medical knowledge, implementation of practice guidelines, better drugs and safer equipment y. There is no evidence that CRNA [certified registered nurse practitioners] independent practice would cause adverse outcomes.’’ Faced with a proliferation of specialty turf battles, the American Medical Association recently formed a coalition of physician organizations, the Scope of Practice Partnership, to mount an organized campaign against all administrative, regulatory, and legislative attempts to expand the use of allied health practitioners in the delivery of health care (Croasdale, 2006).
Professional Politics and the Challenge of Anesthesia Availability
153
Rural areas that are struggling to recruit and retain qualified health practitioners find themselves caught in the crossfire of these turf battles between allied health practitioners seeking the occupational privilege of autonomous practice enjoyed by limited professionals such as dentists and podiatrists, and physician groups defending their traditional right to determine, not only their own scope of practice, but also the scope of practice of other health care workers. The most intensely political of these many inter-occupational turf battles is between anesthesiologists (MDAs) and certified registered nurse anesthetists (CRNAs), who are the sole providers of anesthesia in two-thirds of rural hospitals (American Association of Nurse Anesthetists (AANA), 2007a, 2007b). The need for available qualified anesthesia practitioners in rural areas has figured prominently in this turf battle. Without CRNAs the majority of surgery now performed in rural hospitals would not be possible. CRNAs have won the right to be independent practitioners in 14 states at the time of this writing. MDAs are trying to hold the line in the remaining states, making these states less attractive work locations for CRNAs, thereby jeopardizing the ability of rural hospitals in them to continue to offer surgical services. After providing an overview of the changing structural context of anesthesia delivery, we will present data on the impact of this turf battle and independent practice on the professional interactions of CRNAs practicing in two states, one of which includes a large rural population.
LEGISLATION, REGULATION, AND COURT CASES: THE CHANGING CONTEXT OF ANESTHESIA SERVICES Use of anesthesia spread rapidly after the 1846 demonstration of ether’s ability to render a person insensible to the pain of surgery. Because it fit better with the caretaker model of nursing than the curative model of medicine, anesthesia evolved into the first clinical specialty in nursing (Bankert, 1989; Gunn, 1975; Thatcher, 1953). Nurse Alice Magaw who worked with the Mayo brothers in Rochester, Minnesota became known as the ‘‘mother of anesthesia’’ for her excellent outcomes with thousands of patients using both ether and chloroform. She documented her work in numerous publications, which made surgeons aware that the problems of relatively high morbidity and mortality, particularly with chloroform, could be reduced with training in clinical techniques.
154
DEBORAH SULLIVAN AND LEAH ROHLFSEN
The first formal nurse anesthetist training program was established at St. Vincent Hospital in Portland, Oregon in 1909. Others followed. Most practitioners at the time, however, were trained by employees of the companies selling gas machines or by observing an experienced anesthetist such as Magaw (Bankert, 1989; Thatcher, 1953). The military initiated training programs during World War I that greatly expanded the number of formally trained nurse anesthetists. Demand for nurse anesthetists continued to grow after the war because of the proliferation of surgical procedures that were facilitated by antiseptic practices, reliable blood transfusions, and improved X-rays, as well as anesthesia. Training programs multiplied and attracted some physicians and dentists. Many nurses became directors of anesthesia services in hospitals in the first-half of the twentieth century, reflecting the prevailing view that anesthesia was primarily a nursing specialty, despite a small number of physicians who were interested in specializing in it. During organized medicine’s campaign in the early twentieth century to drive midwives and other ‘‘irregular’’ competitors out of business and elevate ‘‘regular medicine’’ into a scientific profession with uniformly high standards of education and protective licensure laws, some physician anesthetists argued unsuccessfully in court that anesthesia should be a restricted medical practice (Bankert, 1989; Betcher, Ciliberti, Wood, & Wright, 1955 p. 769; Thatcher, 1953, pp. 110–124, 132–152). Frank v South (1917) affirmed that administering anesthesia is a nursing task and ruled that nurse anesthetists could make independent judgments when administering anesthesia because they work under the direction and supervision of the operating physician. Following this decision, some states adopted statutes that required nurse anesthetists to work under the ‘‘supervision’’ or ‘‘direction’’ of a physician (Blumenreich, 2000, p. 405). No state statute requires supervision or direction by a physician anesthetist. Physician anesthetists failed to gain control of anesthesia in the early twentieth century because they lacked the support of the American College of Surgeons (Halpern, 1992), the AMA (Smith, 2000), and the American Hospital Association (Bankert, 1989, pp. 65–107). They also lacked evidence of superior outcomes. As a result, nurse anesthetists continued to administer the vast majority of anesthetics through the middle of the twentieth century, including during World War II when the military reopened training programs to meet its needs (Bankert, 1989; Gunn, 1975; Thatcher, 1953). CRNAs continue to provide the majority of anesthetics to members of the Armed Forces and their dependants. After the war, more physicians began to think of anesthesia as technologically sophisticated and medically challenging. Surgeries had
Professional Politics and the Challenge of Anesthesia Availability
155
become more invasive, extensive, and risky, and innovations in anesthesia replaced the century-old practice of drip inhalation with intravenous anesthetics, spinal epidurals, and muscle relaxants. Physician anesthetists changed their name to anesthesiologists in 1945 to distance themselves from nurse anesthetists and underscore their identity as a fledgling medical specialty. The impact of these changes on the division of labor in anesthesia was substantial. In 1942, there was only one physician anesthetist for every 17 nurse anesthetists (Gunn, 1975). By 1967, the ratio was one MDA for every two CRNAs (Cromwell, Rosenbach, Pope, Butrica, & Pitcher, 1991). After World War II most hospitals in large urban areas complied with a recommendation of the American College of Surgeons that a ‘‘hospital’s anesthesia department be directed by an anesthesiologist or a physician with a special interest in anesthesia y’’ (Thatcher, 1953, p. 166). However, despite the large increase in MDAs between 1942 and 1967, there was still far too few of them to staff the many smaller hospitals, especially those in rural areas, where CRNAs continued to direct all the anesthesia services, technically under the oversight of a surgeon or hospital administrator. The introduction of Medicare radically changed the economics of anesthesia (Cromwell, 1999). CRNAs, who still administered the majority of anesthetics in 1965 as hospital employees, could not bill Medicare for their services, although the hospital could, as long as the CRNAs worked under the direction of a physician. Echoing the court ruling in 1917, the ‘‘directing’’ physician did not have to be a MDA; he could be the operating surgeon. In contrast, not only were MDAs allowed to bill Medicare for all anesthesia services that they personally administered, they also were allowed to bill for every anesthetic delivered by CRNAs under their ‘‘direction,’’ discounted by half the time involved in the actual procedure. MDAs who employed their own CRNAs could bill for their employees’ work as if they personally administered each procedure. These rules allowed MDAs to bill for the simultaneous administration of anesthesia to multiple patients. MDAs’ average earnings jumped from $39,400 in 1970 (American Medical Association, 1977), which was at the low end of physicians’ earnings, to $131,900 in 1982, which was about the same as general surgeons’ earnings, and then to $228,500 in 1992, which put them among the most highly compensated specialties (Anders, 1995). The impact of this new economic incentive is evident in the steep increase in the number of active MDAs from 10,860 in 1970 to 35,715 in 2000 (Bureau of Health Professions, 2003). This increase occurred despite a deliberate reduction in the number of first-year residencies by nearly onehalf between 1992 and 1997 because of concerns about creating an
156
DEBORAH SULLIVAN AND LEAH ROHLFSEN
oversupply in medical climate that had changed from increasing access to cutting costs (Grogono, 2004). The ‘‘featherbedding’’ by MDAs allowed under Medicare became a target of government attempts to rein in run-away health expenditures beginning in the early 1980s. The Tax Equity and Fiscal Responsibility Act of 1982 required MDAs to be physically available to be paid for directing CRNAs (Cromwell, 1999). The act also limited the number of reimbursable concurrent cases to four. The Omnibus Budget Reconciliation Act (OBRA) of 1987 introduced a sliding schedule of reimbursement discounts to discourage MDAs from directing more than three CRNAs concurrently. The 1989 OBRA reduced the base component of MDAs’ Medicare reimbursement. Then, after determining that the cost of reimbursing an MDA-CRNA team was as much as 140 percent more than the cost of reimbursing a solo practitioner, the 1993 OBRA phased in a permanent cap on reimbursement equal to what a solo MDA would make and required that half be paid to a CRNA, if one was involved in the case. The latter stipulation could be interpreted as evidence that both anesthesia providers had equal value, a step toward greater professional recognition for CRNAs. Other government efforts to curb health care costs in the 1980s had some unintended consequences. The DRGs in 1983 encouraged hospitals to terminate CRNAs, despite their greater cost-effectiveness, because the hospitals could no longer bill Medicare separately for CRNAs’ work and had to absorb the costs as part of surgical overhead (Bankert, 1989). Some MDAs took advantage of this new reimbursement structure and negotiated exclusive contracts with hospitals that wanted to shift the cost of anesthesia services from the hospital to independent contractors. CRNAs’ attempts to have such arrangements declared illegal under antitrust laws failed because some MDAs and MDA groups employ CRNAs (Bankert, 1989; Cromwell, 1999; Gunn, 1999). In response to this threat to their professional aspirations, if not their future existence, the American Association of Nurse Anesthetists lobbied for the right to bill Medicare directly for their work (Bankert, 1989). Congress’s desire to halt spiraling health care costs was in their favor. The 1986 OBRA gave CRNAs the right to bill Medicare directly for their services, beginning in 1989. This was a major step toward independent practice. At the end of 1997, the agency that oversees Medicare proposed that CRNAs should be regulated in the same way that all other state-licensed practitioners were regulated – by the state, not the federal government (Centers for Medicare and Medicaid Services, 2001a). To make this happen
Professional Politics and the Challenge of Anesthesia Availability
157
the agency proposed the elimination of the requirement that CRNAs be supervised by a physician to qualify for reimbursement by Medicare. The implications were enormous. In states with no specific statutes or regulations requiring medical supervision or direction, CRNAs, who already won the right to bill Medicare directly, would be independent practitioners. Tense relations erupted into an all out war between the national organizations representing the country’s 35,000 MDAs and 25,000 CRNAs (Bettin, 1999). Both ran aggressive public relations campaigns that issued position papers and press releases. Members wrote editorials and sought media interviews to muster support to pressure legislators and agency administrators. The American Society of Anesthesiologists ranked 30th on Fortune’s list of the most power lobbying groups in the capital in 1998 and 92nd in 1999. The American Association of Nurse Anesthetists ranked 101st in 1999. Bills opposing and supporting the proposal to end the federal supervision requirement were introduced in both the Senate and the House of Representatives every session from 1998 to 2000. Stakeholders such as the National Rural Health Association and the American Hospital Association offered support. On his last working day in office President Clinton signed the proposed rule change. The incoming Bush administration immediately postponed implementation. At that point, the governors of states with many rural hospitals joined the lobbying efforts in favor of the rule change. The Bush administration eventually was pressured into implementing a compromise to the dismay of MDAs. The compromise retains the federal supervision requirement, but allows a state to choose to ‘‘opt out’’ after its governor consults with the boards of medicine and nursing and determines that removal of the supervision requirement is in the best interests of the state and consistent with state law (Centers for Medicare and Medicaid Services, 2001b). The anesthesia war has moved from Washington DC into the states. As of 2006, 14 states (Table 1) have opted out of the federal supervision requirement, most in the first two years after the compromise rule was implemented (AANA, 2007a, 2007b). In his formal letter to the Centers for Medicare and Medicaid announcing that opting out of the federal requirement was in Iowa’s best interest, Governor Vilsack noted that CRNAs are the ‘‘exclusive provider of anesthesia services in 91 out of 118 Iowa hospitals.’’ The Governors of Nebraska, New Hampshire, New Mexico, Kansas, Alaska, Oregon, and South Dakota also mentioned the interests of their rural communities in their declarations of exemption. Most of the remaining states with a large rural population that have not opted out are in the South – West Virginia, Mississippi, Arkansas, Alabama,
158
DEBORAH SULLIVAN AND LEAH ROHLFSEN
Table 1. States That Have Opted Out of the Federal Medicare Supervision Requirement for CRNA Reimbursement: Year and Percent Rural. Year of Opt Out 2001 2002 2002 2002 2002 2002 2003 2003 2003 2003 2003 2004 2005 2005
State
Percent Rural
Iowa Nebraska Idaho Minnesota New Hampshire New Mexico Kansas North Dakota Washington Alaska Oregon Montana South Dakota Wisconsin
38.9 30.2 33.6 29.1 40.7 25.0 28.6 44.1 18.0 34.4 21.3 45.9 48.1 31.7
Kentucky, North and South Carolina, Tennessee, and Oklahoma. Two exceptions are Maine and Vermont. All states now face considerable organized resistance to legislation that would position them to opt out. The MDAs’ professional organization closely monitors and aggressively challenges proposed administrative changes and legislation that might advance the independent practice of CRNAs in more states. For example, the April 2007 issue of the ASA Newsletter warns that pending legislation in Connecticut, Illinois, New York, Pennsylvania, and Utah ‘‘could remove physician involvement in the administration of anesthesia y’’ (Percy, 2007).
DATA The data come from a survey of 224 CRNAs in Arizona and 336 CRNAs in Iowa in 2005. Iowa was the first state to opt out of the federal supervision requirement. CRNAs in Iowa provide anesthesia in most rural hospitals, without which the health care services offered would be severely limited. In contrast to Iowa, there has been no organized effort in Arizona to opt out. Arizona and Iowa were chosen for this study because not only do the policies in these states contrast with each other, but the context in which CRNAs practice are also very different. Most CRNAs in Iowa practice in
Professional Politics and the Challenge of Anesthesia Availability
159
rural areas, whereas most CRNAs in Arizona practice in urban or mid-size areas. The Arizona population consists of CRNAs licensed by the Arizona State Board of Nursing and those on a mailing list provided by the Arizona Association of Nurse Anesthetists. The two lists were used because the Arizona State Board of Nursing’s list is updated only when licenses are renewed and some of the addresses were no longer valid. The Iowa population consists of CRNAs who are registered with the state of Iowa according to the Iowa Board of Nursing. The addresses of people who are retired, overseas, out of state, or whose mail was undeliverable were deleted. Questionnaires were mailed with a cover letter explaining the purpose of the study and a stamped return envelope. A second mailing to non-respondents was sent 3 weeks after the first mailing. The Arizona study had a response rate of 42 percent and the Iowa study had a response rate of 54.8 percent. The purpose of the questionnaire was to assess CRNAs opinions regarding nurse anesthesia practice and the impact of recent changes in the health care system. This study focuses on CRNAs’ professional identity and the impact of the turf battle and other changes in the health care system on their interactions with other health care workers and patients. For example, the respondents were asked to assess the impact of the turf battle on their interactions with anesthesiologists, other health care workers, and patients. In addition, background data that might be expected to influence workplace interactions was collected, including gender, education, income, employment context, and involvement in professional associations.
FINDINGS The socioeconomic and demographic characteristics of respondents in Arizona and Iowa are presented in Table 2. There were not enough rural CRNAs among the Arizona respondents to breakdown their distribution by practice location. Only 19 work in rural areas. This mirrors an important underlying difference in the demographics of Arizona and Iowa; less than 12 percent of Arizonans live in a rural area compared to almost 40 percent of Iowans. This difference in distribution provides a point of comparison to measure the influence of Iowa’s strong rural character on the experience of CRNAs. There are some obvious differences in the characteristics of CRNAs between the two states and between rural and urban practitioners in Iowa.
160
DEBORAH SULLIVAN AND LEAH ROHLFSEN
Table 2. Socioeconomic and Demographic Characteristics of CRNAs in Arizona and in Iowa by Rural and Urban Practice Locationa. Arizona
Gender Male Female Age o34 35–44 45–54 55–64 65+ Years in practice 0–10 11–20 21–30 31–40 41 or more Educational level Certificate/Diploma Bachelor’s Degree Master’s Degree/PhD Practice setting Self-employed contractor Hosp. w/MDA on staff Hosp. w/out MDA on staff Mixed CRNA-MDA group MDA group CRNA group Practice context Supervised by an MDA Supervised by a physician Team member Independent practitioner Income (employed fulltime) $50,000–99,999 $100,000–149,000 $150,000–199,000 Over $200,000 N a
Iowa
State %
State %
Rural %
Urban %
44.2 55.8
63.0 37.0
76.7 23.3
49.5 50.5
6.7 18.9 38.9 25.6 10.0
5.0 16.0 43.6 29.3 6.1
3.4 12.5 45.5 29.5 9.1
6.5 19.6 41.3 29.3 3.3
31.3 21.9 27.1 15.6 4.2
28.4 17.5 35.5 15.8 2.7
21.1 17.8 38.9 17.8 4.4
35.9 17.4 31.5 14.1 1.1
11.6 26.3 62.1
22.3 29.3 48.4
28.9 30.0 41.1
16.1 29.0 54.8
31.6 41.1 4.2 4.2 5.3 0
39.7 15.2 8.7 7.6 15.8 5.4
62.2 3.3 16.7 0.0 2.2 7.8
17.2 26.9 1.1 15.1 29.0 3.2
38.9 6.3 14.7 27.4
22.4 5.5 20.2 50.3
5.6 8.9 3.3 80.0
39.1 2.2 37.0 20.7
5.6 57.7 22.5 14.1 95
2.9 44.0 26.9 26.3 184b
3.5 29.4 28.2 38.8 90
2.2 57.8 25.6 14.4 93
The percentages do not add to 100% because the percent answering other or not answering is not presented. b One respondent did not indicate location of practice.
Professional Politics and the Challenge of Anesthesia Availability
161
The proportion of CRNAs who are male is not only higher in Iowa than in Arizona, it is also significantly higher than the national 49 percent (AANA, 2007a, 2007b). Iowa’s male CRNAs are much more likely to practice in rural areas than female CRNAs. The age profile of Iowa’s CRNAs also varies between rural and urban areas; those in rural areas tend to be older and have more practice experience than those in urban areas. This age difference contributes to the lower proportion of rural CRNAs who have post-baccalaureate training. Until 1979, when the Master of Science in Nurse Anesthesia was established, CRNAs received a bachelor’s degree in Nurse Anesthesia (Bankert, 1989). More recently, CRNAs must complete a two to three year master’s degree in Nurse Anesthesia. Arizona’s CRNAs look like Iowa’s urban CRNAs on some characteristics, such as gender and the higher proportions of both young practitioners and post-baccalaureate training. However, they also look like Iowa’s rural CRNAs on other characteristics, such as higher proportions of older practitioners and more years of experience. Some of these differences may be explained by Arizona’s rapid growth. The state’s population more than doubled between 1980 and 2005, creating a strong demand for health care workers that has made Arizona attractive to young adults earlier in their careers. At the same time, Arizona is a magnet for older workers approaching retirement. The biggest differences between Iowa’s rural and urban practitioners are in the structure of their practices and income. The majority of Iowa’s rural CRNAs report that they are self-employed contractors, compared to only one in six urban CRNAs. Others work as employees of hospitals with no MDAs on staff. This structural arrangement is almost nonexistent in urban areas. Urban CRNAs mostly work in hospitals with MDAs on staff or as part of contracted groups that include MDAs. Few report that they are independent practitioners. In stark contrast, 80 percent of Iowa’s rural CRNAs work as independent practitioners. A surprisingly large proportion of Arizona’s CRNAs also report working as self-employed contractors compared to Iowa’s urban CRNAs, although more work in hospitals with MDAs on staff and few work as part of a contracted group, the arrangement that employs nearly half of Iowa’s urban CRNAs. While the practice settings of CRNAs in Arizona differ substantially from Iowa’s urban CRNAs, the income profiles of the two groups are nearly identical. This is a highly compensated occupation and the data clearly show that the compensation is much higher for CRNAs who work in rural areas. The significantly higher income distribution of rural CRNAs in Iowa should attract other anesthesia practitioners to these areas, resulting in increased competition and lower incomes. Nevertheless, few rural or urban
162
DEBORAH SULLIVAN AND LEAH ROHLFSEN
CRNAs express concern about the increasing supply of CRNAs on either CRNA job opportunities in general or their personal practice (Table 3). The vast majority believes that more CRNAs will improve the quality and reduce the cost of health care. In contrast, half of the rural CRNAs think that the increasing supply of MDAs will reduce job opportunities for CRNAs and more than one in five anticipate that they will personally feel the impact. Iowa’s urban CRNAs, many of whom work as part of a group practice with MDAs or in hospitals with MDAs, are much less likely than rural CRNAs to be concerned about the impact of more MDAs on CRNA jobs in general and their personal practice. Many appear to be comfortable working closely with MDAs. However, a large percent of Arizona’s urban CRNAs also work closely with MDAs, although less often as part of a group practice. Unlike their Iowa counterparts, Arizona’s urban CRNAs are much more likely to express concern about the increasing supply of MDAs. Similarly, they are approximately as likely as Iowa’s rural CRNAs to report that the turf battle over anesthesia has a negative impact on their
Table 3. CRNA Views on the Impact of the Increasing Supply of Anesthesia Providers and the Turf Battle with Anesthesiologistsa. Iowa’s Rural CRNAs Positive % Impact of increasing supply of MDAs on: CRNA job opportunities 2.4 CRNA personal practice 4.6 Quality of health care 12.2 Cost of health care 10.6 Impact of increasing supply of CRNAs on: CRNA job opportunities 44.2 CRNA personal practice 32.2 Quality of health care 79.5 Cost of health care 69.3 Impact of turf battle on interactions with: Administrators 18.6 MDAs 1.3 Other physicians 26.7 Other nurses 18.8 Patients 21.7 N 90 a
Iowa’s Urban CRNAs
Arizona’s Urban CRNAs
Negative %
Positive %
Negative %
Positive %
Negative %
51.2 21.8 24.4 82.4
3.4 5.4 17.8 6.9
28.7 14.1 8.9 79.3
2.9 5.5 17.1 2.9
50.0 28.8 11.4 78.6
3.5 1.1 0.0 2.3
44.0 34.4 72.8 69.2
3.3 2.2 1.1 4.4
43.7 29.2 72.2 72.9
11.3 1.4 0.0 2.9
12.8 61.5 16.3 12.9 9.6
10.2 4.5 20.9 16.5 15.5
12.5 39.8 17.4 7.1 10.7
4.4 2.9 2.8 4.2 5.6
39.7 65.7 35.2 18.3 18.1
93
76
The percentages do not add to 100% because the proportion with a neutral responses or who did not answer is not presented.
Professional Politics and the Challenge of Anesthesia Availability
163
Table 4. Iowa’s CRNA Views on the Impact of the Iowa Opting Out, Providers’ Exclusive Contracts with Hospitals, and Low Medicare Reimbursement ratesa. Iowa’s Rural CRNAs Positive % Impact of Iowa opting out on interactions with: Administrators 50.0 MDAs 7.0 Other physicians 53.2 Other nurses 42.3 Patients 40.0 Impact of Iowa opting out on: CRNA job 83.5 opportunities CRNA personal 69.5 practice Quality of health care 77.5 Cost of health care 77.8 Impact of exclusive contracts on: CRNA job 7.1 opportunities CRNA personal 7.3 practice Quality of health care 9.9 Cost of health care 9.9 Impact of low Medicare reimburse rates on: CRNA job 6.3 opportunities CRNA personal 4.9 practice MDA job opportunities 6.7 Hospitals’ 2.6 sustainability Hospitals’ economic 3.9 well-being
Iowa’s Urban CRNAs
Negative %
Positive %
Negative %
3.8 47.9 5.1 3.8 0.0
23.9 8.5 38.0 22.5 26.1
1.4 26.8 4.2 1.4 0.0
1.3
81.7
0.0
0.0
49.3
0.0
0.0 1.2
80.6 69.4
0.0 2.8
72.6
13.4
54.9
18.3
11.0
11.0
51.9 75.3
10.0 10.0
25.0 56.3
73.8
12.3
54.8
54.9
2.4
44.7
56.7 78.9
2.9 5.3
51.5 72.0
82.9
5.3
73.3
a
The percentages do not add to 100% because the proportion with a neutral responses or who did not answer is not presented.
interactions with MDAs. They also are more likely than Iowa’s CRNAs to report that it has a negative impact on their interactions with administrators and other physicians, even though the turf battle has no public visibility in Arizona. A small minority claims that it has affected their interactions with patients as well.
164
DEBORAH SULLIVAN AND LEAH ROHLFSEN
Unlike CRNAs in Arizona, few in Iowa perceive that the turf battle has a negative impact on their social interactions in the workplace, other than with MDAs. A large portion of the rural practitioners link their positive social interactions with administrators, physicians other than MDAs, other nurses, and patients to Iowa’s decision to waive the federal supervision requirement which made it possible for 80 percent of them to practice as limited professionals with the same level of autonomy as dentists and podiatrists (Table 4). Iowa’s urban CRNAs, few of whom actually practice as limited professionals, are much less likely than their rural counterparts to report that Iowa’s decision to opt out had a positive impact on their social interactions in the workplace. At the same time, very few report a negative impact either. The majority of urban CRNAs view the legal exemption from the supervision requirement as a neutral factor in their social interactions in the workplace and they are much less likely to report that it has a positive impact on their personal practice. In contrast, the vast majority of both rural and urban practitioners report that the opt out increased CRNA job opportunities in Iowa and has a positive impact on quality and cost of health care. Large portions of both groups also express concern about the increasingly common practice in urban Iowa for solo practitioners and groups of practitioners to negotiate exclusive provider contracts with hospitals. Most of these contracts are with MDAs or MDAowned groups. Such contracts remain much less common in rural areas and few rural CRNAs have felt any impact on their personal practice from these arrangements. Nevertheless, rural CRNAs are overwhelmingly concerned about them and believe that they reduce CRNA job opportunities and increase the cost of health care. The final issue of concern to most CRNAs in Iowa is the impact of Medicare’s low reimbursement rates. Despite their high incomes, they fear that these rates will not only affect CRNA and MDA job opportunities, but also hospitals’ economic well-being and financial sustainability.
CONCLUSION It has been six years since Governor Vilsack took the bold step of declaring that Iowa would be the first state to opt out of the federal requirement of physician supervision of CRNAs. With many rural hospitals in his state struggling to maintain adequate staff for surgery, he decided to eliminate a regulatory obstacle to the availability of qualified anesthesia practitioners that lacked scientific justification. His action made it possible for CRNAs
Professional Politics and the Challenge of Anesthesia Availability
165
in Iowa to function as autonomous, limited professionals, a goal long sought by the American Association of Nurse Anesthetists. While most CRNAs in Iowa perceive that job opportunities and the quality and cost of health care have improved as a result of this change, the impact on their social interactions in the workplace depends on location and the structural context of their work. Most CRNAs in Iowa’s urban areas continue to work in a structural context of de facto supervision by MDAs. As a result, only a minority report that the regulatory change has improved their professional interactions in the workplace. The outcome for Iowa’s rural CRNAs is decidedly different. The vast majority now function as independent practitioners and have experienced an improvement in their social interactions in the workplace. As a group the rural CRNAs have gained both professional status and greater economic reward. These occupational privileges should improve the ability of Iowa’s rural hospitals to recruit and retain CRNAs and, as a consequence, surgical services. A comparison of the findings for Arizona and Iowa suggests that Governor Vilsack’s quick action to exempt his state from the federal oversight requirement has spared Iowa’s CRNAs from the repercussions of the on-going turf battle on social interactions in the work place. This is not the case in Arizona. The large majority of Arizona’s CRNAs works in urban areas in close proximity with MDAs. Two-thirds report that their workplace interactions with MDAs have suffered as a result of the turf battle, despite the lack of any action to position Arizona to opt out of the federal regulation. A smaller, but still far greater percentage of CRNAs in Arizona than in Iowa, report collateral damage from the turf battle on their social interactions with other health care workers and patients. Iowa’s data are being scrutinized for evidence that its exemption has put patients at risk. The history of anesthesia indicates that the findings are not likely to change many minds. Professional politics are not about evidencebased medicine; they are about power and privilege.
ACKNOWLEDGMENTS A 2005 ASU Women’s Studies Small Grant, The Turf Battle over Anesthesia: Interpersonal Relations in the Workplace, funded the Iowa survey.
166
DEBORAH SULLIVAN AND LEAH ROHLFSEN
REFERENCES American Association of Nurse Anesthetists (AANA). (2007a). Becoming a CRNA. www.aana.com American Association of Nurse Anesthetists (AANA). (2007b). Fact sheet concerning opt-outs and November 13, 2001 CMS rule. www.aana.com American Hospital Association. (2007). Trends affecting hospitals and health systems: 2007 chartbook. Chart 2.1. www.aha.org American Medical Association. (1977). Profile of medical practice. Chicago: AMA. Anders, G. (1995). Numb and number: Once hot specialty, anesthesiology cools as insurers scale back. Wall Street Journal, A1. Anderson, D., & Hampton, M. (1999). Physician assistants and nurse practitioners: Ruralurban settings and reimbursement for services. Journal of Rural Health, 5, 252–263. Bankert, M. (1989). Watchful care: A history of America’s nurse anesthetists. New York: Continuum. Betcher, A. M., Ciliberti, B. J., Wood, P. M., & Wright, L. H. (1955). Fifty years of organized anesthesiology. JAMA, 159(8), 766–770. Bettin, C. (1999). American association of nurse anesthetists cracks fortune’s list of most powerful lobbying groups. American Association of Nurse Anesthetists, Press Release. Blumenreich, G. (2000). Supervision. Journal of the American Association of Nurse Anesthetists, 68(5), 404–408. Bureau of Health Professions. (2003). United States health personnel fact book: Table 202. Washington, DC: U.S. Department of Health Resources and Services Administration. Centers for Medicare and Medicaid Services. (2001a). Physician supervision of certified registered nurse anesthetists: Fact sheet. www.cms.hhs.gov/media/press/release.asp? Counter=391 Centers for Medicare and Medicaid Services. (2001b). States allowed to set standards for anesthesia. www.cms.hhs.gov/media/press/release.asp?counter=319 Croasdale, M. (2006). Physician task force confronts scope-of-practice legislation. American Medical News, 49. Cromwell, J. (1999). Barrier to achieving a cost-effective workforce mix: Lessons from anesthesiology. Journal of Health Politics, Policy and Law, 24, 1331–1361. Cromwell, J., Rosenbach, M., Pope, G., Butrica, B., & Pitcher, J. (1991). CRNA manpower forecasts: 1990–2010. Medical Care, 29, 628–644. Grogono, A. (2004). Resident numbers and graduation rates from residencies and nurse anesthetist schools in 2004. ASA Newsletter, 68 www.asahq.org/Newsletters.html Gunn, I. (1975). Nurse anesthetists-anesthesiologist relationships: Past, present, and implications for the future. Journal of the American Association of Nurse Anesthetists, 43, 129–139. Gunn, I. (1999). Professional identity and historical roots. CRNA: The Clinical Forum for Nurse Anesthetists, 10(1), 41–47. Halpern, S. (1992). Dynamics of professional control: Internal coalitions and crossprofessional boundaries. American Journal of Sociology, 97(4), 994–1021. Hart, L., Lishner, D., & Rosenblatt, R. (2003). Rural health workforce: Context, trends, and issues. In: E. Larson, K. Johnson, T. Norris, D. Lishner, R. Rosenblatt & L. G. Hart (Eds), State of health workforce in rural America: Profiles and comparison (pp. 7–14). Seattle, Washington: WWAMI Rural Health Research Center, University of Washington.
Professional Politics and the Challenge of Anesthesia Availability
167
Kaiser Commission on Medicaid and the Uninsured. (2003). Health insurance coverage in rural America. www.kff.org/uninsured/upload/Health-Insurance-Coverage-in-RuralAmerica-PDF.pdf Larson, E., & Norris, T. (2003). Rural demography and the health workforce: Interstate comparisons. In: E. Larson, K. Johnson, T. Norris, D. Lishner, R. Rosenblatt & L. G. Hart (Eds), State of the health workforce in rural America: Profiles and comparison (pp. 23–27). Seattle, Washington: WWAMI Rural Health Research Center, University of Washington. Lenz, E., Mundinger, M., Kane, R., Hopkins, S., & Lin, S. (2004). Primary care outcomes in patients treated by nurse practitioners or physicians: Two-year follow-up. Medical Care Research and Review, 61, 332–351. Mundinger, M., Kane, R., Lenz, Elizabeth., Totten, A., Tsai, W.-Y., Cleary, P., Friedewal, W., Siu, A., & Shelanski, M. (2000). Primary care outcomes in patients treated by nurse practitioners or physicians: A randomized trial. JAMA, 283, 59–68. Percy, L. (2007). Legislation seeks to remove physician supervision requirements. ASA Newsletter, 71, 4. www.asahq.org/Newsletters/2007/04-07/stateBeat04_07.html Pine, M., Halt, K., & Lou, Y-B. (2003). Surgical mortality and type of anesthesia provider. Journal of the American Association of Nurse Anesthetists, 71, 109–116. Ricketts, T., Karen, J-W., & Randy, R. (1999). Populations and places in rural America. In: R. Thomas (Ed.), Rural health in the United States (pp. 7–24). New York: Oxford University Press. Rosenblatt, R., Dobie, S., Hart, L., Schneeweiss, R., Gould, D., Raine, T., Benedetti, T., Pirani, M., & Perrin, E. (1997). Interspecialty differences in the obstetric care of low-risk women. American Journal of Public Health, 87, 344–351. Rosenthal, T., James, P., Fox, C., Wysong, J., & FitzPatrick, P. (1997). Rural physicians, rural networks, and free market health care in the 1990s. Archives of Family Medicine, 6, 319–323. Smith, B. (2000). The genesis of the American society of anesthesiologists. ASA Newsletter, 64, 9. www.asahq.org/newsletters/2000/09_00/genesis0900.html Schweitzer, M. (2006). Committee on rural access to anesthesia care: Message from the country. ASA Newsletter, 70, 12. www.asahq.org/Newsletters/2006/12-06/schweitzer12_06.html Thatcher, V. (1953). History of anesthesia with emphasis on the nurse specialist. Philadelphia: JB Lippincott Company. Tompkins, N., James, R., Sam, Z., & Elizabeth, V. (2005). Engaging rural youth in physical activity promotion research in an after-school setting. Preventing Chronic Disease, 2(Special Issue), www.cdc.gov/pcd/issues/2005/nov/05_0075.htm U.S. Census Bureau. (1912). Statistical abstract of the United States: 1911. Washington, DC: Government Printing Office. U.S. Census Bureau. (2007). Statistical abstract of the United States: 2007. Washington, DC, www.census.gov/statab/www/
SOCIOECONOMIC AND RACIAL/ ETHNIC DISPARITIES IN SUBSTANCE ABUSE TREATMENT PROVISION, TREATMENT NEEDS AND UTILIZATION Matthew E. Archibald ABSTRACT Despite continuing socioeconomic and racial/ethnic gaps in many health care services, the National Healthcare Disparities Report (2004) documents parity in substance abuse treatment provision among individuals of varying socioeconomic and racial/ethnic backgrounds. This study investigates that achievement by analyzing the relationship between community socioeconomic and racial/ethnic disadvantage and organizational provision of substance abuse treatment, treatment need and utilization across United States counties, 2000, 2002 and 2003. Results confirm equity in service provision in poorer communities and those with higher concentrations of African Americans. Significant disparities remain, however, in communities with higher concentrations of Hispanics, youth and female-headed households. Limitations and implications for future studies of health care provision are discussed. Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 171–200 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00008-7
171
172
MATTHEW E. ARCHIBALD
INTRODUCTION Recent studies have identified significant socioeconomic and racial/ethnic gaps in health care services (Kaiser Family Foundation, 2003; National Healthcare Disparities Report (NHDR), 2004; Smedley, Stith, & Nelson, 2003). The Surgeon General reports notable disparities in access, quality and availability of health services for minority groups; where disparities are defined as barriers to care for some groups and not for others (Office of the Surgeon General, Substance and Mental Health Services Administration, 2001). Although effective and well-documented health services exist, socioeconomic and racial and ethnic minorities fall below the general population in the care they receive. The NHDR (2004) shows that despite some over-time improvement in several important areas, individuals from lower socioeconomic backgrounds and racial and ethnic minorities with varying backgrounds are more likely to report unmet health care needs and less likely to have a consistent source of healthcare, receive routine care and benefit from insurance coverage. Research in substance abuse services uncovers similar disparities. African Americans and Hispanics experience greater unmet need, poorer quality of care and less access to care for their substance use/abuse (drug and alcohol) and mental health problems than whites (Wells, Klap, Koike, & Sherbourne, 2001). Schmidt, Greenfield and Mulia (2006) argue that ethnic minorities experience more negative consequences than whites, and therefore have greater treatment needs, and receive far less appropriate service. Even under managed care programs, utilization of treatment services remains lower for minority clients than for whites (Daley, 2005). Epidemiological studies of substance abuse treatment reveal utilization gaps between areas that have prominent minority populations (McAuliffe & Dunn, 2004; McAuliffe, Woodworth, Zhang, & Dunn, 2002). Furthermore, mounting evidence indicates that these disparities are linked in a complex manner to socioeconomic resources (Crimmins, Hayward, & Seeman, 2004). In contrast, studies such as the National Household Survey on Drug Abuse (SAMSA 2002 – hereafter NHSDA) show that receipt of some health care services like substance abuse treatment is about the same for blacks, Hispanics and whites. The National Healthcare Disparities Report (2004) using data from the NHSDA survey, argues that unmet need in substance abuse treatment provision is actually less for blacks insofar as the ‘‘selected population receives more care than the comparison population’’ (p.145). The NHSDA study also found that among those with lower educational attainment, inequities in receipt of substance abuse treatment services had
Socioeconomic and Racial/Ethnic Disparities
173
been ameliorated, with high school graduates receiving greater care than the college-educated. Several other studies report research showing no differences in treatment utilization across racial and ethnic groups of problem drinkers (Weisner & Schmidt, 2001 cited in Schmidt et al., 2006). Related research at the community level shows that the provision of health services in metropolitan areas, especially substance abuse treatment, tends to be concentrated in minority areas (Allard, 2004). This implies equity in community-level provision of substance abuse treatment services consistent with the individual-level findings of survey research such as the NHSDA. These conflicting studies highlight the unresolved nature of socioeconomic and racial/ethnic disparities in health care. In addition, little research investigates the provision or availability of substance abuse treatment services per se, and focuses instead on individual access and utilization, since equity in the latter are presumed to represent availability. This exacerbates the problem and suggests the importance of developing an institutional- and community-level research agenda examining the differential structuring of health care services in general, and provision, access and utilization of substance abuse treatment services in particular, for minorities. Part of the problem is that community-level research is still attempting to establish that context matters. This is partly because most health research is framed in a paradigm where individual-level proximal influences – such as behaviors or biomarkers of pathogenic processes – take precedence over contextual factors. (Morenoff & Lynch, 2004, p. 407)
Recent studies such as Jacobson (2006, 2004a, 2004b), Jacobson, Robinson and Blumenthal (2007) and Freisthler, Gruenewald, Johnson, Treno, and LaScala (2005) are beginning to address the issue by showing that the substance abuse treatment environment has a larger role to play in eliminating socioeconomic and racial/ethnic disparities in health care and ameliorating individual outcomes than previously believed. However, none of these studies focuses on the differential supply of health services. This study investigates the socioeconomic and racial/ethnic characteristics of the environmental context of health care service delivery, and examines the extent to which resource advantages and disadvantages shape treatment provision. It presents results of analyses of disparities in substance abuse treatment provision, substance abuse treatment needs and treatment utilization at the county level for the United States during the years 2000, 2002 and 2003. By conceptualizing drug treatment service provision in terms of the availability of drug treatment at the community level, rather than as individual-level access or utilization, this study extends our understanding of
174
MATTHEW E. ARCHIBALD
the relationship between disadvantage and privilege in communities with regard to the provision of substance abuse treatment services and suggests how these characteristics align with various markets (e.g., urban cityscape or gated community) and across social space (e.g., racial and ethnic concentration, unemployment). A key premise of this research is that the structural characteristics of environments are linked to the supply-side of substance abuse treatment provision which then has an impact on individuals’ access, utilization and, ultimately, health. For instance, residential segregation creates disparities in health care through limited distribution of resources and restricted access to services in racially and economically disadvantaged communities. Differential treatment provision in these communities can be uncovered by examining how the availability of treatment organizations and organizational programs is variously distributed across populations, how distribution of services is related to service-needs, and ultimately, what gaps remain between allocation of resources and unmet needs. Investigating substance abuse service provision promotes a better understanding of the institutional origins of disparities in the availability of health care. First, lack of community-level research is due to the emphasis on program performance and individual outcomes, especially in the treatment industry, linked to individual patterns of service utilization (Harrison & Sexton, 2004). Theorizing substance abuse treatment provision, need and utilization as processes at work in the social environment rather than as features of individuals’ consumption patterns uncovers the structure of socioeconomic and racial/ethnic disparities. At any of these points, socioeconomic and racial/ethnic differences may shape the process and discrepancies can be identified both logically and empirically. As noted by Robert and House (2000), there are actual community, regional and societal socioeconomic, racial and ethnic contingencies in health and health care that are not simply the aggregate of individual relationships. Second, as a result, while provision of services is intimately linked to their utilization, since the lack of available services will limit their use (Smaje, 2000), simply examining utilization patterns will not reveal very much about the availability of them, since the determinants of treatment availability (e.g., local housing policies, tax codes, discriminatory social practices) will vary notably between communities. Gaps in provision may reflect availability of resources, organizational responses to different markets (Greve, 2002) and quality of outreach (Gregoire, 2002). Although survey research, such as the NHSDA captures differences in unmet need and utilization across broad aggregates of individuals, hinting at these underlying
Socioeconomic and Racial/Ethnic Disparities
175
institutional processes, surveys are not very good at generating estimates of differences between communities, the true level at which processes such as markets and segregation and discrimination operate (Beshai, 1984; Herman-Stahl et al., 2001). Third, since there is a lack of knowledge about ‘‘the differential adequacy of treatment resources’’ (McAuliffe & Dunn, 2004, p. 1000) and consequently about geographic variation in socioeconomic and racial and ethnic disparities in substance abuse treatment needs, policy is less informed than it might otherwise be. A key issue with regard to the provision of social services is the gap between collective need for services and access to those services, particularly where gaps are based on socioeconomic and racial and ethnic characteristics of place. Institutional and organizational decision-making about location of substance abuse treatment services is anchored in a set of unexamined assumptions about causal relationships between treatment provision, access and outcomes that also reflects perceptions about the source of disparities that may not be supported empirically. By continuing to pursue resource allocation decisions based on a tacit theory of social, economic and cultural relationships, policy makers run the risk of misallocating services and thereby perpetuating the conditions that foster substance abuse rather than ameliorating it.
BACKGROUND Socioeconomic and Racial/Ethnic Disparities in Health Socioeconomic disparities in health care are a major impediment to achieving health in developing societies Robert and House (2000). They are responsible for a number of health problems including low birth weights, cardiovascular disease, arthritis, diabetes, cancer and high mortality rates (Adler & Newman, 2002; Kawachi, Daniels, & Robinson, 2005; Link & Phelan, 1995). Moreover, socioeconomic status when linked to racial and ethnic groups shows a pattern of unequal burden that is particularly acute for those in poverty (Adler & Newman, 2002). Reducing socioeconomic disparities and racial/ethnic health inequities has been a central component of Healthy People 2010 (U.S. Department of Health and Human Services, 2000) and an ongoing research concern for public and private health agencies. Both socioeconomic status and race/ethnicity underlie the three major aspects of health: behavior, environmental exposure and care (Adler & Newman, 2002).
176
MATTHEW E. ARCHIBALD
In 2002, the Institute of Medicine reported that racial and ethnic minorities receive lower-quality health care regardless of economic status (Smedley, Stith, & Nelson, 2003). The study goes on to argue that discriminatory practices have become institutionalized, resulting in unequal access to care, unequal treatment for similar illnesses and conditions and notably different means of health care financing. Studies have found that racial disparities in mortality and mental illness remain when socioeconomic and environmental factors are taken into account (Williams, 1999). This raises the question: To what extent do either or both these factors matter in health inequalities? Research shows clear evidence of racial disparities in health, mostly among African Americans and whites, yet there is much less evidence on the role of socioeconomic factors in understanding the mutually reinforcing influence of socioeconomic and racial/ethnic factors (Crimmins et al., 2004; Kawachi et al., 2005). Studies argue that race and ethnicity play an independent as well as interdependent role in fostering health disparities (Kawachi et al., 2005). Instead of viewing racial and ethnic differences in health as a function of socioeconomic stratification, this research promotes a research agenda that views race and class as co-determinants of inequity. Taking such an approach, a study by American Cancer Society researchers (Ward, Jemal, & Cokkinides, 2004) found that residents in poorer counties had higher cancer mortality rates than those in less impoverished places but that the mortality rates among African American men and other minorities, was higher still, even as areal poverty rates were taken into account. Unfortunately, the study does not explore whether services for early detection are also unequally distributed. As a partial corrective, Sunshine and Bhargavan (2006) found that while breast cancer detection services were equitably distributed across minority communities, poorer black and Hispanic communities were less likely than white areas to have access to these services.
The Provision of Health Care Services An important question therefore arises: To what extent to do these two main factors of socioeconomic and race reflect differences in the distribution of health care services, and can we infer equitable provision from parity in individual access and utilization? The answer is not as obvious as it seems. On the one hand, there is no lack of research at the individual level on the negative relationship between socioeconomic status, race and the utilization of health care services (Kronenfeld, 2005). Care is based on individuals’
Socioeconomic and Racial/Ethnic Disparities
177
ability to pay for it, directly or with insurance, location, management and delivery of service, clinical uncertainties, practitioner beliefs and other factors (NHDR, 2004, p. 14). On the other hand, distribution of health care services shapes access and utilization. Financial and geographic access is limited for marginalized groups and for those who cannot afford it or who experience outright discrimination due to racial or ethnic characteristics (Alliance for Health Reform, 2003). Not surprisingly, socioeconomic and racialized patterns of residential segregation are linked to the location and spatial distribution of health care facilities (Smaje, 2000). Typically, prestigious teaching hospitals, for instance, are located in urban areas, as are minority populations, but it is not clear that the quality of care received is any better for the minority populations who use these facilities for primary care.
The Ecology of Substance Abuse Treatment Provision Like most disparities in health care research, studies investigating the socioeconomic and racial/ethnic differences in substance abuse emphasize outcomes, including service utilization, rates of relapse and attrition (e.g., Howard, LaVeist, & McCaughrin, 1996; Jacobson, 2006), although treatment access, especially in studies evaluating policies aimed at reducing disparities, has been a topic of increasing importance (see e.g., Daley 2005; Heflinger, Chatman, & Saunders, 2006). Less attention has been devoted to investigating the distribution of substance abuse services at the community level. While recent studies have begun to estimate areal provision rates using social indicator data (McAuliffe & Dunn, 2004; McAuliffe et al., 2002), the empirical questions about disparities remain unanswered: Does availability of health care resources have a clear socioeconomic and racial/ethnic patterning, and, do related spatial patterns emerge when examining the distribution of these resources? Two studies directly bearing on the issue are McAuliffe and Dunn’s (2004) examination of state-wide utilization gaps (measuring differences between provision and need) and Allard’s (2004) study of the geographic distribution of substance abuse and mental health service in three metropolitan areas. McAuliffe and Dunn suggest that the greatest unmet need in treatment provision exists in areas with high concentration of minorities, while Allard shows that service providers are already most likely to be located in urban centers, with a high concentration of minorities and high poverty.
178
MATTHEW E. ARCHIBALD
Substance Use and Abuse and Treatment Needs What accounts for the continued existence of disparities in treatment services particularly among geographically circumscribed social groups? Organization theory suggests that location decisions are structured by local labor markets, the presence of competitors, and resource availability (Greve, 2002). The most parsimonious answer, however, is that different communities and their constituent populations have differential service (treatment) needs and treatment providers locate to fill demand, especially when government funding is an important incentive (see DiMaggio & Anheier, 1990). Thus, area disadvantage, delineated by poverty, unemployment, median income, race and ethnicity, housing and levels of education, while likely to shape the provision of services, is also likely to reflect differential substance abuse treatment needs (Storr, Chen, & Anthony, 2004). Silver, Mulvey and Swanson (2002), using National Institute of Mental Health’s Epidemiological Catchment surveys, show that neighborhood disadvantage is associated with higher rates of major depression and substance abuse disorder while McAuliffe and Dunn (2004), McAuliffe et al. (2002), Gregoire (2002), Herman-Stahl et al. (2001) and others uncover persistent differences in drug and alcohol treatment needs across socioeconomic and racial/ethnic dimensions of county and state populations. In Rhode Island, McAuliffe et al. (2002) report that towns with greater urbanization, a higher proportion of those living in poverty, higher percentages of African American and Hispanic residents, and those living in homeless shelters are more likely to have abundant drug and alcohol treatment needs. Herman-Stahl et al. (2001) found that treatment need and intervention were related to the proportion of 15–34 year old males in a population, its urbanicity and density. And Gregoire (2002) argues that alcohol and drug abuse are correlated with a number of areal characteristics, including drop-out rates, murder and property crime, DUIs, motor vehicle fatalities and drug-related deaths. In contrast, other research suggests that marginalized and impoverished areas are not more drug dependent than more affluent areas, and therefore no more nor less in need of treatment services. Some studies argue that substance use/abuse rates are roughly similar among whites, African Americans and Hispanics (Barker et al., 2004; NHDR, 2004). Saxe et al. (2001) show that minority concentration accounts for large differences in visibility of drugs (i.e., sales) but not for any differences with respect to actual use, while the National Comorbidity Study maintains that blacks actually have lower prevalence rates than whites with respect to substance use disorders, among other mental health outcomes (Kessler et al., 1994). Still,
Socioeconomic and Racial/Ethnic Disparities
179
perhaps not surprisingly, minorities experience more negative social and economic consequences of substance use and abuse than whites (Schmidt et al., 2006), and their prospects in treatment are diminished because of much poorer treatment outcomes, including higher program attrition rates and higher rates of relapse (Jacobson et al., 2007). This suggests gaps in our knowledge about treatment needs based on individual outcomes, since need for continued treatment and further provision remains high in the face of differential recidivism rates across social groups.
The Spatial Patterning of Disparities The last question asks whether spatial patterns emerge when examining the relationships between socioeconomic and racial/ethnic features of community substance abuse treatment, substance abuse treatment needs and utilization or unmet need. Limited research and continuing debate about socioeconomic and racial/ethnic features of substance abuse treatment, needs and utilization suggest that mechanisms influencing the spatial distribution of services remain unknown as well. One way of posing the question is in terms of the diffusion of substance abuse, the treatment needs association with it and subsequent treatment. Recent studies in the spatial analysis of substance use and abuse (e.g., Freisthler et al., 2005; Freisthler & Gruenewald, 2005; Gruenewald, Freisthler, Remer, LaScala, & Treno, 2006; Lipton, Gorman, Wieczorek, & Gruenewald, 2003) maintain the importance of delineating the geographic characteristics of addiction and its cures. If diffusion is a fundamental process in the spread of a condition such as substance abuse, then it might also be reversed by strategic distribution of substance abuse treatment, which should then follow the same geographic pattern. Treatment provision is likely to diffuse across communities because location decisions, especially those made at the government level, are a matter of public health policy.
HYPOTHESES This study addresses four research questions. The first question is: Are there socioeconomic and racial/ethnic disparities in substance abuse treatment provision? Substance abuse treatment is an individual or entity that provides alcohol or drug abuse diagnosis, therapy or referral for rehabilitation of persons with addiction or dependence (Brady & Ashley, 2005). In this study it strictly refers to organizations providing services related to addiction and
180
MATTHEW E. ARCHIBALD
dependence and their availability. Given recent survey data results, it might be expected that: (1a) there will be parity among socioeconomic and racial/ ethnic groups with respect to treatment service provision. However, based on McAuliffe’s research and that of Gregoire (2002), Herman-Stahl et al. (2001) and others, it is likely that: (1b) there will be significant disparity among socioeconomic and racial/ethnic groups with respect to treatment service provision. The question naturally arises as to the causes of these disparities. The simplest and most straightforward answer is that differential treatment is a function of differential substance abuse treatment needs between groups. The need for treatment is based on substance abuse, which is the excessive and harmful use of a substance, especially alcohol or a drug, leading to clinically significant impairment or distress and often followed by recurrent use (American Psychiatric Association (1994): Diagnostic and Statistical Manual IV). Posing this answer in the form of the second research question: Are there socioeconomic and racial/ethnic differences in treatment needs? Again, given survey results, it might be expected that: (2a) there will be parity among socioeconomic and racial/ethnic groups with respect to substance abuse treatment needs, although epidemiological research suggests that: (2b) there will be significant disparity among socioeconomic and racial/ethnic groups with respect to substance abuse treatment needs. To get at the question of treatment provision and need more exactly, I ask: Are there socioeconomic and racial/ethnic utilization gaps or unmet needs? Expectations are that: (3a) there will be parity among socioeconomic and racial/ ethnic groups’ service utilization or (3b) there will be significant disparities along this dimension. Lastly, the question remains whether there are spatial inequalities in treatment provision, treatment needs and utilization that are systematically related to the characteristics of the geographic places where treatment is provided, need for services exists and utilization gaps emerge. In short, it is expected that: (4) just as disparities in socioeconomic and racial/ethnic treatment provision, need and utilization exist, these disparities flow across communities that are contiguous.
DATA, MEASURES AND METHODS Data To estimate socioeconomic and racial/ethnic disparities in substance abuse treatment provision, need and utilization, I develop models of these
Socioeconomic and Racial/Ethnic Disparities
181
relationships at the county level and over time. This entailed creating an original database of social indicators of community social and economic and racial and ethnic characteristics, substance abuse treatment provision, substance abuse treatment need, and, utilization for 3,141 counties in the United States (including Alaska and Hawaii). Eliminating cases with odd or discrepant data resulted in a total of 3,137 observations. In order to assure the robustness of the outcomes, I examine these relationships for each of the three years in which complete data were available: 2000, 2002 and 2003. Comparing across the three years demonstrates the extent to which the hypothesized relationships are reliable.1 Data were culled from a number of sources including: The National Center for Health Statistics, The Uniform Crime Reports from Inter-University Consortium for Political and Social Research, National Survey of Substance Abuse Treatment Services, The U.S. Census Bureau, Small Area Income and Poverty Estimates. A complete citation is included below with reference to each measure (Table 1).
Dependent Measures Substance Abuse Treatment Provision Substance abuse ‘‘treatment’’ has a number of dimensions encompassing need, provision, access and utilization. Access and utilization are related to the provision of services through factors such as ability to pay for care (e.g., private and public insurance), location (e.g., distance to services, wait times), management and delivery of services. However, provision of services is primarily a matter of availability and for populations that experience high levels of residential segregation, geographic accessibility is the first step in successful utilization of services. In this study, treatment provision is conceptualized as the per capita rate of availability of organizational services relevant to substance abuse including inpatient and outpatient residential services and detoxification in hospitals, freestanding clinics, and halfway houses (Department of Health and Human Services, National Survey of Substance Abuse Treatment Services, 2004). I measure rate of treatment provision by counting the number of for-profit, nonprofit and government treatment organizations per 10,000 population members using the National Survey of Substance Abuse Treatment Services conducted by U.S. Substance Abuse and Mental Health Services Administration an agency of the Department of Health and Human Services, Office of Applied Studies, in 2000, 2002 and 2003. These organizational providers are distributed
182
MATTHEW E. ARCHIBALD
Table 1. Descriptive Statistics of Data Used in Spatial Regression Models. U.S. Counties 2000, 2002, 2003. Year
Mean
SD
Dependent variables/definitions Treatment 2000 provision 2002 2003 Substance abuse 2000 treatment need 2002 2003 Utilization/unmet 2000 need 2002 2003
1.902 1.847 2.069 8.173 13.475 10.537 0.000 0.000 0.000
2.387 2.387 2.589 5.230 7.585 6.208 5.222 7.560 6.173
0.091 0.091 0.091 0.062 0.066 0.068 13.304 13.753 13.371 36,366 35,890 36,882 4.760 5.928 6.183 10.999 10.858 10.686
0.146 0.145 0.145 0.120 0.122 0.123 5.589 5.592 4.919 9,002 9,401 9,303 2.617 2.679 2.779 3.878 4.571 3.713
0.000 0.000 0.000 0.001 0.001 0.001 1.7 2.3 2.2 15,231 15,025 17,136 0.0 0.7 1.3 0.0 0.0 0.0
0.865 0.861 0.865 0.975 0.974 0.974 42.2 49.1 36.4 91,210 93,927 93,383 27.6 25.3 24.2 49.933 85.861 50.921
0.795
0.066
0.419
0.966
0.149
0.058
0.000
0.446
Explanatory variables/definitions Proportion African 2000 American 2002 2003 Proportion 2000 Hispanic 2002 2003 Poverty rate – 2000 percent below 2002 poverty line 2003 Median income 2000 2002 2003 Unemployment rate 2000 2002 2003 2000 Education 2002 attainment – high 2003 school graduation rate per 1,000 population Proportion married2000 headed households Proportion female2000 headed households Housing unit 2000 density per square mile Proportion 18–29 2000 year olds Number of cases
3,137
86.5
0.151
340.9
0.045
Min
Max
0 0 0 0 0 0 9.13 18.13 14.75
29 29 29 40 40 40 32.05 27.00 29.98
0
0.002
4900
0.691
Socioeconomic and Racial/Ethnic Disparities
183
differentially across social and geographic space. To control for wide differences in county populations and racial and ethnic groups, rates are standardized on a z-score scale and then re-scaled from 0 to 100. While 0 is the lowest observed rate and 100 is the highest by which the distribution is scaled, treatment provision tends to be skewed towards the lower end of the continuum since most counties have only 1 or 2 providers. Consequently, the three or four highest counties were down-scaled so that a score of 22 represents the greatest amount of per capita treatment provision possible.
Substance Abuse Need There is no consensus in the research literature on indicators of areal characteristics linked to substance abuse treatment need although there are a number of possibilities (Herman-Stahl et al., 2001). I conceptualize the need for treatment as the necessity of professional care for serious adverse effects of an alcohol or drug-use disorder. I then apply this concept to aggregate data for the years 2000, 2002 and 2003 measuring mean rates of drug and alcohol mortality (per 100,000), and alcohol and drug arrests. This is known as the social indicator approach. McAuliffe and Dunn (2004), McAuliffe et al. (2002), Gregoire (2002) and others, create distinct measures of alcohol and drug abuse and then combine their measures with mortality data to form a substance abuse need index encompassing both types of addictions. The source of mortality data for this study is the Compressed Mortality File provided by the Department of Health and Human Services, Center for Disease Statistics, National Center for Health Statistics (2003). It is a county-level file containing national mortality data spanning the years 1968–2003. As McAuliffe and Dunn (2004) explain a person who died from alcohol and drug-induced causes probably had a severe substance abuse problem that needed treatment. For substance abuse arrest data, this study culled county-level arrests from the Uniform Crime Report available through the Inter-University Consortium for Political and Social Research, University of Michigan (2006). Alcoholrelated abuse is defined as driving under the influence and liquor law violations, including public intoxication. Drug-related abuse stems from arrests for possession and sale of illicit drugs. Again, as McAuliffe and Dunn (2004) argue, being arrested for drug and alcohol use is a symptom of an abuse disorder according to the American Psychiatric Association (1994). These measures have been validated at varying levels of aggregation including across counties and states.
184
MATTHEW E. ARCHIBALD
Substance abuse need is a composite index of per capita mortality and alcohol and drug arrests for each of the three years. The index is standardized by creating z-scores and then converting these to the same metric as the substance abuse treatment provision scores which range from 0 to 100. Zero indicates no substance abuse mortality and arrest for the county and 100 equals a score that is the highest observed rate. Using Cronbach’s alpha, the reliability of the raw components of the index for each year was assessed at .67, .65 and .65, respectively. The scales were refashioned, like the treatment provision scores, to fit the distribution, so that the observed range was actually 0–40, rather than 100. Substance Abuse Provision Gap – Unmet Need Measures of unmet need developed in survey studies usually explore the gap between individuals’ need for alcohol or drug abuse treatment and whether or not they have received treatment. This can take the form of perceived needs (Wells et al., 2001) or responses to questions on standard measures of substance dependence such as those provided in the DSM-IV and incorporated into the NHSDA (2002). In McAuliffe’s work, gaps between provision and need imply failure of service supply. The utilization or unmet need gap is measured by regressing the standardized index of treatment provision on treatment need and then using the residuals as indicators of the difference. Positive scores indicate that provision lags behind need while negative scores suggest that treatment provision is sufficient to meet the treatment needs of that community.
Independent Measures Areal Disadvantage Socioeconomic and racial/ethnic disparities in treatment provision reflect underlying environmental processes such as the pressures that markets place on decision-making with regard to locating facilities for either competitive advantage or to meet demand. Given these assumptions, it is possible to measure the level of social and economic disadvantage, and racial/ethnic composition of each county using a number of indicators related to health care provision and expected to influence the provision of substance abuse treatment. Race and ethnicity are primary indicators of disadvantage and measured in several ways. Using the Census Bureau, National Center for Health Statistics’ Bridged-race Vintage, 2003 Postcensal Population Estimates for
Socioeconomic and Racial/Ethnic Disparities
185
July 1, 2000–July 1, 2003, I calculate the proportion nonwhite/nonHispanics (African American) for each county as well as the proportion of Hispanics. At the same time, socioeconomic disadvantage, both independent of and in combination with race and ethnicity, is expected to shape substance abuse treatment needs and provision by acting as a barrier to investment in the community. While greater socioeconomic disadvantage may signal the need for more human services such as substance abuse treatment, it is rather likely to act as a barrier to investment. There are a number of standard indicators drawn from the literature that measure this construct. The rate of poverty, median income and unemployment rate are primary indicators of disadvantage as is educational attainment. Poverty, median income and unemployment were drawn from the U.S. Census Bureau, Small Area Income & Poverty Estimates – Model-based Estimates for States, Counties & School Districts, 2000, 2002, 2003. Education data were drawn from the Department of Education, National Center for Education Statistics (2004). Educational attainment is measured by the percent high school graduates and reflects community capacity – the extent to which local institutions contribute to the social and economic well-being of the area. Additional indicators of socioeconomic disadvantage include the proportion of the population that is not married/married, the percent of female-headed households as well as the age structure, measured by the proportion of youth in the population between 18 and 29. While most health care research shows that this population has less of a health disadvantage than other age and gender strata (Robert & House, 2000), this population has the highest substance abuse treatment needs (NHSDA, 2002). These data were obtained from the U.S. Census Bureau’s, Census (2000). Overcrowding in housing because of segregation and inequities in availability is likely to have a direct effect on risk factors associated with poor health (O’Campo, Xue, Wang, & Caughy, 1997) as well as lowering property values and acting as a barrier to potential investors. Community capacity is therefore diminished, limiting social and economic investment in health care treatment services, especially private substance abuse rehabilitation. In the aggregate, socioeconomic and racial/ethnic disparities in health will vary with the degree to which places are located in highly concentrated, urban areas, or less concentrated, rural ones. This type of geography can play an important role in provision of substance abuse treatment services which influences health outcomes as Jacobson (2004a) and others (e.g., McAuliffe & Dunn, 2004) have shown. The NHDR (2004), for example, argues that rural populations are more likely to have worse access
186
MATTHEW E. ARCHIBALD
to health care services and therefore poorer quality health care, a finding consistent with the concentration of unmet substance abuse need in rural sections of the country shown in McAuliffe’s work. To maximize these features of disadvantage, I measure the degree of urbanicity using housing density per square mile from the Department of Health and Human Services, Health Resources and Services Administration, Area Resource File, 2005.
Methods Analyzing the disparities of treatment provision across counties raises a crucial geographic assumption that things nearby are more related than things far away. One important implication is that observations will cease to be independent of one another, they will show autocorrelation, violating an important assumption of ordinary least squares regression that the residuals are uncorrelated.2 In addition, when observations are geographically proximate, the error variances will be heteroscadastic.3 To overcome these problems, spatial regression models are used in the following analyses to account for spatial dependence; the issue of serial autocorrelation is adjusted by inclusion of a lagged dependent variable, rho (r), and spatial error is adjusted through the use of a special error term, lambda (l). Spatial regression is ideally suited for modeling substance abuse and treatment data since the central units of analysis share contiguous boundaries with one another (e.g., neighborhoods, census tracts, counties or states). It is expected that substance abuse treatment provision diffuses across contiguous boundaries, and the inconsistencies and bias produced by lack of independence can be adjusted by explicitly modeling spatial error. The basic model from Anselin (1988) is: y ¼ rWy þ ¼ lW þ x
ð1Þ
where y is a vector of observations on a dependent variable, Wy a spatially lagged endogenous variable with a weights matrix W, r the spatial autoregressive parameter for the spatially lagged dependent variable, e a vector of error terms, We a spatially lagged error term with spatial weights matrix W, l the spatial autoregressive parameter for the spatially lagged error term, and xB N(0,O), where Oii=hi(za). When a=0, h=s2, and the errors are homoscedastic. Maximum likelihood estimates are used to model the spatially lagged dependent variable and the autoregressive error term.
Socioeconomic and Racial/Ethnic Disparities
187
Exploratory analyses showed that poverty, unemployment and median income are more highly related with one another than with the outcome variables. The prime suspect is median income which tends to fluctuate downwards as both poverty and unemployment increase. In addition, the proportion of female-headed households and those headed by married couples strongly covary. In the regressions that follow, I therefore orthogonalize median income and poverty, and female- and married-headed households using a modified Gram–Schmidt procedure. Both GeoDa (Anselin, 2004), a program that facilitates analysis of spatially distributed data, and Stata (StataCorp 8.0, 2003) were used in the analyses.
RESULTS Tables 2, 3 and 4 present results of spatial regression analyses of socioeconomic, racial and ethnic differences in substance abuse treatment provision, need and utilization, taking into account the underlying diffusion of these characteristics between communities. To address the issue, Tables 2, 3 and 4 consist of two types of models: spatial error models and spatial lag models. First, I developed the diagnostics reported at the bottom of each table. These are based on an OLS model and determine whether error or lag models should be fitted. Initial results across the three specifications showed that there was considerable spatial dependence and heterogeneity suggesting the importance of modeling both error and lag specifications. Second, since this was the case, Tables 2, 3 and 4 contain a parameter (lambda, l) assessing the degree of spatial autoregressive error for each treatment, treatment need and utilization outcome, 2000, 2002 and 2003. The presence of a significant spatial error term indicates correlated errors among neighboring communities. The models also include maximum likelihood estimates of a weighted spatial lag term (rho, r), representing the dependent variable of interest. The lag term reflects the extent to which commonalities in the covariates are due to spillovers across neighboring counties. A significant spatial lag parameter is evidence that the original OLS estimate (not shown) of the impact of the covariates on the outcomes is overstated and must be adjusted (Anselin, 1988). Inclusion of the weighted lag variable contributes to the adjustment. Its significance in all models implies that some of the influence of racial/ethnic and socioeconomic covariates of substance abuse treatment provision need and utilization is due to proximity and that the pattern of relationships between covariates and outcomes is
188
MATTHEW E. ARCHIBALD
Table 2. Regression Coefficients and Standard Errors for Racial/Ethnic and Socioeconomic Analyses of Substance Abuse Treatment Provision: U.S. Counties 2000, 2002, 2003. Treatment Provision Spatial Error 2000 Explanatory variables Proportion 5.507 (.596) African American Proportion 1.769 Hispanic (.495) .086 Poverty rate – proportion (.063) below poverty line Median income .128 (.054) Unemployment .000 (.016) rate Education .033 (.012) attainment – high school graduation rate per 1,000 Proportion .166 (.046) married-headed households 2000 .730 Proportion female-headed (.092) households 2000 Housing unit .001e1 density per (.000) square mile in 2000 Proportion 18–29 5.702 (.952) year olds Lambda (l) .302 (.026) Rho (r) Constant
1.270 (.238) 127.7
Log likelihood ratio (model vs. OLS baseline) AIC 14027.4
2002
Spatial Lag 2003
2000
2002
2003
5.024 (.599)
5.431 (.664)
4.661 (.522)
4.130 (.521)
4.413 (.567)
1.782 (.470) .079 (.060)
1.209 (.535) .167 (.077)
1.464 (.398) .056 (.057)
1.358 (.369) .068 (.054)
.867 (.408) .167 (.066)
.079 (.056) .053 (.020) .002 (.010)
.063 (.064) .048 (.021) .028 (.014)
.094 (.045) .002 (.016) .037 (.012)
.079 (.045) .043 (.017) .002 (.009)
.048 (.050) .041 (.018) .033 (.013)
.254 (.046)
.141 (.050)
.179 (.043)
.264 (.043)
.155 (.047)
.639 (.091)
.766 (.107)
.648 (.088)
.530 (.088)
.653 (.102)
.008e2 (.000)
.007e2 (.000)
.001e1 (.000)
.007e2 (.000)
.007e2 (.000)
6.033 (.952) .312 (.026)
6.522 (1.037) .351 (.026)
4.946 (.942)
5.202 (.938)
5.625 (1.019)
1.216 (.241) 137.3
1.060 (.284) 180.9
.277 (.026) .734 (.231) 115.1
.293 (.026) .708 (.229) 131.0
.331 (.025) .392 (.265) 170.4
13976.9
14498.0
14041.9
13985.2
14510.5
Socioeconomic and Racial/Ethnic Disparities
Table 2.
189
(Continued ) Spatial Model Diagnosticsa
Moran’s I Robust Lagrange multiplier (spatial error) Robust Lagrange multiplier (spatial lag) Number of cases
2000
2002
2003
12.594 .173
13.097 1.573
15.161 0.370
10.244
3.615
6.620
3,137
3,137
3,137
3,137
3,137
3,137
Notes: Numbers in parentheses are standard errors. a Diagnostic tests are run on the OLS model prior to analyzing the error and lag models. po.05. po.01. po.001: two-tailed tests.
more pronounced between communities than within. More specifically, the spatial multiplier rho indicates the precise degree (i.e., 1/1r) by which the estimated marginal effect of the covariates is increased (see Mobley, Root, Anselin, Lozano, & Koschinsky, 2006 for a detailed discussion). Consequently, the spatial hypothesis (Hypothesis 4) is supported by these data. I discuss its implications for the other findings in the concluding section. With respect to socioeconomic, racial and ethnic differences in substance abuse treatment provision, Table 2 provides evidence to support several claims. Communities with a greater proportion of African Americans as well as those with a high proportion of Hispanics are less likely to have treatment services provided. This finding is quite robust: it is consistent across all three years and for both types of model specifications. Other robust findings include the negative relationship between the proportion of houses headed by married couples and treatment provision, and the positive weight that communities with female-headed households and younger residents have on proximity of services. Since female-headed households, those run by single parents, and high concentrations of youth are associated with resource disadvantage (Brooks-Gunn, Duncan, & Aber, 1997), it appears that there is some evidence that treatment services are being targeted to those places with greater disadvantage – and therefore some parity has been achieved (suggested by Hypothesis 2b). Trends in unemployment (positive with provision) and median income (negative) further support this
190
MATTHEW E. ARCHIBALD
Table 3. Regression Coefficients and Standard Errors for Racial/Ethnic and Socioeconomic Analyses of Substance Abuse Treatment Needs: US Counties 2000, 2002, 2003. Substance Abuse Treatment Needs Spatial Error 2000
2002
Spatial Lag 2003
2000
2002
2003
Explanatory variables Proportion African 7.657 12.905 11.847 6.035 1.131 9.053 (1.342) (1.967) (1.618) (1.043) (1.517) (1.259) American Proportion 1.342 11.790 7.804 7.456 8.070 5.345 Hispanic (1.182) (1.662) (1.374) (.829) (1.111) (.927) Poverty rate – .663 .572 .396 .562 .481 .271 proportion below (.135) (.186) (.183) (.114) (.160) (.148) poverty line Median income .251 .263 .428 .076 .394 .469 (.126) (.188) (.162) (.090) (.133) (.114) Unemployment rate .005 .250 .214 .015 .221 .183 (.033) (.062) (.049) (.032) (.051) (.041) .032 .045 .036 .001 .027 .019 Education (.026) (.029) (.032) (.024) (.028) (.029) attainment – high school graduation rate per 1,000 Proportion .176 .252 .184 .121 .134 .112 (.096) (.140) (.115) (.086) (.126) (.104) married-headed households 2000 Proportion female1.941 2.735 2.246 1.639 2.326 1.854 headed (.189) (.273) (.242) (.177) (.258) (.227) households 2000 .001 e-1 .001e1 .001e1 .002e1 .001e1 .001e1 Housing unit (.000) (.000) (.000) (.000) (.000) (.000) density per square mile in 2000 1.844 19.740 15.348 Proportion 18–29 8.006 15.070 12.054 year olds (1.909) (2.787) (2.312) (1.905) (2.762) (2.282) Lambda (l) .503 .515 .509 (.022) (.022) (.022) Rho (r) .466 .478 .474 (.022) (.022) (.022) Constant 6.896 12.844 1.476 3.325 6.775 5.536 (.498) (.743) (.660) (.487) (.726) (.634) Log likelihood ratio 439.6 475.8 455.1 396.7 432.2 414.3 (model vs. OLS baseline) AIC 18506.6 20812.4 19602.6 18551.5 20858.0 19645.4
Socioeconomic and Racial/Ethnic Disparities
Table 3.
191
(Continued ) Spatial Model Diagnosticsa
Moran’s I Robust Lagrange multiplier (spatial error) Robust Lagrange multiplier (spatial lag) Number of cases
2000
2002
2003
24.304
25.545
.491
0.149
24.808 0.280
44.984
46.806
42.860
3,137
3,137
3,137
3,137
3,137
3,137
Notes: Numbers in parentheses are standard errors. a Diagnostic tests are run on the OLS model prior to analyzing the error and lag models. po .05. po.01. po.001: two-tailed tests.
possibility. Our measure of the poverty rate, however, was not significant and tended to be negatively related to service provision. It may be that substance abuse treatment is limited to communities with greater substance abuse treatment need given the high cost of making services more widely available, which would explain the disparities in provision. Table 3 addresses the issue by examining socioeconomic and racial/ethnic differentials in substance abuse need, as measured by the extent of community use and abuse. Interestingly, counties with more African American residents have lower levels of treatment need, and as Table 2 showed, matched by lower levels of provision, confirm results of individual-level surveys reporting parity between blacks and whites with regard to service provision (NHDR, 2004). Hispanics, on the other hand, have consistently higher rates of treatment service-needs (as well as bigger gaps in provision), consistent with expectations that there are disparities in substance abuse treatment needs that emerge along racial/ethnic lines (Hypothesis 2a). As for socioeconomic indicators of community disadvantage, in Table 3, poverty is consistently related to greater treatment need – but in a direction that suggests that greater affluence and less disadvantage fosters higher rates of treatment need. Similarly, although less robustly, as the unemployment rate climbs, substance abuse treatment need declines. In contrast, again, a robust relationship holds between female-headed households, youth and treatment needs. Communities with a greater
192
MATTHEW E. ARCHIBALD
Table 4. Regression Coefficients and Standard Errors for Racial/Ethnic and Socioeconomic Analyses of Utilization/Unmet Treatment Needs: U.S. Counties 2000, 2002, 2003. Utilization/ Unmet Treatment Needs Spatial Error 2000
2002
Spatial Lag 2003
2000
2002
2003
Explanatory variables Proportion African 7.020 11.652 1.536 4.661 9.187 8.036 (1.342) (1.962) (1.612) (.522) (1.511) (1.249) American Proportion 1.533 12.266 8.102 1.464 8.316 5.478 Hispanic (1.182) (1.658) (1.369) (.398) (1.109) (.922) Poverty rate – .676 .548 .361 .056 .463 .235 proportion below (.135) (.186) (.182) (.057) (.159) (.147) poverty line Median income .272 .243 .404 .094 .370 .450 (.126) (.188) (.162) (.045) (.133) (.113) Unemployment rate .005 .262 .224 .002 .229 .190 (.033) (.062) (.049) (.016) (.051) (.041) .035 .043 .042 .037 .028 .026 Education (.025) (.029) (.032) (.012) (.027) (.029) attainment – high school graduation rate per 1,000 Proportion .190 .307 .211 .179 .191 .143 married-headed (.096) (.139) (.115) (.043) (.126) (.104) households 2000 Proportion female1.853 2.558 2.049 .648 2.192 1.695 headed (.188) (.272) (.241) (.088) (.257) (.225) households 2000 .001e1 .001e1 .001e1 .001e1 .001e1 .001e1 Housing unit (.000) (.000) (.000) (.000) (.000) (.000) density per square mile in 2000 1.101 18.016 13.635 Proportion 18–29 4.946 13.684 1.640 year olds (1.905) (2.780) (2.300) (.942) (2.752) (2.266) Lambda (l) .505 .514 .512 (.022) (.022) (.022) Rho (r) .277 .481 .480 (.026) (.022) (.022) Constant 1.195 .458 .183 .734 .102 .231 (.498) (.741) (.657) (.231) (.659) (.580) Log likelihood ratio 445.6 476.2 461.0 115.1 439.3 428.6 (model vs. OLS baseline) AIC 18497.8 20798.3 19571.1 14041.9 20837.2 19605.5
Socioeconomic and Racial/Ethnic Disparities
Table 4.
193
(Continued ) Spatial Model Diagnosticsa
Moran’s I Robust Lagrange multiplier (spatial error) Robust Lagrange multiplier (spatial lag) Number of cases
2000
2002
2003
24.516
25.603
0.103
0.052
25.026 0.208
39.607
37.410
30.127
3,137
3,137
3,137
3,137
3,137
3,137
Notes: Numbers in parentheses are standard errors. a Diagnostic tests are run on the OLS model prior to analyzing the error and lag models. po .05. po.01. po.001: two-tailed tests.
proportion of female-headed households and youth have greater treatment needs. While Table 3 can only address the treatment provision – treatment need gap indirectly, Table 4 investigates disparities in utilization explicitly. Consistent with our previous findings, communities featuring a higher proportion of African Americans had lower utilization rates and less unmet need for treatment, supporting NHDR (2004) findings. Similarly, as poverty increases, unmet needs decline (the unemployment rate follows a comparable if less robust negative pattern). In contrast, communities with femaleheadedness and youth have greater unmet needs, and therefore ongoing disparities with regard to getting substance abuse treatment services.
DISCUSSION AND CONCLUSION Using data from U.S. counties for the years 2000, 2002 and 2003, results of spatial analyses of socioeconomic and racial/ethnic disparities in treatment provision, need and utilization provide support for survey research findings suggesting parity in service provision insofar as poorer communities and those with higher concentrations of African Americans had lower unmet substance abuse needs. Yet, significant disparities remain in communities with higher concentrations of Hispanics, youth, and female-headed households.
194
MATTHEW E. ARCHIBALD
These findings were robust across years and models. Communities with high concentrations of Hispanics, youth, and female-headed households, had lower substance abuse treatment provision, greater treatment needs and therefore, the biggest gaps in utilization or unmet need. These findings add nuance to McAuliffe et al.’s (2002) study of sub-state areas, and Allard’s (2004) research on service provision in metropolitan centers, confirming the positive relationship between treatment need and proportion Hispanic and young, but disconfirming expectations that African Americans and impoverished areas (based on poverty rate, median income and unemployment) have the greatest treatment needs. What accounts for the unexpected relationships between treatment need, the proportion African American and socioeconomic disadvantage? First, although McAuliffe et al. (2002) show a positive relationship between proportion black and drug and alcohol treatment needs, and Schmidt et al.(2006) argue that ethnic minority populations have greater treatment needs than whites, debate still surrounds differences in substance abuse treatment needs with implications for service provision. Survey and epidemiological research often show no differences between populations in the extent of dependence, use/abuse, and therefore need for treatment services (see e.g., National Comorbidity Study in which blacks have lower prevalence rates than whites, Kessler et al., 1994). Consequently, studies such as the NHDR (2004) find that regardless of difference in the degree of treatment need (which is not analyzed in the report itself) among those who need care ‘‘blacks and people with less education are more likely than white or college attendees to receive treatment’’ (p. 129). Second, treatment need and socioeconomic disadvantage is equally difficult to pinpoint since it is not apparent that the extent of individuals’ substance dependence and therefore need for treatment services, narrowly defined, is related to socioeconomic status. At the community level, neoclassical economic theory would argue that higher levels of unemployment and lower earnings (e.g., higher rates of poverty) will be associated with a decrease in consumption patterns of these kinds of goods. Saffer and Chaloupka (1999), for example, show that a consistent pattern of negative price-effects across socioeconomic and racial/ethnic groups diminishes drug and alcohol use, which should reduce the need for treatment services. Third, an additional issue emerges concerning the quality of services by which treatment needs are being met. While parity has been achieved for some groups and not for others, it is worthwhile to ask whether the quality of services is equitable. Disadvantaged groups’ prospects in treatment may be diminished because of lack of resources to cover appropriate services.
Socioeconomic and Racial/Ethnic Disparities
195
This leads to higher program attrition rates and higher rates of relapse (Jacobson et al., 2007) which are not reflected in research focused on provision, need and utilization, narrowly defined. Part of the problem is a levels-of-analysis issue. This is an ecological study using areal characteristics, so establishing causal linkages at the individual level is not tenable. There is also the question of accuracy of the social indicators (McRae, Beebe, & Harrison, 2001), since states have differential reporting criteria. McAuliffe’s research, however, has been exceptionally rigorous in validating social indicators, like the ones used in this study, with survey measures. An additional issue raised by the use of community-level measures is the extent to which individual variation needs to be controlled. Spatial autocorrelation might indicate that an unobserved county-level effect has been omitted from the model. Moreover, it is unclear what communitylevel socioeconomic and racial/ethnicity effects there are on health, independent of those of residents of such communities (Robert & House, 2000). Research suggests, with some exceptions, that socioeconomic features of the community, and not just of its individual residents, influence health and mortality (e.g., Krieger, 1991). We might infer that the level of substance abuse could be high because of a skew in the socioeconomic and racial/ethnic characteristics of abusers or because on average most people in the community with certain key characteristics are substance abusers. The issue is important with regard to the discrepancy between individual-level surveys of substance abuse which condition on whether people have received services to address that need, and, more macro-level measures of substance abuse and the systemic provision of services. The solution to all of these issues lies with multilevel analysis that controls for treatment outcomes at the individual level and incorporates their consequences into population-level models. Mapping client characteristics such as geographic origin and merging this information with characteristics of substance abuse offenders would enhance our ability to understand these relationships immeasurably. Future research should address the issue by combining survey and epidemiological data. In this way, more careful operationalization of what constitutes ‘‘treatment need’’ and its provision can be developed. If public resources are allocated to areas on a per capita basis rather than on the basis of some other compelling formula such as high-risk or unmet need, then some areas will have less access to critical services even when their unmet needs are high (McAuliffe et al., 2002). From the point of view of public policy, developing allocation criteria that reflect differential population needs will be important in reduction of inequalities in access to
196
MATTHEW E. ARCHIBALD
treatment and retention in treatment translating into better outcomes for minority populations. Therefore, studies that examine gaps in treatment provision, and, seek to describe its socioeconomic and racial/ethnic dimensions make an important contribution towards solving the persistent dilemma of how to provide adequate health care in places that have a great deal of unmet treatment needs. In a broad sense, the theoretical background for the study of disparities in substance abuse treatment provision begs the questions: Does the location of public service organizations directly contribute to provision of community goods, in this case, treatment of substance abuse, and, what are the factors that determine where (and sometimes, when) new organizations will be established to meet these needs? Both sets of questions focus on the spatial ecology of organizations and their embeddedness in community settings. Since socioeconomic and racial/ethnic disparities in treatment provision reflect underlying institutional processes, both economic and political, that foster the unequal distribution of services, it will therefore require considerable political will to rectify these disparities. Studies extending this framework and asking questions about the supply and demand of health care services can provide firm evidence upon which to make decisions ameliorating these inequities.
NOTES 1. There are several gaps in the data that I have filled: (1) Florida has no UCR data with ICPSR, but can be located at http://www.fdle.state.fl.us/fsac/data_statistics. asp#UCR%20Arrest%20Data; Source: Florida Department of Law Enforcement (2001). Crime in Florida, 2000 Florida uniform crime report (Computer program). Tallahassee, FL: FDLE. (2) Illinois has UCR data for major urban counties like Cook but not the smaller ones – these smaller counties are treated as if they had no arrests in these years. (3) The mortality data from the CDC is ‘‘suppressed,’’ because the number of cases per county is too small for reliable estimation, however, I combined the three years of interest, 2000, 2002, 2003 and calculated the average rate for each county before merging mortality with UCR data on arrests. (4) Finally, there were two cases whose standardized scores on substance abuse treatment provision was rescaled because their contribution dwarfed all other counties with respect to the ratio of treatment provision to population size. These are the Dade and Hillsborough counties in Florida. See discussion of dependent variables in the next section. 2. Autocorrelation is problematic in that while U and X are supposed to influence Y, with autocorrelation Ut1 influences Ut and then Y: the errors are not unrelated. 3. To illustrate the practical implication of this error, imagine there is a problem estimating the error term for treatment service provision in poor counties but not the affluent counties. The Y values (treatment service provision) that correspond to the poor counties will therefore fit worse than those for the affluent counties.
Socioeconomic and Racial/Ethnic Disparities
197
ACKNOWLEDGMENTS The author would like to thank Julia Koschinsky for providing the GIS shapefiles used in spatial analysis, and Kendralin J. Freeman and Sonal Nalkur for their help in organizing the data. Thanks especially to Jennie Kronenfeld for her feedback and support.
REFERENCES Adler, N. E., & Newman, K. (2002). Socioeconomic disparities in health: Pathways and policies. Health Affairs, 21(2), 69. Allard, S. W. (2004). Access to social services: The changing urban geography of poverty and service provision. Washington, DC: The Brookings Institution, Center on Urban and Metropolitan Policy, Survey Series. Alliance for Health Reform. (2003). Closing the gap: Racial and ethnic disparities in health care. Washington, DC: Alliance for Health Reform. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: American Psychiatric Association. Anselin, L. (1988). Spatial econometrics, methods and models. Dordrecht: Kluwer. Anselin, L. (2004). GeoDa 0.95i release notes. Spatial Analysis Laboratory (SAL). Department of Agricultural and Consumer Economics, University of Illinois, Urbana-Champaign, IL. Barker, P.R., Epstein, J. F., Hourani, L. L., Gfroerer, J., Clinton-Sherrod A. M., West N., & Shi, W. (2004). Patterns of mental health service utilization and substance use among adults, 2000 and 2001. Substance Abuse and Mental Health Services Administration, Office of Applied Studies, DHHS Publication No. SMA 04-3901, Analytic Series A-22. Rockville, MD. Beshai, N. (1984). Assessing needs of alcohol-related services: A social indicators approach. American Journal of Drug and Alcohol Abuse, 10(03), 417–427. Brady, T. M., & Ashley, O. S. (Eds). (2005). Women in substance abuse treatment: Results from the Alcohol and Drug Services Study (ADSS). Substance Abuse and Mental Health Services Administration, Office of Applied Studies, DHHS Publication No. SMA 04-3968, Analytic Series A-26. Rockville, MD. Brooks-Gunn, J., Duncan, G. J., & Aber, J. L. (Eds). (1997). Neighborhood poverty. New York: Russell Sage Foundation. Crimmins, E. M., Hayward, M. D., & Seeman, T. E. (2004). Race/ethnicity, socioeconomic status, and health. In: N. B. Anderson, R. A. Bulatao & B. Cohen (Eds), Critical perspectives on racial and ethnic differences in health in late life (pp. 310–352). Washington, DC: The National Academies Press, National Academy of the Sciences. Daley, M. (2005). Race, managed care, and the quality of substance abuse treatment. Administration and Policy in Mental Health, 32(4), 457–476. DiMaggio, P. J., & Anheier, H. K. (1990). The sociology of nonprofit organizations and sectors. Annual Review of Sociology, 16, 137–159. Florida Department of Law Enforcement. (2001). Retrived 2/5/07 from http://www.fdle.state.fl.us/ fsac/data_statistics.asp#UCR%20Arrest%20
198
MATTHEW E. ARCHIBALD
Freisthler, B., & Gruenewald, P. J. (2005). Social ecology and the invention of new regulatory strategies for preventing drug and alcohol problems. In: T. Stockwell, P. J. Gruenewald, J. W. Toumbourou & W. Loxley (Eds), Preventing harmful substance use: The evidence base for policy and practice (pp. 291–306). Chichester, West Sussex, England: Wiley. Freisthler, B., Gruenewald, P. J., Johnson, F. W., Treno, A. J., & LaScala, E. A. (2005). An exploratory study examining the spatial dynamics of illicit drug availability and rates of drug use. Journal of Drug Education, 35(1), 15–27. Gregoire, T. K. (2002). The validity of a social indicator approach to substance misuse needs assessment. Substance Use and Misuse, 37(February 3), 357–380. Greve, H. R. (2002). An ecological theory of spatial evolution: Local density dependence in Tokyo banking, 1894–1936. Social Forces, 80, 847–879. Gruenewald, P. J., Freisthler, B., Remer, L., LaScala, E. A., & Treno, A. (2006). Ecological models of alcohol outlets and violent assaults: Crime potentials and geospatial analysis. Addiction, 101(5), 666–677. Harrison, J. P., & Sexton, C. (2004). The health care manager. Frederick, 23(3), 192. Heflinger, C. S., Chatman, J., & Saunders, R. C. (2006). Racial and gender differences in utilization of Medicaid substance abuse services among adolescents. Psychiatric Services, 57, 504–511. Herman-Stahl, M., Wiesen, C. A., Flewelling, R. L., Weimer, B. J., Bray, R. M., & Rachal, J. V. (2001). Using social indicators to estimate county-level substance use intervention and treatment needs. Substance Use and Misuse, 36(March 4), 501–521. Howard, D. L., LaVeist, T. A., & McCaughrin, W. C. (1996). The effect of social environment on treatment outcomes in outpatient substance misuse treatment organizations: Does race really matter? Substance Use and Misuse, 31(5), 617–638. Jacobson, J. O. (2004a). Place and attrition from substance abuse treatment. Journal of Drug Issues, 34(Winter), 23–50. Jacobson, J. O. (2004b). The ecological context of substance abuse treatment outcomes: Implications for NIMBY disputes and client placement decisions. Dissertation, The Rand Graduate School, Santa Monica, CA. Jacobson, J. O. (2006). Do drug treatment facilities increase clients’ exposure to potential neighborhood-level triggers for relapse? A small-area assessment of a large, public treatment system. Journal of Urban Health, 83(2), 150–161. Jacobson, J. O., Robinson, P., & Blumenthal, R. N. (2007). A multilevel decomposition approach to estimate the role of program location and neighborhood disadvantage in racial disparities in alcohol treatment completion. Social Science and Medicine, 64(January 2), 462–476. Kaiser Family Foundation. (2003). Key facts: Race ethnicity and medical care. Washington, DC: Kaiser Family Foundation. Kawachi, I., Daniels, N., & Robinson, D. (2005). Health disparities by race and class: Why both matter. Health Affairs, 24(March–April 2), 343–352. Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., Hughes, M., & Eshelman, S. (1994). Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity study. Archives of General Psychiatry, 51, 8–19. Krieger, N. (1991). Women and social class: A methodological study comparing individual, household, and census measures as predictors of black/white differences in reproductive history. Journal of Epidemiology of Community Health, 45, 35–42.
Socioeconomic and Racial/Ethnic Disparities
199
Kronenfeld, J. J. (2005). Health care disparities and underserved populations: Issues of social factors. In: J. J. Kronenfeld (Ed.), Health care services, racial and ethnic minorities and underserved populations patient and provider perspectives, research in the sociology of health care (Vol. 23, pp. 3–13). Netherlands: Elsevier Ltd. Link, B. G., & Phelan, J. (1995). Social conditions as fundamental causes of disease. Journal of Health and Social Behavior, 35(extra issue), 80–94. Lipton, R., Gorman, D. M., Wieczorek, W. F., & Gruenewald, P. (2003). The application of spatial analysis to the public health understanding of alcohol and alcohol-related problems. In: O. A. Khan & R. Skinner (Eds), Geographic information systems and health applications (pp. 57–79). Hershey, PA: Idea Group Publishing. McAuliffe, W. E., & Dunn, R. (2004). Substance abuse treatment needs and access in the USA: Interstate variations. Addiction, 99, 999–1014. McAuliffe, W. E., Woodworth, R., Zhang, C., & Dunn, R. (2002). Identifying substance abuse treatment gaps in sub-state areas. Journal of Substance Abuse Treatment, 23(33), 199–208. McRae, J. A., Jr., Beebe, T. J., & Harrison, P. A. (2001). Estimating the prevalence of substance abuse with social indicators. Journal of Drug Issues, 31(4, Fall), 977–988. Mobley, L., Root, E., Anselin, L., Lozano, N., & Koschinsky, J. (2006). Spatial analysis of elderly access to primary care services. International Journal of Health Geographics, 5(19). Morenoff, J. D., & Lynch, J. (2004). What makes a place healthy? Neighborhood influences on racial/ethnic disparities in health over the life course. In: N. B. Anderson, R. A. Bulatao & B. Cohen (Eds), Critical perspectives on racial and ethnic differences in health in late Life (pp. 406–449). Washington, DC: National Academies Press. National Healthcare Disparities Report (NHDR): Summary. February 2004. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/qual/nhdr03/ nhdrsum03.htm O’Campo, P., Xue, X., Wang, M., & Caughy, M. O. (1997). Neighborhood risk factors for lowbirthweight in Baltimore: A multilevel analysis. American Journal of Public Health, 87, 1113–1118. Robert, S. A., & House, J. A. (2000). Socioeconomic inequalities in health: Integrating individual-, community-, and societal-level theory and research. In: G. L. Albrecht, R. Fitzpatrick & S. C. Scrimshaw (Eds), Handbook of social studies in health and medicine (pp. 115–135). London: Sage. Saffer, H., & Chaloupka, F. (1999). State drug control and illicit drug participation. NBER Working Papers 7114. National Bureau of Economic Research, Inc. Saxe, L. K. C., Beveridge, A., Livert, D., Tighe, E., Rindskopf, D., Ford, J., & Brodsky, A. (2001). The visibility of illicit drugs: Implications for community-based drug control strategies. American Journal of Public Health, 91(12), 1987–1994. Schmidt, L., Greenfield, T., & Mulia, N. (2006). Unequal treatment: Racial and ethnic disparities in alcoholism treatment services. Alcohol Research Health, 29(1), 49–54. Silver, E., Mulvey, E. P., & Swanson, J. W. (2002). Neighborhood structural characteristics and mental disorder: Faris and Dunham revisited. Social Science and Medicine, 55(8), 1457–1470. Smaje, C. (2000). Race, ethnicity, and health. In: D. E. Bird, P. Conrad & A. M. Fremont (Eds), Handbook of medical sociology (pp. 114–128). Upper Saddle River: Prentice-Hall, Inc. Smedley, B. D., Stith, A. Y., Alan, R., & Nelson, A. R. (Eds). (2003). Unequal treatment: Confronting racial and ethnic disparities in health care. Institute of Medicine. Washington, DC: National Academies Press.
200
MATTHEW E. ARCHIBALD
StataCorp. (2003). Stata Reference Manuals Release 8, Stata Press, College Station TX. Storr, C. L., Chen, C. Y., & Anthony, J. C. (2004). Unequal opportunity: Neighborhood disadvantage and the chance to buy illegal drugs. Journal of Epidemiology and Community Health, 58, 231–237. Substance Abuse and Mental Health Services Administration. (2002). Results from the 2001 National Household Survey on Drug Abuse: Volume I. Summary of national findings. DHHS Publication No. SMA 02–3758, NHSDA Series: H–17. Rockville, MD. Sunshine, J., & Bhargavan, M. (2006). Location decisions of mammography facilities: Racial/ ethnic disparities across zip-codes. Paper presented at the annual meeting of the Economics of Population Health: Inaugural Conference of the American Society of Health Economists, TBA, Madison, WI, USA, Jun 04. Uniform Crime Reports from Inter-University Consortium for Political and Social Research (n.d.) Retrieved February 25, 2006, from http://www.icpsr.umich.edu. Note: (n.d.)=no date U.S. Census Bureau. Census 2000, Summary File 1, generated by Sonal Nalkur, using American FactFinder, http://factfinder.census.gov (20 February 2004). U.S. Census Bureau, National Center for Health Statistics’ Bridged-race Vintage 2003 Postcensal Population Estimates for July 1, 2000–July 1, 2003. U.S. Department of Education, National Center for Education Statistics, generated by Sonal Nalkur, http://nces.ed.gov/surveys/sdds/disclaimer.asp?t=selectgeo.asp (20 February 2004). U.S. Department of Health and Human Services, Health Resources and Services Administration, Bureau of Health Professions, Area Resource File (ARF). (2005). Rockville, MD. U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, Office of Applied Studies. National Survey Of Substance Abuse Treatment Services (N-Ssats), (2004): [United States] [Computer File]. Icpsr04256-V2. Arlington, VA: Synectics for Management Decisions, Inc. [producer], 2004. Ann Arbor, MI: Interuniversity Consortium for Political and Social Research [distributor], 2006-08-01. U.S. Department of Health and Human Services. Healthy People 2010. (2000). With Understanding and Improving Health and Objectives for Improving Health (2nd ed., Vol. 2.). Washington, DC: U.S. Government Printing Office, November 2000. Ward, E., Jemal, A., & Cokkinides, V. (2004). Cancer disparities by race/ethnicity and socioeconomic status. CA: A Cancer Journal for Clinicians, 54, 78–93. Weisner, C., & Schmidt, L. A. (2001). Rethinking access to alcohol treatment. Recent Developments in Alcoholism, 15(1), 107–136. Wells, K. B., Klap, R., Koike, A., & Sherbourne, C. D. (2001). Ethnic disparities in unmet need for alcoholism, drug abuse, and mental health care. American Journal Psychiatry, 158(12), 2027–2032. Williams, D. R. (1999). Race, SES, and health: The added effects of racism and discrimination. In: N. E. Adler, M. Marmot, B. S. McEwen & J. Stewart (Eds), Socioeconomic status and health in industrial nations: Social, psychological and biological pathways (Vol. 896, pp. 109–114). New York, NY: Annals of the New York Academy of Science.
GEOGRAPHIC DISPARITIES IN ADULT MENTAL HEALTH UTILIZATION AND NEED FOR SERVICE Stephanie L. Ayers, Jennie Jacobs Kronenfeld, Sam S. Kim, Jemima A. Frimpong and Patrick A. Rivers ABSTRACT The purpose of this chapter is to examine geographic variations in utilization and need for mental health services. Data for this study were obtained from the 2002 National Survey of American Families. The total sample size was 23,327 adults of aged 18 years and older. Both logistic and linear regression were used to test the possibility of geographical variations. Disparities were found among the 13 U.S. states examined in this study. Results also showed that the percentage of African Americans, state mental health budgets, and mean length of stay in psychiatric hospitals in the state are important predictors of variations in mental health utilization and need variables. These findings suggest that although individual sociodemographic characteristics are important in examining mental health utilization, state characteristics (especially percentage of Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 201–223 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00009-9
201
202
STEPHANIE L. AYERS ET AL.
African Americans, state mental health laws, and mean length of stay in psychiatric hospitals) are also important predictors of variation in utilization of mental health services.
INTRODUCTION An attempt was made in 1963 by the Community Mental Health Center Act to improve the delivery of services to the socially, culturally, and economically disadvantaged. In 1978, the President’s Commission on Mental Health restated the need for mental health services delivery to disadvantaged groups (Worthington, 1992). According to the 1999 U.S. Surgeon General’s report, over 54 million Americans had a mental disorder, however fewer than eight million received treatment (U.S. Department of Health and Human Services, 1999). Improving delivery of effective, community-based mental health services was also identified as an avenue through which the quality of mental health can be improved. In 2002, The President’s New Freedom Commission on Mental Health was created in order to examine health services delivery systems. One of the goals of the commission is the elimination of disparities in availability of resources and outcomes with the aim of facilitating mental health utilization in communities (Executive Order 13263, 2002). The annual prevalence of mental health disorders for adults 18 and older in the United States is estimated at 22% (Regier et al., 1993) with depression as the leading mental health disorder (National Institute of Mental Health, 2001). In 2002, there were 40 million visits to physician offices for mental disorders (Woodwell & Cherry, 2004), 2.5 million discharges with an average length of stay of about 7 days (DeFrances & Hall, 2004), and about 31,655 suicides attributed to mental disorders (Arias, Anderson, Hsiang-Ching, Murphy, & Kochanek, 2003). Studies show disparities between need and use of mental health services and are greatest among socioeconomically disadvantaged ethnic minority groups (Alegria et al., 2002; Bao & Sturm, 2004; Hines-Martin, Brown-Piper, Sanggil, & Malone, 2003; Vega, Kolody, Aguilar-Gaxiola, & Catalano, 1999). Hispanics and African Americans have significantly lowered use of mental health services when compared to white non-Hispanics, and these services are incongruent with their needs (Alegria et al., 2002; Rosen, Tolman, & Warner, 2004; Scheffler, Zhang, & Snowden, 2002; Vega et al., 1999; Hines-Martin et al., 2003). Looking closer at this relationship, Scheffler and
Geographic Disparities in Adult Mental Health Utilization
203
Miller (1991) found that non-Hispanic Whites use more outpatient services while African Americans use more inpatient mental health services. This relationship remains when income is controlled. Specifically, African Americans who were not poor had significantly less mental health use than Whites of the same income category (Alegria et al., 2002). For African Americans, the low use of mental health services has been attributed to fewer financial resources in any income category, greater mistrust, low availability of minority providers, and experiences of racism within the health care system (Alegria et al., 2002). For Hispanics, the low use of mental health services is attributed to ‘‘language fluency, cultural differences such as self-reliance, access to Medicaid specialty services, y differences in recognition of mental health problems, lower quality of mental health care’’ (Alegria et al., 2002, p. 1550), the use of natural healers, and a denser more nurturing support system (Vega et al., 1999). In addition to racial and ethnic disparities, other variations in mental health utilization have been noted. The level of severity and number of mental health disorders are determinants of using mental health services (Bao & Sturm, 2004; Scheffler et al., 2002; Rosen et al., 2004) and have the strongest effect on utilizing mental health services (Albizu-Garcia, Alegria, Freeman, & Vera, 2001). Income is positively associated with using mental health services and is the strongest determinant of mental health status (Rosen et al., 2004; Williams & Collins, 1995). Individuals with higher levels of educational attainment are least likely to have a need for mental health services, but when the need arises, they are more likely to use mental health services (Rosen et al., 2004). Type of insurance also influences use of mental health services. Historically, private insurance plans have been very restrictive in their coverage of mental health services (Bao & Sturm, 2004). Private health insurance plans that include mental health services provide an increased access to outpatient care (Goldman, McCulloch, & Sturm, 1998). Nonetheless, having any type of health insurance is positively related to using mental health services (Albizu-Garcia et al., 2001). Those ages 65 years and older, who have private health insurance, have lower use of mental health services including both inpatient and outpatient mental health services because older aged individuals are more likely to attribute mental health symptoms as problems of ‘‘old age’’ (Sarkinsian, Lee-Henderson, & Mangione, 2003). Many studies have been conducted on factors that influence need and use of mental health services. Mainly, these studies examined the impact of sociodemographic variables including race/ethnicity as the main independent variable, as noted above. Although the correlation between sociodemographic
204
STEPHANIE L. AYERS ET AL.
variables and utilization and need for mental health services is important, there are other factors that have not been explored in depth. Studies have identified variations in sociodemographic characteristics as predictors of disparity across states. There are, however, limited studies that investigate the impact of geographic factors on use and needs of mental health services for adults. Geographic location influences mental health services utilization when sociodemographic, need, availability, and accessibility are controlled (Sommers, 1989). In general, as compared to the Midwest, those geographic regions of the Northeast, South, and West have less mental health use (Alegria et al., 2002). Specifically, those areas that have ‘‘stressful environments, particularly a poverty-ridden urban one, [produce] proportionally more mental ill-health’’ (Philo, 2005, p. 586). Urban and industrialized areas have greater access to, demand for, and expenditures of mental health services (Scheffler et al., 2002). Variations in services utilization between rural and non-rural patients have been attributed to less availability or accessibility of mental health services in rural areas (Chumbler, Cody, Booth, & Beck, 2001; Sullivan, Jackson, & Spritzer, 1996) which results in an underutilization of mental health services. (Rosen et al., 2004; Chumbler et al., 2001). Conversely, urban and industrialized areas have greater access to, demand for, and expenditures of mental health services (Scheffler et al., 2002). For availability, mental health professionals are limited; therefore, general practitioners are usually the first medical care professional that residents in rural areas go to for mental health symptoms (Chumbler et al., 2001). Typically, rural residents must travel great distances, usually to urban areas, to seek mental health help (Chumbler et al., 2001). In rural areas, family incomes are less than family incomes in urban areas, which makes mental health services less affordable (Chumbler et al., 2001). In addition, because residents in rural areas are more likely to know each other, there is a fear of stigma in receiving local mental health services (Chumbler et al., 2001). These geographical variations have been linked to public mental health policy. Recently, there has been a shift in policies. Public mental health systems, like private mental health systems, have tried to cut costs. One such way that states have adopted cutting costs is through capitation payment systems (Bloom et al., 2002). Capitation payments resulted from the ‘‘inflation of medical costs y the surge in hospital utilization, y the increase of uninsured and indigent heavy service users, y [and] the tax reform initiatives that reduced available dollars for safety net services not covered by Medicaid’’ (Cohen & Bloom, 2000, pp. 65–66). In 2002, at least
Geographic Disparities in Adult Mental Health Utilization
205
32 states had enacted as a capitation payment system. In most instances, capitation resulted in a decentralization of mental health funding from the federal level to state and local levels (Cohen & Bloom, 2000). Specifically at the local level, this decentralization made counties responsible for provision and monitoring of mental health services (Scheffler et al., 2002) which included a focus on crisis services and community support rather than inpatient hospitalization (Cohen & Bloom, 2000). Findings are mixed if capitation reduces mental health utilization. In a study of Minnesota Medicaid recipients, there was no difference in mental health utilization for those in a capitated payment plan versus a fee-for-service plan (Lurie, Moscovice, Finch, Christianson, & Popkin, 1992). On the contrary, in a study of Colorado Medicaid recipients, those in a direct capitation plan had significantly less mental health visits over time as compared to those Medicaid recipients using a fee-for-service plan (Bloom et al., 2002). Additionally, as compared to fee-for-service, those Medicaid recipients using a capitation plan reported more services that were refused, discontinued, or reduced and a longer time between when the appointment was made to the appointment date (Bloom et al., 2002). Like Colorado, in California, mental health use decreased by 77% when the public mental health system shifted from fee-for-service to capitation (Scheffler et al., 2002). Traditionally, research on mental disorders has examined the impacts of sociodemographic factors such as race/ethnicity when assessing delivery of services. Although important, it is necessary to include other variables such as geographic location in order to improve understanding of variation in type and level of mental health services received. The purposes of this chapter are to describe the differences in utilization of mental health services, to examine need and unmet needs of mental health services, to explore intensity of mental health services, and to understand overuse of mental health services. Andersen’s Sociobehavioral Model The theoretical basis for this chapter is Andersen’s Sociobehavioral Model (SBM), which is shown in Fig. 1 (Andersen, 1995). There are three main components that make up the SBM: environment, population characteristics, and health behavior. The environment includes the health care system and the external environment. In this study, the environment includes 13 geographical states. Within population characteristics, there are predisposing characteristics, enabling characteristics, and need. Predisposing characteristics are the social and cultural factors that contribute to an individual’s probability of
206
STEPHANIE L. AYERS ET AL.
ENVIRONMENT
POPULATION CHARACTERISTICS
HEALTH BEHAVIOR
External Environment 13 states
Predisposing Characteristics
Enabling Resources
Gender Race Education Age
Income Insurance
Need
Poor Mental Health Symptoms Use of Mental Health Services
Fig. 1.
Andersen’s Sociobehavioral Model.
seeking mental health care. In this study, race, education, and age are included as predisposing characteristics. Enabling characteristics are the direct means needed to seek mental health treatment. Income and health insurance have been included in this study as enabling characteristics. Finally, need characteristics are categorized as those mental health symptoms that individuals define as problematic. In this study, poor mental health symptoms are used as an indicator of need. These factors (environmental and population characteristics) directly and indirectly affect the probability of individuals using mental health services (Andersen, 1995).
DATA AND METHODS Data for this study were obtained from the 2002 National Survey of American Families (NSAF) collected by the Urban Institute. The purpose of NSAF was to seek a better understanding of the well-being of children and adults in 13 states. We only used the child sample for our study because this is the only section of the NSAF study that asks mental health questions for adults. Adults aged 18 years and older who lived in the same household as the focal child were included in our study. The population included in the survey was non-institutionalized, civilian adults 18 years of age and older.
Geographic Disparities in Adult Mental Health Utilization
207
The sample was collected using random digit dialing of households with telephones. The telephone interview included questions regarding one randomly selected focal child aged 6–17 years. One adult within the sampled household was designated as the most knowledgeable adult (MKA) for the focal child. The total sample size was 23,327 with a 51.9% overall response rate (Urban Institute, 2005). Clustering effects exist within the 2002 NSAF data. Households were clustered into segments and segments were clustered into primary sampling units, thereby causing an increase in design effects due to the probability that households within each segment appeared to be similar. As a result, STATA 9.0 was used to control for design effects (StataCorp, 2005).
Dependent and Independent Variables Five dependent variables were used in this study. These included need for services, any use of services, unmet need of services, over-utilization of services, and intensity of services. Need for Mental Health Services This variable was preconstructed in the NSAF data. It was derived from the following five questions: How often in the past month have you: (a) been very nervous person, (b) felt calm or peaceful, (c) felt downhearted and blue, (d) been a happy person, and (e) felt so down in the dumps nothing could cheer you up? Responses were assigned to the following codes: 1 (all of the time), 2 (most of the time), 3 (some of the time), and 4 (none of the time). Responses to the questions about feeling calm or peaceful and being a happy person are reverse coded to ensure scale direction compatibility. Responses were totaled, creating a scale score ranging from 5 to 20 and then multiplied by five in order to create a mental health symptoms scale with a high score indicating better mental health symptoms. This variable was then dichotomized by NSAF. Respondents who had a score of 68–100 were assigned a code of 0 (good mental health symptoms) and respondents who had a score of 25–67 were assigned a code of 1 (having poor mental health symptoms). Respondents who had poor mental health symptoms were categorized as needing mental health services. Any Use of Mental Health Services This variable assessed the use of mental services in the last 12 months. The following question was asked: During the past 12 months, how many times
208
STEPHANIE L. AYERS ET AL.
have you received mental health services from a doctor, mental health counselor, or therapist? This variable was dichotomized into (0) no mental health visits and (1) had one or more mental health visits. Unmet Need of Mental Health Services This variable assessed respondents who had poor mental health but did not use any mental health services in the last 12 months. It was constructed using the two dependent variables, need for mental health services and any use of mental health services. This variable was dichotomized into (0) no unmet need for mental health services and (1) having an unmet need for mental health services. Respondents who were coded as 0 included those who exhibited poor mental health symptoms and had used of mental health services, those who exhibited good mental health symptoms and had no use of mental health services, and those who exhibited good mental health symptoms and had used mental health services. Respondents who had poor mental health symptoms and no mental health visits were assigned a code of 1. Overuse of Mental Health Services This variable described respondents who had good mental health and used mental health services in the last 12 months. It was constructed using the two dependent variables, need for mental health services and any use of mental health services. Respondents exhibiting poor mental health symptoms and those with good mental health symptoms and no use of mental health services were assigned a code of 0 (no overuse of mental health services). Respondents who exhibited good mental health symptoms and had mental health visits were assigned a code of 1 (overuse of mental health services). Intensity of Mental Health Services Intensity measured the number of mental health visits in the last 12 months by those who used at least one mental health service within this time frame. The following question was asked: During the past 12 months, how many times have you received mental health services from a doctor, mental health counselor, or therapist? This variable was a continuous measure of the number of mental health visits. A log transformation was performed on this variable due to skewness. Because this variable only included those who had used at least one mental health service, the sample size was 1,866. Six independent variables were used in this study. These included state, income, race, education, insurance, and age. The main independent variable
Geographic Disparities in Adult Mental Health Utilization
209
was the 13 geographic states (i.e., Alabama, California, Colorado, Florida, Massachusetts, Michigan, Minnesota, Mississippi, New Jersey, New York, Texas, Washington, and Wisconsin). State pertained to where the respondent lived at the time of the interview. Mississippi was chosen as the reference category based on the characteristics of that state in the logistic regression analysis. The coding for the remaining five independent variables is shown in Table 1. Subsequent analyses examined state-level characteristics. These characteristics were not included in the NSAF data and had to be derived from outside sources and include eight continuous variables: median state income (U.S. Census Bureau, 2000), dollars per person spent on mental health services (U.S. Department of Health and Human Services, 2001), percent of African Americans in the state (U.S. Census Bureau, 2000), percent of the state with high school degree (U.S. Census Bureau, 2000), percent of the state receiving Medicaid (U.S. Census Bureau, 2000), and mean length of stay in state psychiatric hospitals services (U.S. Department of Health and Human Services, 2002). Additionally, a categorical variable of state mental health laws was created. This variable was based on the 1996 Mental Health Parity Act (National Alliance on Mental Illness, 2007) which requires that employers with more than 50 employees who offer health insurance also offer mental health coverage as well; however the definition of mental health coverage at the discretion of each state. States with full comprehensive mental health coverage are defined as ‘‘comprehensive parity’’ and served as the reference group. Comprehensive parity is defined as those states whose mental health laws require an ‘‘equal coverage of a broad range of mental health conditions, including substance abuse disorders’’ (National Alliance on Mental Health, 2007, p. 1). States that have enacted laws for only a small set of serious mental problems are defined as ‘‘limited parity.’’ States with little or no mandated mental health coverage are defined as ‘‘no parity’’ (National Alliance on Mental Health, 2007). We calculated odds ratios for the dependent variables of need, any use, unmet need, and overuse of mental health services using a series of logistic regression models in order to identify independent associations between both mental health factors and individual sociodemographic factors, mental health factors and states, and mental health factors and state characteristics. Linear regression was used for the dependent variable of intensity of mental health services in order to identify the associations between intensity and individual sociodemographic factors, states, and state characteristics.
210
STEPHANIE L. AYERS ET AL.
Table 1. Descriptive Statistics for Dependent Variables: Need for Mental Health Services, Any Use of Mental Health Services, Unmet Need of Mental Health Services, Overuse of Mental Health Services, and Intensity of Services.
Geographic state Alabama California Colorado Florida Massachusetts Michigan Minnesota Mississippi New Jersey New York Texas Washington Wisconsin Income o100% poverty line 100–199% poverty line 200–299% poverty rate 300% and greater Race White Black Hispanic Other Education Less than high school High school diploma Some college Bachelors degree
Need
Any Service
Unmet Need
Overuse
Intensity
N=23,327
N=23,327
N=23,327
N=23,327
N=1,866
% (S.E.)
% (S.E.)
% (S.E.)
% (S.E.)
Mean (S.E.)
20.66 15.55 12.05 17.99 16.85 15.45 10.02 26.95 17.15 19.76 17.81 13.94 13.80
(0.85) (0.68) (0.78) (0.77) (0.80) (0.70) (0.81) (1.53) (0.62) (0.67) (0.76) (0.70) (0.75)
18.47(1.46) 14.02 (1.15) 10.02 (0.89) 15.81 (1.55) 13.52 (0.98) 13.33 (0.80) 8.10 (0.78) 22.98 (3.38) 15.20 (0.88) 17.08 (1.29) 15.51 (1.03) 12.42 (1.01) 11.19 (0.70)
3.44 5.34 7.12 4.97 7.28 5.41 6.35 2.47 4.58 4.59 4.98 5.71 6.81
34.07 (1.35)
8.84 (0.85)
29.43 (1.34)
4.20 (0.54)
8.52 (0.53)
21.60 (0.84)
8.80 (0.85)
18.12 (0.85)
5.32 (0.50)
8.62 (0.62)
16.98 (0.92)
7.30 (0.62)
14.76 (0.90)
5.08 (0.67)
8.52 (0.88)
9.22 (0.43)
7.25 (0.39)
7.76 (0.41)
5.78 (0.34)
8.18 (0.48)
6.15 4.71 2.81 4.05
7.49 8.74 6.76 7.87
15.43 19.44 21.69 14.79
(1.45) (1.17) (0.90) (1.61) (1.07) (0.88) (0.93) (3.44) (0.90) (1.41) (1.09) (1.04) (0.77)
(0.49) (1.09) (1.17) (1.95)
5.62 6.84 9.18 7.23 10.61 7.56 8.27 6.43 6.51 7.25 7.26 7.21 9.43
8.90 7.11 4.98 4.57
(0.40) (0.76) (0.61) (0.89)
12.69 17.05 19.51 14.26
(0.45) (1.09) (1.07) (1.89)
(0.62) (0.59) (0.68) (0.70) (0.72) (0.64) (0.75) (0.59) (0.55) (0.54) (0.65) (0.65) (0.67)
(0.31) (0.71) (0.38) (0.88)
6.33 7.95 9.84 7.67 10.84 9.08 7.90 4.98 8.95 10.60 7.22 8.30 7.28
(0.86) (0.91) (0.86) (0.82) (0.72) (0.81) (0.68) (0.70) (0.79) (0.98) (0.85) (0.96) (0.63)
(1.65) (0.39) (0.67) (0.65)
29.27 (1.62)
6.43 (0.69)
26.49 (1.57)
3.65 (0.54)
7.54 (0.85)
20.56 (0.78)
7.41 (0.59)
17.71 (0.69)
3.65 (0.44)
7.88 (0.55)
16.10 (0.67) 8.81 (0.58)
9.50 (0.59) 6.96 (0.47)
13.09 (0.59) 7.37 (0.57)
6.49 (0.50) 5.53 (0.45)
9.12 (0.71) 8.13 (0.89)
Geographic Disparities in Adult Mental Health Utilization
211
Table 1. (Continued ) Need
Any Service
Unmet Need
Overuse
Intensity
N=23,327
N=23,327
N=23,327
N=23,327
N=1,866
% (S.E.)
% (S.E.)
% (S.E.)
% (S.E.)
Mean (S.E.)
Type of insurance Medicaid Medicare Uninsured Private
28.18 15.25 23.98 12.20
Gender Male Female
10.91 (0.89) 18.12 (0.47)
Age 18–34 35–44 45–54 55–64 65 and older
18.92 16.55 14.79 17.01 12.40
(1.05) (2.14) (1.73) (0.43)
(0.90) (0.63) (0.79) (2.10) (3.31)
11.43 10.32 4.31 7.00
(0.67) (1.68) (0.61) (0.35)
5.12 (0.54) 7.91 (0.31) 6.66 7.44 10.01 12.67 10.83
(0.48) (0.36) (0.70) (2.17) (2.66)
23.09 11.60 22.71 10.47
(0.98) (1.66) (1.73) (0.40)
6.34 6.67 3.03 5.24
(0.47) (1.32) (0.62) (0.29)
10.04 (0.88) 15.69 (0.45)
4.25 (0.51) 5.47 (0.26)
16.43 13.98 12.66 13.18 10.24
4.17 4.87 7.87 8.84 8.67
(0.87) (0.53) (0.76) (1.90) (3.17)
(0.35) (0.30) (0.67) (1.75) (2.47)
9.84 10.57 6.81 7.45
(0.55) (1.18) (0.91) (0.38)
8.20 (0.86) 8.40 (0.33) 6.73 8.17 8.20 9.83 13.85
(0.75) (0.49) (0.45) (0.83) (2.15)
FINDINGS Descriptive statistics are presented in Fig. 2 and Table 1. Sample averages for all figures were derived from the percentage of those individuals having a need, use, and unmet need for the total sample. Geographic variations in need for mental health services are shown in Fig. 2a. Mississippi had the highest percentage of need (26.95%), while Minnesota had the lowest (10.02%). Fig. 2b illustrates the geographic variations in any use of mental health services. Massachusetts had the highest use of services (10.61%) while Alabama had the lowest (5.62%). Geographic variation in unmet need for mental health services is shown in Fig. 2c. Minnesota had the lowest unmet need while Mississippi had the highest, 8.10% and 22.98%, respectively. Fig. 2d illustrates the geographic variations in overuse of services. Massachusetts had the highest overuse (7.28%), and Mississippi had the lowest (2.47%). Fig. 2e shows intensity of services. The sample mean of 8.29 visits was constructed by using the mean of the sub sample population. Massachusetts had the highest intensity (10.84 visits), while Mississippi had the lowest (4.98 visits).
212
STEPHANIE L. AYERS ET AL.
(a)
(b) 12 Sample Average:16.98%
ts ga n ne M sot is si a N ss ew i p Je p i N rse ew y Yo T rk W exa as s hi n W gto is co n ns in
et
hi
ic
in
M
us
ch
sa as M
M
o
a id
ad
or
or
Fl
a
rn ifo
am ab Al
in
ol
ts ga n M sot is si a N ss ew ip Je pi N rse ew y Yo T rk W exa as s hi n W gto is co n ns in ne
et
hi
us
ic
M
ch
sa as M
M
a id
ad
or
or
Fl
a
rn
am
ifo
C
C
al
ab Al
o
0 ia
2
0
ia
4
5
State
(c)
State
(d) 8
25 Sample Average:14.48%
20
Percent adults
Percent adults
6
al
10
8
C
15
C
20
Sample Average:7.82%
10
Percent adults
25
ol
Percent adults
30
15 10 5
Sample Average:5.31%
6 4 2
hi ga n n M eso is si ta N ssip ew p Je i rs ey N ew Yo rk Te W as xas hi ng to W n is co ns in
ic M
in
et us
M
M
ts
o
a id or
Fl as
sa
ch
rn
or
ifo
ol
C
al
ad
a am ab Al
C
et ts g in an ne M so ta is s N iss e w ip J e pi r N sey ew Yo rk W Tex as as hi n W gto is co n ns in hi
us
ic
M
ch
sa M
as
M
a
o
id
ad
or
or ol
Fl
ia
a
rn
am
ifo
al
C
C
ab Al
ia
0 0
State
State
(e) Mean number of visits
Sample Average: 8.29 12 10 8 6 4 2
ol
al
C
Al
C
ab
am
ifo a rn ia or ad M as Flo o sa ri ch da us M etts ic M hig in an ne s M is ota N siss e w ip J e pi N rse ew y Yo r W Te k as xa hi s n g N at Wis ton io na con l A sin ve ra ge
0
State
Fig. 2. Descriptive Statistics for Dependent Variables: (a) Need for Mental Health Services, (b) Any Use of Mental Health Services, (c) Unmet Need of Mental Health Services, (d) Overuse of Mental Health Services, and (e) Intensity of Mental Health Services by 13 Geographic States.
Table 1 presents the descriptive statistics for each state and sociodemographic variables. For income, those respondents who have income less than 100% of the federal poverty level had the highest percentage of need for mental health services, 34.07%, the highest percent of any use of mental health services, 8.84%, and the highest percent of unmet need of use of service, 29.43%. In contrast, those respondents with the income 300% or
Geographic Disparities in Adult Mental Health Utilization
213
higher than the federal poverty level had the lowest percentage of need, any use, and unmet need, 9.22%, 7.25%, and 7.76%, respectively. Respondents whose race/ethnicity is non-Hispanic White have the highest percent of any use of mental health services, 8.90%, while respondents whose race/ethnicity is Other have the lowest use of mental health services, 4.57%. Hispanics have the highest need for mental health services, 21.69%, and the highest unmet need, 19.51%. African-Americans have the highest mean of intensity of services, 8.74. For education, those respondents who have some college had the highest use of mental health services, 9.50%, and the highest mean intensity of services, 9.12. For need and unmet need, those respondents with less than a high school diploma had the highest percentages, 29.27% and 26.49%, respectively. For age, those respondents ages 55–64 have the highest use of mental health services, 12.67%, while those ages 18–34 have the highest need and unmet need of mental health services, 18.92% and 16.43%, respectively. Those respondents’ ages 65 and older have the highest mean of intensity of services, 13.85. For type of health insurance, those respondents having Medicaid had the highest percent of any use of mental health services, 11.43%, need for mental health services, 28.18%, and the highest percent of unmet need of mental health services, 23.09%. Those respondents having Medicare have the highest mean intensity of services, 10.57. The uninsured have the lowest use of mental health services, 4.31%, and the lowest mean for intensity of mental health visits, 6.81. Those respondents with private health insurance have the lowest need for and unmet need of mental health services, 12.20% and 10.47%, respectively.
Need for Mental Health Services Table 2, Model 1, shows an inverse association between income and education and need for mental health services. As income and education increased the odds of having a need for mental health services decreased. Women have significantly greater odds of needing mental health services. Hispanic respondents were less likely than non-Hispanic whites to need mental health services. When compared to those respondents with private insurance, respondents with Medicaid, as well as, the uninsured had significantly greater odds of needing mental health services. Age was not a significant predictor for needing mental health services. California, Colorado, Michigan, Minnesota, Texas, Washington, and Wisconsin had significantly lower odds of needing mental health services
214
STEPHANIE L. AYERS ET AL.
Table 2. Individual Characteristics. Logistic Regression for Variables: Need, Any Use, Unmet Need, and Overuse (N=23,327). Linear Regression for Variable: Intensity (N=1,866). Need
Any Use
Unmet Need
Overuse
Intensity
Model 1
Model 2
Model 3
Model 4
Model 5
OR
OR
OR
OR
B
Gendera Income Education
1.60 0.81 0.80
1.62 0.94 1.13
1.48 0.82 0.80
1.47 0.98 1.20
0.05 0.05 0.07
Raceb Black Hispanic Other
0.86 0.74 1.22
0.56 0.44 0.58
0.92 0.84 1.44
0.53 0.45 0.71
0.19 0.12 0.18
Insurancec Medicaid Medicare Uninsured
1.57 1.01 1.65
1.57 1.21 0.80
1.47 1.00 1.76
1.49 1.26 0.96
0.55 0.28 0.18
1.01
1.03 2.17
1.01
1.03
0.02
Age Poor mental healthd Intercept R
2
0.74 0.06
0.04
0.06
0.03
0.06
a
Males=reference category. Non-Hispanic whites=reference category. c Private insurance=reference category. d Good mental health symptoms=reference category. po0.05. po0.01. po0.001. b
when compared to Mississippi (Table 3, Model 1). Even when controlling for individual characteristics, there were still state-level variations in the need for mental health services. Table 4, Model 1 examines the state-level characteristics and need for mental health services. Those states with limited parity mental health laws, as compared to comprehensive mental health laws, have significantly greater odds of needing mental health services. As the percentage of African Americans in the state increased, the need for mental health services increased by 1% (odds=1.01). A one-day increase in the mean length of
Geographic Disparities in Adult Mental Health Utilization
215
Table 3. Geographic States. Logistic Regression for Variables: Need, Any Use, Unmet Need, and Overuse (N=23,327). Linear Regression for Variable: Intensity (N=1,866). Need
Any Use
Unmet Need
Overuse
Intensity
Model 1
Model 2
Model 3
Model 4
Model 5
OR
OR
OR
OR
B
a
State Alabama California Colorado Florida Massachusetts Michigan Minnesota New Jersey New York Texas Washington Wisconsin
0.83 0.61 0.56 0.74 0.86 0.70 0.48 0.87 0.84 0.66 0.62 0.63
0.93 1.50 1.81 1.32 1.80 1.33 1.48 1.20 1.23 1.57 1.25 1.71
0.88 0.63 0.55 0.77 0.81 0.72 0.47 0.90 0.86 0.66 0.67 0.60
1.38 2.55 2.85 2.07 2.62 2.11 2.30 1.83 1.78 2.44 2.10 2.63
0.34 0.47 0.83 0.60 0.92 0.77 0.63 0.80 0.80 0.44 0.49 0.53
Genderb Income Education
1.59 0.81 0.80
1.62 0.94 1.13
1.47 0.82 0.79
1.48 0.97 1.19
0.02 0.05 0.07
Racec Black Hispanic Other
0.81 0.79 1.27
0.59 0.42 0.57
0.87 0.88 1.48
0.56 0.42 0.68
0.19 0.12 0.11
Insuranced Medicaid Medicare Uninsured
1.03 1.56 1.67
1.20 1.58 0.79
1.02 1.46 1.78
1.23 1.50 0.95
0.34 0.57 0.13
1.01
1.03 2.20
1.01
1.03
0.01
Age Poor mental healthe Intercept R a
2
0.24 0.07
0.03
Mississippi=reference category. Males=reference category. c Non-Hispanic whites=reference category. d Private Insurance=reference category. e Good mental health symptoms=reference category. po0.05. po0.01. po0.001. b
0.06
0.03
0.10
216
STEPHANIE L. AYERS ET AL.
Table 4. State Characteristics. Logistic Regression for Variables: Need, Any Use, Unmet Need, and Overuse (N=23,327). Linear Regression for Variable: Intensity (N=1,866).
State income State MH budget
Need
Any Use
Unmet Need
Overuse
Intensity
Model 1
Model 2
Model 3
Model 4
Model 5
OR
OR
OR
OR
B
0.99 1.00
State mental health lawsa No parity 0.85 Limited parity 1.26
0.98 1.00
0.99 1.00
0.99 1.00
0.005 0.004
1.10 1.43
0.87 1.15
1.23 1.34
0.067 0.238
State black State high school State medicaid Median days stay
1.01 1.04 1.00 1.01
0.97 1.03 1.00 1.00
1.02 1.02 1.00 1.01
0.96 1.01 1.00 1.00
0.012 0.008 0.031 0.001
Genderb Income Education Racec Black Hispanic Other
1.59 0.81 0.80
1.62 0.94 1.13
1.47 0.82 0.80
1.48 0.97 1.19
0.018 0.046 0.066
0.81 0.79 1.28
0.59 0.42 0.57
0.87 0.89 1.50
0.56 0.42 0.68
0.192 0.121 0.113
Health insuranced Medicaid Medicare Uninsured
1.57 1.04 1.68
1.58 1.20 0.79
1.47 1.02 1.79
1.50 1.23 0.95
0.565 0.335 0.128
1.01
1.03 2.19
1.01
1.03
0.014
Age Neede Intercept R a
2
0.807 0.07
0.03
Comprehensive parity law=reference category. Males=reference category. c Non-Hispanic whites=reference category. d Private insurance=reference category. e Good mental health symptoms=reference category. po0.05. po0.01. po0.001. b
0.06
0.03
0.09
Geographic Disparities in Adult Mental Health Utilization
217
stay in state psychiatric hospitals increased, the odds of needing mental health services increased by 1% (odds=1.01). The remaining state-level characteristics (e.g., median state income, mental health budget, percent of the state with a high school degree, percent of the state receiving Medicaid) were not significant with respect to the need for mental health services. Any Use of Mental Health Services Although income is not a significant predictor of using mental health service, increased education resulted in increased odds for using mental health services (Table 2, Model 2). Women have significantly greater odds of needing mental health services. When compared to non-Hispanic whites, racial and ethnic minority groups have significantly lower odds of using mental health services. Having Medicaid as opposed to private health insurance was associated with greater odds of using mental health services. As age increased, the odds of using a mental health service also increased. Having a need for mental health services significantly increases the odds of using mental health services. In Table 3, Model 2, only Colorado had significantly greater odds of using mental health services when compared to Mississippi (Model 2, Table 3). Table 4, Model 2 examines the state-level characteristics and any use of mental health services. States with limited parity laws utilize mental health services 57% more than states with full parity laws (odds=1.43). As the percentage African American increased, use of mental health services decreased by 3% (odds=0.97). The remaining state-level characteristics (e.g., median state income, mental health budget, percent of the state with a high school degree, percent of the state receiving Medicaid, and mean length of stay in state psychiatric hospitals) were not significant with respect to use of mental health services. Unmet Need of Mental Health Services Individual characteristics for having an unmet need of mental health services are shown in Table 2, Model 3. Income and education exhibited an inverse association with unmet need of mental health services. Women have significantly greater odds of needing mental health services. Race/ethnicity and age were not significant predictors of unmet need for mental health services. Respondents with Medicaid, as well as the uninsured, had significantly greater odds of having unmet needs.
218
STEPHANIE L. AYERS ET AL.
California, Colorado, Minnesota, Texas, Washington, and Wisconsin had significantly lower odds of having an unmet need for mental health services when compared to Mississippi (Table 3, Model 3). These states all had significantly lower need for mental health services and significantly lower odds of having an unmet need for those services. Table 4, Model 3 examines the state-level characteristics and unmet need of mental health services. Examination of state-level characteristics showed that as the state percentage of African American increased, the need for mental health services decreased by 2% (odds=1.02). A one-day increase in the mean length of stay in state psychiatric hospitals increased, the odds of needing mental health services increased by 1% (odds=1.01). The remaining state-level characteristics (e.g., median state income, mental health budget, type of mental health law, percent of the state with a high school degree, and percent of the state receiving Medicaid) were not significant with respect to unmet need of mental health services.
Overuse of Mental Health Services Unlike income, education was a significant predictor of overuse of mental health services (Table 2, Model 4). As education increased, the odds of having an overuse of mental health services increased. Women have significantly greater odds of overusing mental health services. African Americans and Hispanics had significantly lowered odds of overusing mental health services, 47% and 55%, respectively, when compared to nonHispanic whites. Receipt of Medicaid was significantly associated with greater odds of overuse of mental health services. Likewise, as age increased, the odds of overuse of mental health services increased. Residents of California, Colorado, Florida, Massachusetts, Michigan, Minnesota, New Jersey, New York, Texas, and Washington, Wisconsin had higher odds of overuse of mental health services when compared to Mississippi (Table 3, Model 4). The state with significantly higher odds in any use of mental health services (i.e., Colorado) also exhibited significantly higher overuse of mental health services. Table 4, Model 4 examines the state-level characteristics and overuse of mental health services. Examination of state-level characteristics showed that as the percentage of African Americans in the stated increased, the odds of overuse of mental health services decreased by 4% (odds=0.96). The remaining state-level characteristics (e.g., median state income, mental health budget, type of mental health law, percent of the state with a high
Geographic Disparities in Adult Mental Health Utilization
219
school degree, percent of the state receiving Medicaid, and mean length of stay in state psychiatric hospitals) were not significant with respect to overuse of mental health services.
Intensity of Mental Health Services In Table 2, Model 5, shows that only health insurance and age are significant predictors of intensity of mental health visits. Having Medicaid was associated with an increase in the number of mental health visits. Residents of California, Colorado, Florida, Massachusetts, Michigan, Minnesota, New Jersey, New York, Washington, and Wisconsin had significantly more mental health visits than residents of Mississippi (Table 3, Model 5). Table 4, Model 5 examines the state-level characteristics and intensity of mental health services. Examination of state-level characteristics showed that as the state mental health budget (i.e., dollars per person spent on mental health services) increased, the number of visits increased by 0.004 visits. As the percentage of Medicaid recipients in the state increases, the intensity of services decreases by 0.031 visits. Additionally, as the mean length of stay in state psychiatric hospitals increased, the number of visits to mental health professionals increases. The remaining state-level characteristics (e.g., median state income, type of mental health law, percent of the state African American, and percent of the state with a high school degree) were not significant with respect to intensity of mental health services.
DISCUSSION This chapter examined a variety of factors involving mental health utilization: any use of services, need for services, unmet need of services, overuse of, and intensity of visits. By using Andersen’s SBM (Andersen, 1995) and including various measures of mental health utilization, this chapter provided a comprehensive examination of both geographic variation and sociodemographic factors on the mental health system. This study strengthens previous findings regarding individual variations in sociodemographic characteristics as predictors of mental health utilization. Similar to previous research, results of this study show that a need for mental health services is the strongest predictor for using mental health services (Scheffler et al., 2002; Rosen et al., 2004; Bao & Sturm, 2004;
220
STEPHANIE L. AYERS ET AL.
Albizu-Garcia et al., 2001). Likewise, Hispanics and African Americans have a significantly lowered use of mental health services when compared to whites (Scheffler et al., 2002; Rosen et al., 2004; Bao & Sturm, 2004; Albizu-Garcia et al., 2001; Albizu-Garcia et al., 2001; Hines-Martin et al., 2003; Vega et al., 1999). Contrary to previous studies, having Medicaid is associated with increased use of mental health services (Goldman et al., 1998; Albizu-Garcia et al., 2001). The theoretical basis for this chapter was Andersen’s SBM, which was supported by this research. The Andersen’s SBM theorized that in order to properly predict utilization of mental health services both population characteristics and the external environment must be taken into account (Andersen, 1995). This study examined and found numerous disparities among 13 states (i.e., the external environment). For example, Colorado had lower need and unmet need for mental health services, but had higher use, overuse, and intensity of services than Mississippi. These findings suggest that although individual sociodemographic characteristics are important in examining mental health utilization, there are other broader factors that impact utilization. State-level characteristics were also examined to link specific state-level characteristics with utilization of and need for services. These results show a strong and significant association between the percentage of African Americans in the state and mental health need and utilization. An increased percentage of African Americans in the state results in increased need and unmet need and a decreased use and overuse of mental health services. These relationships remained significant after controlling for all individuallevel characteristics. These results also show a positive significant relationship between mean length of stay in psychiatric hospitals and need, unmet need, and intensity of services. These relationships remained significant after controlling for all individual-level characteristics. Although there were significant associations between states and use and need of mental health services, the mechanisms at work within the state were not clear. Additionally, those states with limited parity laws have a significantly greater need for services and a greater use of services. Individual characteristics have more of an impact on utilization and need for mental health services, however, the totality of state characteristics impact utilization and need for mental health services as well. Additional intrastate and interstate studies are needed in order to understand what state-level characteristics, other than percentage of African Americans, mean length of stay in psychiatric hospitals, and state mental health laws, help to explain variation in mental health service utilization and need
Geographic Disparities in Adult Mental Health Utilization
221
patterns. However, given the current interest in health care disparities, this study demonstrates both geographic disparities across states and the importance of percentage of African Americans in the state as one factor in creating this disparity. The sample used for this study was a major limitation. Because only adults with children under the age of 18 were used, these findings may not be generalizable to adults without children. Future research should examine differences between adults with children and adults without children with respect to need for mental health services, use of service, unmet need for mental health services and intensity of those services. Additionally, future research should better assess state-level characteristics in order to promote a clearer understanding of the variations in mental health utilization. In conclusion, this study provides information for conducting research as well as the delivery of service. First, the findings of this study supplement the body of literature regarding the relationships between sociodemographic characteristics and need for and utilization of mental health services. Furthermore, this study adds to the body of health literature by demonstrating a clear relationship between the presence of Medicaid and greater utilization and need for mental health services. Third, the findings establish broader factors that impact need and utilization of mental health services. State-level characteristics, specifically percentage of African Americans, state mental health laws, and mean length of stay in psychiatric hospitals are also important for understanding the need for and utilization of mental health services.
REFERENCES Albizu-Garcia, C. E., Alegria, M., Freeman, D., & Vera, M. (2001). Gender and health services use for a mental health problem. Social Sciences and Medicine, 53, 865–878. Alegria, M., Canino, G., Rios, R., Vera, M., Calderon, J., Rusch, D., & Ortega, A. N. (2002). Inequalities in use of specialty mental health services among Latinos, African Americans, and non-Latino whites. Psychiatric Services, 53, 1547–1555. Andersen, R. M. (1995). Revisiting the behavioral model and access to medical care: Does it matter? The Journal of Behavioral Health Services and Research, 36, 1–10. Arias, E., Anderson, R. N., Hsiang-Ching, K., Murphy, S. L., & Kochanek, K. D. (2003). Deaths: Final data for 2001. National vital statistics reports; Vol 52, no 3. Hyattsville, MD: National Center for Health Statistics. Bao, Y., & Sturm, R. (2004). The effects of state mental health parity legislation on perceived quality of insurance coverage, perceived access to care, and use of mental health specialty care. Health Services Research, 39, 1361–1377. Bloom, J. R., Hu, T.-W., Wallace, N., Cuffel, B., Hasuman, J. W., Sheu, M.-L., & Scheffler, R. (2002). Mental health costs and access under alternative capitation systems in colorado. Health Services Research, 37, 315–340.
222
STEPHANIE L. AYERS ET AL.
Chumbler, N. R., Cody, M., Booth, B. M., & Beck, C. K. (2001). Rural-urban differences in service use for memory-related problems in older adults. The Journal of Behavioral Health Services and Research, 28, 212–221. Cohen, E., & Bloom, J. R. (2000). Managed care and service capacity development in a public mental health system. Administration and Policy in Mental Health, 28(2), 63–74. DeFrances, C. J., & Hall, M. J. (2004). 2002 National Hospital Discharge Survey. Advance data from vital and health statistics; no 342. Hyattsville, MD: National Center for Health Statistics. Executive Order 13263. (2002). President’s New Freedom Commission on Mental Health. Federal Register, 67, 22337–22338. Goldman, W., McCulloch, J., & Sturm, R. (1998). Costs and use of mental health services before and after managed care. Health Affairs, 17, 40–52. Hines-Martin, V., Brown-Piper, A., Sanggil, K., & Malone, M. (2003). Enabling factors of mental health service use among African Americans. Archives of Psychiatric Nursing, 5, 197–204. Lurie, N., Moscovice, I. S., Finch, M., Christianson, J. E., & Popkin, M. K. (1992). Does capitation affect the health of the chronically mentally ill? Results from a randomized clinical trial. Journal of the American Medical Association, 267, 3300–3304. National Alliance on Mental Illness. (2007). State Mental Health Parity Laws 2007. Retrieved on May 29, from http://www.nami.org National Institute of Mental Health. (2001). The Impact of Mental Illness on Society. Retrieved on December 2005, from http://www.nimh.nih.gov/publicat/burden.cfm Philo, C. (2005). The geography of mental health: An established field? Current Opinion in Psychiatry, 18, 585–591. Regier, D. A., Narrow, W. E., Rae, D. S., Manderscheid, R. W., Locke, B. Z., & Goodwin, F. K. (1993). The de facto mental and addictive disorders service system. Epidemiologic Catchment Area prospective 1-year prevalence rates of disorders and services. Archives of General Psychiatry, 50(2), 85–94. Rosen, D., Tolman, R. M., & Warner, L. A. (2004). Low-income women’s use of substance abuse and mental health services. Journal of Health Care for the Poor and Underserved, 15, 206–219. Sarkinsian, C. A., Lee-Henderson, M. H., & Mangione, C. M. (2003). Do depressed older adults who attribute depression to ‘‘old age’’ believe it is important to seek care? Journal of General Internal Medicine, 18, 1001–1005. Scheffler, R. M., & Miller, A. B. (1991). Differences in mental health services utilization among ethnic subpopulations. International Journal of Law and Psychiatry, 14, 363–376. Scheffler, R., Zhang, A., & Snowden, L. (2002). The impact of realignment on utilization and cost of community-based mental health services in California. Administration and Policy in Mental Health, 29, 129–143. Sommers, I. (1989). Geographic location and mental health services utilization among the chronically mentally ill. Community Mental Health Journal, 25(2), 132–144. StataCorp. (2005). Stata Statistical Software: Release 9. College Station, TX: StataCorp Lp. Sullivan, G., Jackson, C. A., & Spritzer, K. L. (1996). Characteristics and service use of seriously mentally ill persons living in rural areas. Psychiatric Services, 47(1), 57–61. Urban Institute. (2005). The 1997, 1999 and 2002 National Survey of America’s Families methodology reports. The Urban Institute web site: http://www.urban.org/content/ Research/NewFederalism/NSAF/Methodology/2002MethodologySeries/2002.htm
Geographic Disparities in Adult Mental Health Utilization
223
U.S. Census Bureau. (2000). U.S. Census Bureau: State and County Quick Facts. Data derived from Population Estimates, Census of Population and Housing, Small Area Income and Poverty Estimates, State and County Housing Unit Estimates, County Business Patterns, Nonemployer Statistics, Economic Census, Survey of Business Owners, Building Permits, Consolidated Federal Funds Report. U.S. Department of Health and Human Services. (1999). Mental health: A report of the surgeon general. Rockville, MD: Department of Health and Human Services (US), National Institutes of Health, National Institute of Mental Health. U.S. Department of Health and Human Services. (2001). State Mental Health Agency, Mental Health Actual Dollar & Per Capita Expenditures, 2001. Rockville, MD: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, Center for Mental Health Services. Retrieved on December 2006, from http://mentalhealth.samhsa.gov/databases/ U.S. Department of Health and Human Services. (2002). 2002 CMHS Uniform Reporting System Output Tables. Rockville, MD: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, Center for Mental Health Services. Retrieved on December 2006, from http://mentalhealth.samhsa.gov/cmhs/ MentalHealthStatistics/ Vega, W. A., Kolody, B., Aguilar-Gaxiola, S., & Catalano, R. (1999). Gaps in service utilization by Mexican-Americans with mental health problems. American Journal of Psychiatry, 156, 958–962. Williams, D. R., & Collins, C. (1995). U.S. socioeconomic and racial differences in health. Annual Review of Sociology, 21, 349–386. Woodwell, D. A., & Cherry, D. K. (2004). National Ambulatory Medical Care Survey: 2002 Summary. Advance data from vital and health statistics; no 346. Hyattsville, MD: National Center for Health Statistics. Worthington, C. (1992). An examination of factors influencing the diagnosis and treatment of black patients in the mental health system. Archives of Psychiatry Nursing, 6, 195–204.
TRANSGENDER BODIES, IDENTITIES, AND HEALTHCARE: EFFECTS OF PERCEIVED AND ACTUAL VIOLENCE AND ABUSE Tarynn M. Witten ABSTRACT ‘‘Disparity’’ implies the existence of a ‘‘markedly distinct in quality or character,’’ difference between one group and another. Some groups, due to elevated stigma associated with group membership, are invisible as a disparate minority and therefore, while there may be a great inequity in healthcare between that group and the normative population, the invisible minority is ignored. This chapter addresses the issue of healthcare for the transgender-identified population. We address how the normative viewpoint of mental illness and unacceptable religious status, along with lifelong perceived and actual abuse and violence, creates a socially sanctioned inequality in healthcare for this population.
Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 225–249 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00010-5
225
226
TARYNN M. WITTEN
INTRODUCTION Invisibility, Minority, and Disparity The issue of healthcare disparity has become a central research focus in recent years (Victoria, 2006; Irwin et al., 2006; Alliance for Health Reform, 2006). Reports from various federal agencies, as well as private think tanks (Smedley, Stith, & Nelson, 2003) carefully document disparities in healthcare delivery, health status, and treatment outcome for various ‘‘US racial and ethnic minority groups’’ (Atdjian & Vega, 2005; Sheikh, 2006). Atdjian and Vega (2005) further point out that all such research articles and reports ‘‘urge immediate action to overcome these disparities.’’ Central to this discussion are two words: ‘‘minority’’ and ‘‘disparity.’’ The term ‘‘minority’’ recognizes the existence of a socially sanctioned ‘‘outgroup’’ or ‘‘group/sub-population differing from others in some characteristics and often subjected to differential treatment’’ (Merriam-Webster Online, 2006), yet now deemed important enough to study. The term ‘‘disparity’’ implies the existence of a ‘‘markedly distinct in quality or character’’ (Merriam-Webster Online, 2006), difference between one group and another. Both definitions (1) imply that a unique sub-group of individuals exists and that (2) there is/are inequity/inequities between that group and the ‘‘normative’’ or ‘‘in-group’’ population. However, some ‘‘out-groups,’’ due to the elevated stigma associated with membership in that group, are invisible with respect to being defined as a disparate minority (Witten & Eyler, 1999) and therefore, while there may be a great disparity in healthcare between that group and the normative population (U.S. Department of Health and Human Services, 2000), the invisible minority is not sanctioned as ‘‘studyable’’ (GLMA, 2000). This may be due to the fact that the in-group finds the existence of a particular minority identity to be intolerable for religious, moral, or other psychosocio-economic-political reasons, thereby causing it to be invisible with respect to healthcare research and delivery (Witten, 2004).
OVERVIEW OF GENDER MINORITY IDENTITIES Traditional Western Biomedical Perspective of Sex/Gender The traditional Western biomedical construction of identity routinely conflates sexuality, gender, and birth body or ‘‘birth sex/reproductive sex’’
Transgender Bodies, Identities, and Healthcare
227
(Basu, 2000; Doyal, 2001; Greenberg, 1998; Grant, 2001; Pesquera, 1999; Pryzgoda & Chrisler, 2000; Witten, 2004, 2005; Witten & Eyler, 1999). A trivial example is the conflation of sex and gender on numerous medical forms worldwide that routinely ask for ‘‘gender’’ when they obviously mean ‘‘birth sex/birth body,’’ as suggested in Witten (2005). This is best illustrated when examining the National Institutes of Health PHS398 personal investigator information form. This form is required of all researchers who submit a grant proposal to the NIH. As part of the information collected, the NIH asks for the investigator’s ‘‘sex/gender.’’ Furthermore, even if a survey asks for birth sex, it provides only the two choices of male and female, thereby ignoring the existence of the worldwide intersex population. This lack of attention to the intersex population is particularly important as intersex birth incidence is estimated between 1/1200-1/200 (ISNA, 2007).
Multi-Cultural Minority Gender Identities Gender minority persons include transsexuals, transgenders, cross-dressers, and others with gender self-perceptions other than the traditional Western dichotomous gender world-view (i.e. identifying only masculine and feminine) such as members of some Native American and other indigenous groups (Langevin, 1983; Satterfeld, 1988; Godlewski, 1988; Hoenig & Kenna, 1974; Kro¨hn, Bertermann, Wand, & Wille, 1981; Kockott & Fahrner, 1988; Sigusch, 1991; Tsoi, 1988; van Kesteren, Gooren, & Megens, 1996; Walinder, 1971, 1972; Weitze & Osburg, 1996; Witten et al., 2003). Breadth of cultural competence is important in transgender (HBIGDA, 2007) and intersex (ISNA, 2007) healthcare, as many indigenous peoples also recognize genders other than male and female. For example, adult members of the Tewa tribal culture identify as women, men, or members of the third gender, ‘‘kwido,’’ although their American birth records and other government documents recognize only females and males (Jacobs & Cromwell, 1992). The Chukchi, in early 20th century North America, recognized seven genders in addition to female and male. The traditional cultures of Tonga and Samoa identify Fa’afafine and Fa’afatama as additional genders (Witten et al., 2003; Schmidt, 2003). Recently the Hijra (India, Pakistan, and Bangladesh) have been acknowledged as a third gender by the Indian government, thereby paving the way for their passports to indicate this status. Japanese culture contains a number of ‘‘folk categories’’ that are considered to be transgendered, for example ‘‘okama, gei bli, bur4bli and ny4h#fu’’ (McLelland, 2004). The Mak Nyah
228
TARYNN M. WITTEN
are Malaysian male transsexuals (Teh, 2001; Zucker & Blanchard, 2003; Poasa & Blanchard, 2004). Identity and Thai transgender is discussed in the work of Winter (2005, 2006), while Turkish transsexual status is covered in Atamer (2005). Similar non-traditional gender identities exist in indigenous New Zealand peoples (Maori). A more detailed discussion of worldwide gendered identities may be found in Witten and Eyler (2007b). Roughgarden (2004) provides a particularly comprehensive discussion of transgender organisms across the animal kingdom.
Aging and Transgender Identities Estimates of US and worldwide transgender population sizes are discussed in Witten (2003). Based upon these estimates, Witten finds that the worldwide population of non-normative (non-Western) gender identities may exceed 20 million individuals and therefore constitutes a non-trivial minority population that remains excluded from international healthcare research efforts. Physicians and other healthcare/caregiving professionals should remain aware of the possibility of culturally normative gender variance when discussing gender identity with their patients, particularly as elderly persons are more likely to have retained traditional cultural beliefs and practices than are their younger peers. Multicultural aspects of gender identity are especially important in countries where the rates of aging are high, such as Malaysia, Bangladesh, India, Pakistan, and Thailand. We will momentarily see why the intersection of aging and trans-identities/bodies is a topic of importance.
VIOLENCE, ABUSE, HATE SPEECH, AND HATE CRIMES AGAINST TRANSGENDERED IDENTITIES Transgender Hate Violence and Abuse Hate crimes against a lesbian, gay (Herek, 1989; Herek et al., 2002), bisexual, or transgender (LGBT) person result in both short-term and longterm psychological effects for the victim(s) as well as for society as a whole. The fear and/or trauma engendered by a hate crime can impede an individual’s ability to carry out normal day-to-day activities (Bradford,
Transgender Bodies, Identities, and Healthcare
229
Ryan, & Rothblum, 1994). Moreover, it can have longer term life effects (Witten, 2004). Hate crimes, violence, and abuse are also a fact of life for a great number of transgender-identified individuals. Witten and Eyler (1999) state that, in the TranScience Longitudinal Aging Research Study (TLAR) survey; a sample of snowball sample of 213 transgender-identified individuals, 91% of the respondents stated that they had suffered perceived and actual violence and abuse. Sadly, much of this abuse and violence is suffered prior to the age of 18 years old and falls into multiple categories and is of multiple occurrence. 69.76% of the TLAR respondents stated that they had suffered some sort of violence or abuse (multiple choices of form of violence/abuse could be checked) prior to age 18. The top perpetrators of this violence/ abuse were – in order of importance – the father, another adult, a relative, the mother, and a peer. Consider the following quotations from both the TLAR and the FTM survey:1 The abuse was exploitation by a brother. I was defrauded of money (approx. $2000) and though I would not have taken action to recover it, he assured my silence by threatening to present a letter to my employer and ‘‘outing’’ me. I would call it extortion. It was several yrs ago. Not reported to authorities. Family members voiced their disapproval. Stabbed in eighth grade by schoolmate mugged by a group in 1973. While crossdressed verbally abused 1995, 1990. My early experiences in cross-dressing were discovered y and reported to my father. He caus[ed] me great embarrassment in front of the whole family. The second [time] I was caught resulted in a private consultation where I was issued the ultimatum: Stop dressing or be sent to a psychiatric institution y
These violence and abuse results are supported by the more recent work of Lombardi, Wilchins, Priestling, and Malouf (2001) and the Washington Transgender Needs Assessment Survey (WTNAS) (Xavier & Simmons, 2001, personal communication). More recent results from the Virginia Transgender Health Initiative Study (VTHIS, 2007) add additional support with 40% of the VTHIS respondents reported being physically attacked since the time they were 13 years old, including 45% of the female-to-male and 36% of the male-to-female respondents. Similar results have been reported by Kenagy (2005) for Philadelphia and for Los Angeles (LACCHR, 2006). Dang and Vianney (2007) in a sample of GLBT Asian and Pacific Islanders state that 98% of the respondents report at least one form of harassment or discrimination.
230
TARYNN M. WITTEN
Reporting of Transgender Violence and Abuse Respondents of the TLAR survey were also asked to identify whether or not they had ever told another individual about the violence, abuse, or mistreatment that they had received, and to whom these events had been reported. Of the participants who answered this question, 23% indicated that they had not told others of their abuse experiences. With respect to reasons for non-reporting, 21% indicated that they were afraid to report for fear of reprisal by the perpetrator, 11% feared abuse by the medical/legal system, 4% were unable to report, 29% felt that it would not make a difference if they had reported the incident or incidents, 8% wanted to protect the perpetrator, and 17% indicated that there had been reasons other than those listed. For the VTHIS (2007) study, over 70% of the respondents who were attacked did not report any assault to the police. The published results do not provide a breakdown of the reasons for not reporting. Fear of reprisal and fear of abuse from the systems that are supposed to protect people was frequently mentioned in both the TLAR and the FTM survey respondents: Arrested a few yrs ago for possession of cocaine – I was verbally harassed by police (‘‘you mean you have a pussy and not a dick?’’ and forced to pull my pants down in front of 4–5 cops to prove my gender status. 4 yrs ago at a demonstration cops began beating on me with clubs. Police verbal: paraded around police station for amusement – ‘‘This guy is really a woman.’’ Police also informed my employer of my transsexualism. I had been stopped and asked for ID – There had been no crime nor suspicion of crime, just a request for I.D. I had a female drivers license, so I was taken into custody for proof of identity. Released without charges.
The more recent study of Lombardi et al. (2001) reported that 59.5% of the sample experienced either violence or harassment (26.6% experienced a violent incident, 14% reported rape or attempted rape, 19.4% reported assault without a weapon, 17.4% reported having items thrown at them, and 10.2% reported assault with a weapon) and 37.1% reported some form of economic discrimination. The National Coalition of Anti-Violence Programs (2005) found that 10% of the crimes tracked by the organization in 2004 were transgender victims. While this number represents a 3% decline from the 2003 report, the researchers noted that the decline may actually be a result of many transgender people attempting to remain undetected (go stealth) rather than an actual decrease in anti-transgender attitudes. This conclusion is not surprising, given the perceptions and experiences
Transgender Bodies, Identities, and Healthcare
231
illustrated in the following cross-sectional sample of quotations from both the longitudinal TLAR survey and from this FTM survey: Mugged in NYC by a gang of black people who took all my cash. Brutally sexually mutilated in what the police said was a ‘‘drug related’’ hit on the wrong person. Police didn’t consider it serious enough to follow up on even though my penis was bisected several centimeters with a knife or razorblade. Numerous assaults while growing up. Was sexually harassed at work place, employer and employees found out that I was a transsexual, and co-workers tried to find out if I was really a man or woman by grabbing at my chest and hair and other body parts.
Gender Education and Advocacy (2005) report that, ‘‘Over the last decade, more than one person per month has died due to transgender-based hate or prejudice, regardless of any other factors in their lives.’’ Given the significant degree of perceived and actual violence and abuse against the transgenderidentified population, how does this affect healthcare utilization and healthcare delivery?
HEALTHCARE-GENDERED BODIES AND GENDERED IDENTITIES Healthcare Perceptions of Non-Western Normatively Gendered Person/Identities The institution of healthcare is not immune from participation in transgender abuse and violence. In fact, as the GLMA (2000) document clearly points out, the federal government routinely marginalizes the GLBT population and in doing so silently sanctions anti-GLBT behaviors (Witten, 2002; Belongia & Witten, 2006). Many transgender-identified individuals have experienced a variety of both subtle and overt abuse and violence at the hands of healthcare workers. Witten and Eyler (1999) demonstrate that hate crimes involving transgender people are similar in many ways to hate crimes involving lesbian, gay, and bisexual victims. This similarity is rooted in the commonality of the two groups’ transgression of traditional gender norms; whether this takes the form of sexual intimacy with a person of the ‘‘nonopposite’’ gender or if one’s own gender identity is more closely associated with another gender. Despite these similarities, Witten and Eyler (1999) concluded, from both anecdotal and survey evidence, that transgender
232
TARYNN M. WITTEN
people were simultaneously more likely to be victimized and less likely to have access to medical care and legal services. Among the numerous types of healthcare response, TLAR respondents indicated that 5.2% were placed in a psychiatric hospital, 15.7% were forced to see a counselor or therapist who tried to change them, and 2.4% were forced to have surgery (intersex identification, Greenberg, 1998; ISNA, 2007). Consider the following comment from ‘‘B’’ (an FTM-identified respondent in the TLAR study): It is always important to realize that, within the trans-population, different subpopulations will have different healthcare related problems. For example, female-tomale transsexuals who have had mastectomy will always have the problem of secrecy y Either his chest scars are obvious, or his genitals give him away. Thus, accessing normatively sexed and gendered healthcare services is nearly impossible. Add to this the difficulty of FTMs who have taken only hormones but could not afford or do not want surgeries. Billy Tipton comes to mind as one who never accessed healthcare in his lifetime and probably died prematurely because of it. There are scads of FTMs who suffer in isolation because they refuse to subject themselves to medical scrutiny, possible mistreatment and ridicule. Also, there is Robert Eades who recently died of medical neglect, after seeking help from at least 20 doctors who refused to treat him for ovarian cancer.
TLAR and FTM respondents detail a diverse distribution of abuse types ranging from non-inclusion to outright abuse and violence to denial of services (as in the case of Robert Eades). The following examples from the TLAR and FTM surveys illustrate these experiences: Went to counseling – and was taken out of the home at age 15 to mental hospital – Went back home for 5 months – went back to the hospital and then to foster parents. They [my therapists] would try to convince me to remain a man (biological sex) as it would be the most healthful and totally discourage any cross dressing.
Among the most famous healthcare abuse stories is that of Tyra Hunter, a Washington, DC hit and run victim, who was allowed to bleed to death by an EMT team when they discovered that she was a pre-operative male-tofemale transsexual. The EMT team argued that they thought she was gay and had AIDS (Fernandez, 1998).
Transgender Elder Healthcare Abuse and Violence Elderly transgender people were also noted as victims of abuse and/or violence, as their access to healthcare and caregiving services is often
Transgender Bodies, Identities, and Healthcare
233
reduced because of their transgender status as well as their elder status (Bradley, 1996; Cahill, South, & Spade, 2000; Cooke-Daniels, 1995; Witten, 2002, 2003; Witten & Whittle, 2004). More recently, Belongia and Witten (2006) reported that transgender elders are invisible with respect to eldercare facilities (see also Shankle, Maxwell, Katzman, & Landers, 2003; Watt, 2001; Witten, 2003; Witten, Eyler, & Weigel, 2000). In their study of 29 regional eldercare facilities, Belongia and Witten reported that 80% of the facilities contacted stated that participation in a one-hour lunchtime training in transgender eldercare was not relevant to their patient population and/or staff. One facility Director of Nursing had the misperception that transgender was ‘‘a homosexual thing.’’ Her disapproval of the topic was quite evident, and she refused to reconsider her position that ‘‘these people’’ are not part of her patient population. In fact, the nursing director stated that, ‘‘we don’t have that kind of client here.’’ The common institutional response seems to be a firm belief that ‘‘these people’’ are not ever patients in nursing homes or other eldercare facilities. It is important to understand that violence and abuse against trans-persons and against elderly transgender-identified persons is not just a US problem. Rather, it is a worldwide problem (Witten & Whittle, 2004). This process of making a population invisible results in the in-group’s failure to allow the out-group to be sanctioned as a minority, for example the Gay/Lesbian/Bisexual-identified population (GLMA, 2000). It then follows that, by not sanctioning the existence of such minorities, the healthcare system, as well as other macro-level institutions further condemn these groups to a future of healthcare disparity and healthcare disenfranchisement. Such a dynamic clearly flies in the face of Atdjian and Vega’s (2005) call to ‘‘immediate action to overcome disparities.’’
Institutionalized Bias, Terminology Conflation, and Marginalization in Healthcare Systems The problem of institutionalized bias and terminology conflation with respect to gender identity and sex (Basu, 2000; Doyal, 2001; Gannon, Luchetta, Rhodes, Pardee, & Segrist, 1992; Grant, 2001; Pryzgoda & Chrisler, 2000; Velkoff & Kinsella, 1998; Witten, 2003, 2005, 2007a) can be demonstrated early on in healthcare students. Witten (2004) describes a recent study in which over 2000 anonymous response surveys were sent out to all the students in the five colleges (Medicine, Nursing, Dentistry, Allied Health Professions, and Biomedical Sciences) of a major southwestern
234
TARYNN M. WITTEN
university medical center. Among other questions, survey respondents were asked to rate their perception of their gender using the Eyler-Wright gender continuum measurement instrument (Eyler & Wright, 1997). Qualitative comments were also collected from the individuals who responded to the survey. Of those that responded, a number of them expressed vehement emotions concerning the concepts of gender and sexuality. One selfidentified 24 year-old biological female medical student stated that: I feel that this survey is very sad, because the world as a whole does not understand that God in the book of Genesis made ‘‘Adam and Eve’’ not ‘‘Adam and Steve’’! I hope that you turn from your immoral ways and know that God loves you and can deliver your from this evil immoral way of thinking. There is no way to survey people on what is wrong and ungodly! Turn away from your evil ways and submit yourself to the Lord before it is too late! God bless you! God is coming SOON!!
Observe the command to ‘‘submit’’ to the Judeo-Christian-Islamic proscription of sexuality as defined within the construct of the proscribed genital sex dyad (Witten, 2005). One 19-year old, self-identified male nursing student wrote: If you were born a woman, you’re a woman, If you were born a man your a man That’s that.
Here we see the inability of the ‘‘normative’’ type to function within a conflicted reality in which the new proscription is based upon norms that are in conflict with the accepted norms of the larger cultural institution. A 22-year old, self-identified biological male medical student wrote: Biology teaches us that men are XY and women are XX. There are no other possibilities, anything else is sick!
It is important to understand that this type of response is frequently the normative response experienced by members of the transgender-identified communities (Witten, 2004; Witten & Eyler, 1999). However, this type of viewpoint is not exclusive to transgender-identified individuals. Intersexidentified individuals frequently experience similar types of pejorative remarks as well. Cheryl Chase (2002, personal communication; ISNA, 2007) tells the following story about a young intersex-identified college student and her visit to her university clinic: A college student visited the university clinic for back pain problems. When the doctor discovered that she had been treated for the intersex condition he wrote, in capital letters on her chart, ‘‘Ambiguous Genitalia.’’ The student stopped attending the clinic because of the reasonable expectation that she would be treated as a freak.
Transgender Bodies, Identities, and Healthcare
235
Atdjian and Vega (2005) further point out that the ‘‘discourse on disparities is not an academic exercise but rather a matter of life and death y it is our responsibility to our patients, to our communities and to the pursuit of social justice.’’ This is consistent with the call to arms seen in the work of Witten and Eyler (1999), who point out that transgender violence is a public health problem. These results are further supported by the work of Lombardi et al. (2001) and in a more recent publication by Patton (2006). At the writing of this chapter, transgender-identified persons continue to be murdered for their transgender identification and intersex babies continue to be surgically sexed (ISNA, 2007; see also McGhee, 2003). Given this environment, how can the healthcare system have discourse about a group that is being made invisible by that very system? This type of dynamic begs the question, ‘‘What do members of an invisible minority’’ (Shankle et al., 2003) do when the very systems of healthcare professionals who profess such a ‘‘life & death’’ viewpoint, simultaneously refuse to recognize/treat these out-group members? Or, if and when they do treat these transgender/intersex-identified individuals, the healthcare experience is perceived by the client as strongly negative, from the care recipient’s perspective (Greenberg, 1998; Goodnow, 2000; Witten & Eyler, 1999; Fernandez, 1998; Willging, Salvador, & Kano, 2006a; Willging, Salvador, & Kano, 2006b). For example, a male-to-female transsexual TLARS survey respondent stated that, I obtained an inappropriate surgery because I lied to my M.D. about being a TS. I did this because the last time I told a medical professional (University student mental health counselor) the truth they wanted to institutionalize me.
while a female-to-male transsexual TLAR survey respondent stated that: I have experienced a wide variety of abuse. From being beaten and sexually assaulted by a police officer to being gawked at by doctors, dismissed by mental health professionals
Another female-to-male transsexual responded: Previously stated at gynecological exams as requirement for testosterone shots – also laughed at by emergency staff – treated unnecessarily roughly and ignored during hospitalization.
Finally, another TLARS respondent reported that, Notations re: gender are always disclosed in medical records. Whenever insurance applications are filled out, an authorization for release of all medical records is included. Once the info is disseminated to the insurance carrier, all hope of confidentiality is lost ... providers are not TG friendly.
236
TARYNN M. WITTEN
Transgender Identities and Multiple Marginalizations: Emergent Complexities We have already established that transgender-identified persons frequently suffer a broad spectrum of life course abuse and violence (ISNA, 2007; Witten & Eyler, 1999; Lombardi et al., 2001; Witten, 2003, 2005). Further, we have seen how these individuals are further marginalized by the healthcare system as they age (Yagoda, 2005; Velkoff & Kinsella, 1998; Witten, 2002; Witten & Whittle, 2004; Cahill et al., 2000). However, these effects can be further exacerbated and confounded by additional life factors. An excellent overview of some of the relevant issues can be found in Cahill et al. (2000). Important factors to consider include such items as race (Chadiha, Proctor, Morrow-Howell, Darkwa, & Dore, 1996; Lombardi et al., 2001; Witten & Eyler, 1999), socio-economic status (Turrell, Lynch, & Kaplan, 2002; Witten, 2004; Witten & Eyler, 1999), frailty and functional limitation status (Burbank, 2006; Kelley-Moore & Ferraro, 2001), HIV/ AIDS status (Bockting et al., 1999; Earth, 2006; Manfredi, 2002; Melendex et al., 2006; Whipple & Scoura, 1989; Witten & Eyler, 1999), developmental and physical disability status (Allen, 2003; Sobsey, 1994), non-Western cultural status (Connor & Sparks, 2004; Earle, Bradigan, & Morgenbesser, 2001; Jacobs, Thomas, & Lang, 1997; Kleinman & Sung, 1979; Kulick, 1998a, 1998b; Lancaster, 1998; Teh, 2001; Wikan, 1991; Wilhelm, 2004; Winter, 2006; Witten & Eyler, 2007b), military status (Settersten, 2006; Witten, 2007b), physical location (Butler & Hope, 1999; Goins & Krout, 2006; Willging et al., 2006a, 2006b), social network structure and social status (Everard, Lach, Fisher, & Baum, 2000; Fiori, Antonucci, & Cortina, 2006; Grossman, D’Augelli, & Hershberger, 2000; Holtzman et al., 2004; Kubzansky, Berkman, & Seeman, 2000; Pinquart & Sorenson, 2000; Rautio, Heikkinen, & Heikkinen, 2001; Seeman, Kaplan, Knudsen, Cohen, & Guralnik, 1987), substance abuse status (Abrams & Alexopoulos, 1988; Earth, 2006; Ettrich & Fischer-Cyrulies, 2005; Finlon, 2002; Kausch, 2002; Lawson, 1989; Elason, 2000; Witten & Eyler, 2007a, 2007b), and prison status (Earle et al., 2001; Witten, 2007b). One excellent example of the complexities of the multiply marginalized trans-person (Nemoto, Operario, Keatley, & Villegas, 2004; Oggins & Eichenbaum, 2002) is illustrated in the following quotation from the TLAR survey: Report from the war zone I was an outreach worker on a volunteer basis with the High Risk Project Society. y Few transgendered women would go into drug rehab programs because they were housed with the males. y Sex trade workers are regularly attacked
Transgender Bodies, Identities, and Healthcare
237
and beaten and a number have died in the last year. More have died of HIV infections than were murdered.
As evidenced by the preceding discussion, successful aging is strongly affected by numerous factors that are negatively mediated by both perceived and actual violence and abuse. Moreover, it is clear, from the presented data that transgender-identified individuals experience a significant amount of perceived and actual violence and abuse throughout their life course.
TRANSGENDER PERCEPTIONS OF HEALTHCARE Part 2 of the TLAR Study (Witten & Eyler, 2007b) asked questions about healthcare needs, utilization, insurance, and problems. It also gave the respondents the opportunity to write in options that were not specified and to provide written supplemental commentary on all questions. Lastly, it asked for a summary comment addressing anything that the respondent felt had not been covered either adequately or at all in the survey.
Primary Struggles with the Healthcare System The primary respondent issues fell into five distinct areas. It is easy to see why these areas are critical to a population that falls outside of the traditional Western biblical models of sex and gender and why these areas would not necessarily come up for those who fall within the traditional definitions. Confidentiality Respondents fear that by going to the medical provider, their ‘‘secret’’ would be out and they would suffer serious ramifications (loss of job, for example). TLAR survey results show that 80% of the FTM respondents express serious concerns regarding medical confidentiality while 48% of the MTF respondents express serious concerns. The more recent VTHIS (2007) results show that 20% of the respondents felt that they were denied their job because of their transgender status or gender expression and 13% reported being fired from a job due to the employer’s reaction to their transgender status or gender expression. Thus, the need for confidentiality is central to the transgender life course.
238
TARYNN M. WITTEN
Experience/Qualifications of Healthcare Providers Most medical practitioners have no idea how to treat the healthcare needs of minority communities in general, and have little to no experience or training in dealing with the needs of the gender community. Some of the survey comments in this area were illustrative of the transgender-identified person’s experience with healthcare deliverers: Most physicians are clueless y I think the physician who prescribes my testosterone knows less than I do about relevant care issues, blood tests, etc.
Need to Educate Provider It follows as a consequence of the previous item that it is up to the transperson to train the provider, assuming the provider is willing to treat the client and is willing to be trained by a ‘‘non-professional.’’ TLAR survey results showed that 74% of the FTM respondents and 35% of the MTF respondents report having had to ‘‘educate’’ a physician about transgender healthcare in order to receive appropriate services. Safe Environment Most medical environments are not safe in that the trans-person risks being ‘‘outed’’ and, as a consequence of that, risks confidentiality and therefore all of the subsequent ramifications of privacy violation: I spent about 10 years lying to doctors and getting inappropriate treatment y I was convinced I would be institutionalized if I told the truth. I believe this fear was reasonable and based in real experience. However, since coming out as a TS, I have met several responsible and sympathetic health care professionals. I believe now that if I had told the physician who did my endometrial resection the truth, it might have been helpful, to say the least. I honestly do not know how to reconcile this conflict. I believe my experience is not unique.
Cost/Economics Since being outed is a significant risk, most trans-persons will not use their medical insurance, even if they have medical coverage. Unfortunately, because trans-related items are typically not covered under insurance, these are considered out of pocket expenses. The large number of low-income respondents support the hypothesis that many members of the gendered population are most likely not on medical insurance and probably cannot afford either the insurance or the out of pocket expenses to obtain their
Transgender Bodies, Identities, and Healthcare
239
healthcare needs, much less the added needs of being a member of the transpopulation. My insurance specifically excludes TS care, so I’m having trouble with money for medical care. Oregon Health plan excludes mental health, so I can’t afford therapy, which I need for surgery. I obtained an innapropriate surgery because I lied to my M.D. about being a TS. I did this because the last time I told a medical professional (University student mental health counselor) the truth they wanted to institutionalize me. I had serious complications from the surgery, possibly because I was on birth control pills because I could not get testosterone.
Respondent Healthcare Stories Some of the most poignant commentary concerning the experience of transgender-identified persons with the healthcare system comes from the respondents themselves. There needs to be sensitivity training for hospital personnel in particular re: transgender issues. The greatest fear I have is receiving substandard care in the event of trauma. A list of care providers sensitive to TG patients y would be helpful.
A female-to-male transsexual wrote If they have questions, they should ask and not assume knowledge they don’t have – they should know that FTMs get yeast infections, etc.
Confidentiality and the consequences of its violation are of major concern to all members of the gender community. Because being transgender-identified is not socially acceptable, the need for invisibility becomes crucial, especially during the early stages of the transition period. As we saw earlier, fear of reprisal and its economic consequences is weighed heavily when seeking out healthcare. In considering using health insurance to cover the cost of my surgery I feared I’d lose my job if word got back to my employer.
It is clear that the consequences of viewing the constructs of sex and gender from a Western biomedical model, coupled with the biblical models of sex and gender, gives rise to a total failure of the healthcare community in assessing the needs of both the intersex and transgender communities.
240
TARYNN M. WITTEN
DISCUSSION As one of the TLAR survey respondents stated: Condoned social institutions that foster hate and intolerance should be looked at. They cause much psychological damage as anything. Prevailing attitudes by society need to be changed so that all people can fit in without fear of violence, loss of job/family etc. There is room enough for everybody to live peaceable lives as they see fit.
From a macro-sociological perspective, it is well documented that ‘‘health’’ is intimately tied to position in the power hierarchy. That is, top people live longer than bottom people. Thus, in the scheme of the power hierarchy of all social institutions, including healthcare, transgender-identified individuals are invisible and therefore would be equally invisible in the health hierarchy. Tightly integrated with this status in the hierarchy or ‘‘socio-ecological embedding’’ is the critical role of early experience in influencing health and well-being over the course of the life cycle. Our research, and the research of others has demonstrated that the transgender community exists in a sociocultural-political environment that carries with it implicit daily struggles surrounding the issues of perceived and actual violence and abuse. We have demonstrated that the impact of psychosocial, biomedical, temporalcultural issues all have an impact on the life course of a transgender and/ or intersex individual. Moreover, we have shown that these factors impact the generative processes of both health and aging as a human being. The impact of our previous discussion is not localized only to the current cohort of transgender and intersex persons. It extends into the future generations to come. For example, in 1999, in the United States, the size of the age 65 years and older population was 34.7 million individuals. This subpopulation represents approximately 13% of the total population of the United States. There were 4.2 million people who were over age 85 years. The age 65 years and older population is projected to reach over 70 million individuals over the next three decades. Centenarians, individuals 100 years old or more, represent a special component of the aging population. They are the fastest growing segment of the aging population; the second fastest being the 85 plus year old population segment. For centenarians, the current estimate is 50,000–75,000 individuals. This group is expected to reach 834,000 by the year 2050. Moreover, 90% of the centenarians are women and 10% are men. This prevalence rate is approximately the same or a little higher than that of other industrialized countries. Based upon estimates of the demographics of the US population as a whole and of the demographics
Transgender Bodies, Identities, and Healthcare
241
of the transgender and intersex populations it is possible to construct a reasonable demographic of the aging transgender and intersex populations. Back of the envelope calculations demonstrates that the numbers of potentially older transgender and intersex persons is not negligible (Witten, 2002, 2003). Furthermore, if we allow for the more broad interpretation of transgender as including cross-dressing, non-surgical, gender queer, and non-Western gender, then these estimates would increase substantially. Moreover, given the demonstrated preponderance of the lack of medical coverage in all three studies, the VTHIS, the WTNAS and the TLARS surveys, given the large proportions of the population with marginal to no income, given the perceived and actual abuse, violence and marginalization experienced at the hands of the healthcare institution, and given the stigma associated with being transgendered, it is not unreasonable to project that the long-term quality of life and the success at meeting the HP2010 goals will be marginal to non-existent given the current federal policies with respect to the transgender population in general and the elders of that population in particular.
CONCLUSIONS In this chapter we have presented a study of the impact of invisibility, violence, and abuse, coupled with aging, and healthcare disparity issues for the transgender-identified community. Using the sociological argument of lifespan health effects as mediated by power inequality in the socioeconomic/political hierarchy of society, coupled with the inherent violence and abuse suffered by the transgender/intersex community, we have demonstrated that these populations are at risk for significant healthcarerelated problems, in a variety of areas; risks that may well exceed those of the ‘‘normative’’ control populations. We have seen how the stigma and social isolation of being transgender-identified leads to significant social isolation and that this isolation, coupled with the generative processes of aging, the concomitant risks associated with the transgender/intersex lifestyle, and the fiscal insecurity associated with these lifestyles are profound covariates with respect to what would be expected life cycle issues for a normative control individual. There are little to no data on international populations, with respect to mid-to-late life aging issues in the transgender and intersex communities (Witten et al., 2003). Transgender and intersex persons must go through a great deal to survive. Those that manage to live long lives as transgender-identified must have
242
TARYNN M. WITTEN
developed coping and survival strategies that were highly effective in the face of all that is against them. Understanding these coping and survival strategies can potentially benefit the normative population, particularly if these strategies can be extended to any individual in the mid-to-later stages of the life cycle. Understanding how members of the community manage to live fulfilling lives can also help us to better understand the abilities of the human being to deal with complex difficult situations and to resolve them in a fashion that can allow the person to not just simply survive, but to also have a satisfactory quality of life. The findings from this chapter and the studies therein contained clearly have important and wide-ranging implications for the US and worldwide research community. The non-traditional gender-identity population is a worldwide, non-negligible population and represents an invisible and highly stigmatized and disenfranchised minority that needs to be included in future research efforts. Healthcare institutions must understand that the costs of dealing with such an effort are not as overwhelming as might be perceived (Horton, 2005). Funders must, for example appreciate that research into healthcare minorities, must include populations that are non-traditional and may therefore not fit into preconceived socio-cultural mores. Moreover, these populations can teach us much about what is traditionally seen as the ‘‘normative’’ gendered populations. What is now needed is a deeper understanding of latent assumptions of the healthcare research system and a focus on inclusion rather than exclusion in order to address the marginalization of this worldwide population. Those with the power to change the way in which research is performed should include the implications of what is discussed in this chapter in their efforts to extend to all invisible minorities the inclusion and participation rights that they deserve as human beings.
NOTE 1. For all survey quotations, spelling and grammar has been preserved as written.
ACKNOWLEDGMENTS First and foremost, the author would like to thank all the FTM survey respondents as well as the TLARS survey respondents. Without their honesty and willingness to participate in this research effort, we would be
Transgender Bodies, Identities, and Healthcare
243
unable to provide any information on this subject of importance. The author would also like to acknowledge her colleagues A. Evan Eyler and David Bromley for their ongoing dialogue and support of this research. To find out more about the TranScience Research Institute, the research being sponsored, conducted, and/or to participate in any of its projects, you may visit the TSRI website at http://www.transcience.org/ or you may reach Dr. Tarynn M. Witten at any of the following email addresses:
[email protected] or
[email protected].
REFERENCES Abrams, R. C., & Alexopoulos, G. S. (1988). Substance abuse in the elderly: Over-the-counter and illegal drugs. Hospital and Community Psychiatry, 39, 822–823. Allen, J. D. (2003). Gay, lesbian, bisexual and transgender people with developmental disabilities and mental retardation: Stories of the rainbow support group. Binghampton, NY: Haworth Press. Alliance for Health Reform. (2006). Covering health issues. Available at http://www. allhealth.org/sourcebook2006/pdfs/sourcebook2006.pdf Atamer, Y. M. (2005). The legal status of transsexuals in Turkey. International Journal of Transgender, 8(1), 65–71. Atdjian, S., & Vega, W. A. (2005). Disparities in mental health treatment in US racial and ethnic minority groups: Implications for psychiatrists. Psychiatric Services, 56(12), 1600–1602. Basu, A. M. (2000). Gender in population research: Confusing implications for health policy. Population Studies, 54, 19–38. Belongia, L., & Witten, T. M. (2006). We don’t have that kind of client here: Institutionalized bias against and resistance to transgender and intersex aging research and training in elder care facilities. Gerontological Health Newsletter (Fall), American Public Health Association (downloaded from http://www.transcience.org/). Bockting, W.O., Rosser, S., & Coleman, E. (1999). Transgender HIV prevention: Community involvement and empowerment. International Journal of Transgender, 3(1+2): http:// symposion.com/ijt/hiv_risk/bockting.htm Bradford, J., Ryan, C., & Rothblum, E. (1994). National lesbian health care survey: Implications for mental health care. Journal of Consulting and Clinical Psychology, 62(2), 228–242. Bradley, M. (1996). Caring for older people: Elder abuse. British Medical Journal, 313, 548–550. Burbank, P. M. (Ed.) (2006). Vulnerable older adults: Health care needs and interventions. New York, NY: Springer. Butler, S. S., & Hope, B. (1999). Health and well-being for late middle-aged and old lesbians in a rural area. Journal of Gay and Lesbian Social Services, 9(4), 27–46. Cahill, S., South, K., & Spade, J. (2000). Outing aging. Report of the NGLTF task force on aging. Washington, DC: NGLTF (available at http://www.ngltf.org). Chadiha, L. A., Proctor, E. K., Morrow-Howell, N., Darkwa, O. K., & Dore, P. (1996). Religiosity and church-based assistance among chronically ill African-American and White elderly. Journal of Religious Gerontology, 10(1), 17–36.
244
TARYNN M. WITTEN
Connor, R. P., & Sparks, D. H. (2004). Queering creole spiritual traditions: Lesbian, gay, bisexual, and transgender participation in African-inspired traditions in the Americas. Binghampton, NY: Harrington Park. Cook-Daniels, L. (1995). Lesbian, gay male, and transgender elder abuse. Available at http:// www.amboyz.org/articles/elderabuse.html Dang, A., & Vianney, C. (2007). Living in the margins: A national survey of lesbian, gay, bisexual and transgender Asian and Pacific Islander Americans. New York, NY: National Gay & Lesbian Task Force Policy Institute. Doyal, L. (2001). Sex, gender, and health: The need for a new approach. British Medical Journal, 323, 1061–1063. Earle, K. A., Bradigan, B., & Morgenbesser, L. I. (2001). Mental health care for American Indians in prison. Journal of Ethnic and Cultural Diversity in Social Work, 9(3/4), 111–132. Earth, B. (2006). Diversifying gender: Male to female transgender identities and HIV/AIDS programming in Phnom Penh, Cambodia. Gender and Development, 14(2), 259–271. Elason, M. J. (2000). Substance abuse counselor’s attitudes regarding LGBT clients. Journal of Substance Abuse, 12, 311–318. Ettrich, K. U., & Fischer-Cyrulies, A. (2005). Substance abuse in middle and old age – everyday drug alcohol and nicotine: Use and abuse. Zeitschrift fu¨r Gerontologie und Geriatrie, 38(1), 47–59. Everard, K. M., Lach, H. W., Fisher, E. B., & Baum, M. C. (2000). Relationship of activity and social support to the functional health of older adults. Journal of Gerontology, 55B(4), S208–S212. Eyler, A. E., & Wright, K. (1997). Gender identification and sexual orientation among genetic females with gender-blended self-perception in childhood and adolescence. International Journal of Transgender, 1(1). Available at http://symposion.com/ijt/ijtc0102.htm Fernandez, M. E. (1998). Death suite costs city $2.9 million; Mother of transgendered man wins case. Washington Post, 12 December, Metro Section, p. C01. Finlon, C. (2002). Substance abuse in lesbian, gay, bisexual and transgender communities. Journal Gay & Lesbian Social Services, 14(4), 109–116. Fiori, K. L., Antonucci, T. C., & Cortina, K. S. (2006). Social network typologies and mental health status among older adults. The Journals of Gerontology, Psychological Sciences and Social Sciences, 61B, P25–P32. Gannon, L., Luchetta, T., Rhodes, K., Pardee, L., & Segrist, D. (1992). Sex bias in psychological research. American Psychologist, 47(3), 389–396. Gender Education and Advocacy. (2005). Transgender Day of Remembrance: About DOR. Retrieved on 20 February 2006 (downloaded from http://www.gender.org/remember/ day/what.html). GLMA. (2000). Healthy People 2010: Companion document for lesbian, gay, bisexual, and transgender (LGBT) health. U.S. San Francisco, CA. Gay and Lesbian Medical Association. Available at http://www.glma.org Godlewski, J. (1988). Transsexualism and anatomic sex ratio reversal in Poland. Archives of Sexual Behavior, 17(6), 547–548. Goins, R. T., & Krout, J. A. (Eds). (2006). Service delivery to rural older adults: Research, policy and practice. New York, NY: Springer. Goodnow, C. (2000). A tragically maimed boy, raised as a girl, comes to terms with his identity, Seattle Post-Intelligencer, 13 March.
Transgender Bodies, Identities, and Healthcare
245
Grant, A. M. (2001). Health of socially excluded groups: Lessons must be applied. British Medical Journal, 323, 1071. Greenberg, J. A. (1998). Defining male and female: Intersexuality and the collision between law and biology. Arizona Law Review, 41(2), 265–328. Grossman, A. H., D’Augelli, A. R., & Hershberger, S. L. (2000). Social support networks of Lesbian, gay, and bisexual adults 60 years of age and older. Journal of Gerontology, 55B(3), P171–P179. Harry Benjamin International Gender Dysphoria Association. (2007). http://www.hbigda.org Herek, G. (1989). Hate crimes against lesbians and gay men. American Psychologist, 44(6), 948–955. Herek, G., et al. (2002). Victims’ experiences in hate crimes based on sexual orientation. Journal of Social Issues, 58(2), 319–339. Hoenig, J., & Kenna, C. (1974). The prevalence of transsexualism in England and Wales. The British Journal of Psychiatry: The Journal of Mental Science, 124, 181–190. Holtzman, R. E., Rebok, G. W., Saczynski, J. S., Kouzis, A. C., Doyle, K. W., & Eaton, W. W. (2004). Social network characteristics and cognition in middle-aged and older adults. The Journals of Gerontology, Psychological Sciences and Social Sciences, 59B(6), P278–P284. Horton, M. A. (2005). The cost of transgender health benefits. International Journal of Transgender, 8(1). Irwin, A., Valentine, N., Brown, C., Loewenson, R., Solar, O., Brown, H., Koller, T., & Vega, J. (2006). The commission on social determinants of health: Tackling the roots of health inequities. PLOS Medicine, 3(6), 749–751. ISNA-Intersex Society of North America. (2007). http://www.isna.org/ Jacobs, S.-E., & Cromwell, J. (1992). Visions and revisions of reality: Reflections on sex, sexuality, gender, and gender variance. Journal of Homosexuality, 23, 43–69. Jacobs, S.-E., Thomas, W., & Lang, S. (Eds). (1997). Two-spirit people: Native American gender identity, sexuality and spirituality. Urbana, IL: University of Illinois Press. Kausch, O. (2002). Cocaine abuse in the elderly: A series of three case reports. Journal of Nervous and Mental Disease, 190(8), 562–565. Kelley-Moore, J. A., & Ferraro, K. F. (2001). Functional limitations and religious service attendance in later life: Barrier and/or benefit mechanism. The Journals of Gerontology, Psychological Sciences and Social Sciences, 56B(6), S365–S373. Kenagy, G. P. (2005). Transgender health: Findings from two needs assessment studies in Philadelphia. Health and Social Work, 30(1), 19–26. van Kesteren, P. J., Gooren, L. J., & Megens, J. A. (1996). An epidemiological and demographic study of transsexuals in the Netherlands. Archives Sexual Behavior, 25(6), 589–600. Kleinman, A., & Sung, L. H. (1979). Why do indigenous practitioners successfully heal? Social Science and Medicine, 13B, 7–26. Kockott, G., & Fahrner, E.-M. (1988). Male-to-Female and Female-to-Male transsexuals: A comparison. Archives of Sexual Behavior, 17(6), 539–546. Kro¨hn, K., Bertermann, H., Wand, H., & Wille, R. (1981). Nachtuntersuchung bei operierten Transexuellen. Der Nervenarzt, 52, 26–31. Kubzansky, L. D., Berkman, L. F., & Seeman, T. E. (2000). Social conditions and distress in elderly persons: Findings from the MacArthur studies of successful aging. Journal of Gerontology, 55B(4), P238–P246.
246
TARYNN M. WITTEN
Kulick, D. (1998a). Transgender in Latin America: Persons, practices and meanings. Sexualities, 1, 259–260. Kulick, D. (1998b). Fe/male trouble: The unsettling place of lesbians in the self-images of Brazilian travesti prostitutes. Sexualities, 1, 299–312. Lancaster, R. N. (1998). Transgenderism in Latin America: Some critical introductory remarks on identities and practices. Sexualities, 1, 261–274. Langevin, R. (1983). Sexual strands: Understanding and treating sexual anomalies in men. Hillsdale, NJ: Lawrence Erlbaum Associates. Lawson, A. W. (1989). Substance abuse problems of the elderly: Considerations for treatment and prevention. In: G. W. Lawson & A. W. Lawson (Eds), Alcoholism and substance abuse in special populations. Rockville, MD: Aspen Publishers. Lombardi, E. L., Wilchins, R. A., Priestling, D., & Malouf, D. (2001). Gender violence: Transgender experiences with violence and discrimination. Journal of Homosexuality, 42(1), 89–101. Los Angeles County Commission on Human Relations. (2006). 2005 Hate Crime Report. Manfredi, R. (2002). HIV disease and advanced age: An increasing therapeutic challenge. Drugs & Aging, 19(9), 647–669. McGhee, D. (2003). Joined-up government, ‘community safety’ and lesbian, gay, bisexual and transgender ‘active citizens’. Critical Social Policy Ltd., 23(3), 345–374. McLelland, M. J. (2004). From the stage to the clinic: Changing transgender identities in postwar Japan. Japan Forum, 16(1), 1–20. Melendez, R. M., Exner, T. A., Erhart, A. A., & Dodge, B., et al. (2006). Health and healthcare among male-to-female transgender persons who are HIV positive. American Journal of Public Health, 96(6), 1034–1037. Merriam-Webster Online Dictionary and Thesaurus. http://www.m-w.com/dictionary/ (last accessed on 10 August 2006). National Coalition of Anti-Violence Programs. (2005). Anti-lesbian, gay, bisexual and transgender violence in 2004 (downloaded from http://www.ncavp.org). Nemoto, T., Operario, D., Keatley, J., & Villegas, D. (2004). Social context of HIV risk behaviours among male-to-female transgenders of colour. AIDS Care, 16(6), 724–735. Oggins, J., & Eichenbaum, J. (2002). Engaging transgender substance users in substance use treatment. International Journal of Transgender, 6(2). http://www.symposion.com/ijt/ ijtvo06no02_03.htm Patton, C. (2006). Anti-lesbian, gay, bisexual and transgender violence in 2005: A report of the national coalition of anti-violence programs. http://www.ncavp.org Pesquera, A. (1999). Court to decide what’s changed in sex operations. San Antonio Express News, 3 September. Pinquart, M., & Sorenson, S. (2000). Influences of socio-economic status, social network, and competence on subjective well-being in later life: A meta-analysis. Psychology and Aging, 14(2), 187–224. Poasa, K. H., & Blanchard, R. (2004). Birth order in transgendered males from polynesis: A quantitative study of Samoan Fa’af¯afine. Journal of Sex and Marital Therapy, 30, 13–23. Pryzgoda, J., & Chrisler, J. C. (2000). Definitions of gender and sex: The subtleties of meaning. Sex Roles, 43(7/8), 553–569. Rautio, N., Heikkinen, E., & Heikkinen, R.-L. (2001). The association of socio-economic factors with physical and mental capacity in elderly men and women. Archives of Gerontology and Geriatrics, 33, 163–178.
Transgender Bodies, Identities, and Healthcare
247
Roughgarden, J. (2004). Evolution’s rainbow: Diversity, gender and sexuality in nature and people. Berkeley, CA: University of California Press. Satterfeld, S. B. (1988). Transsexualism. Special Issue: The sexually unusual: Guide to understanding and help. Journal of Social Work and Human Sexuality, 7(1), 77–87. Schmidt, J. (2003). Paradise lost? Social change and the Fa’afafine in Samoa. Current Sociology, 51, 417–432. Seeman, T. E., Kaplan, G. A., Knudsen, L., Cohen, R., & Guralnik, J. (1987). Social network ties and mortality among the elderly in the Alameda county study. American Journal of Epidemiology, 126(4), 714–723. Settersten, R. A. (2006). When nations call: How wartime military service matters for the lifecourse and aging. Research on Aging, 28(1), 12–36. Shankle, M. D., Maxwell, C. A., Katzman, E. S., & Landers, S. (2003). An invisible population: Older lesbian, gay, bisexual, and transgender individuals. Clinical and Regulator Affairs, 20(2), 159–182. Sheikh, A. (2006). Why are ethnic minorities under-represented in US research studies? PLOS Medicine, 3(2), 166–167. Sigusch, V. (1991). Die Transsexuellen und unser nosomorpher Blick. Z. Sexualforsch, 4, 225–256. Smedley, B. D., Stith, A. Y., & Nelson, A. R. (Eds). (2003). Unequal treatment: Confronting racial and ethnic disparities in healthcare. Washington, DC: The National Academies Press. Sobsey, D. (1994). Violence and abuse in the lives of people with disabilities: The end of silent acceptance? Baltimore, MD: P.H. Brookes. Teh, Y. K. (2001). Mak Nyahs (male transsexuals) in Malaysia: The influence of culture and religion on their identity. International Journal of Transgender, 5(3). http://www. symposion.com/ijt/ijtvo05no03_04.htm Tsoi, W. F. (1988). The prevalence of transsexualism in Singapore. Acta Psychiatrica Scandinavica, 78, 501–504. Turrell, G., Lynch, J. W., Kaplan, G. A., et al. (2002). Socioeconomic position across the lifecourse and cognitive function in late middle age. The Journals of Gerontology, Psychological Sciences and Social Sciences, 57B(1), S43–S51. U.S. Department of Health and Human Services (USDHHS). (2000). Healthy People 2010. With Understanding and Improving Health and Objectives for Improving Health (2nd ed., Vol. 2). Washington, DC: U.S. Government Reprinting Office. Healthy People 2010 Online Documents. http://www.healthypeople.gov/document/ Velkoff, V. A., & Kinsella, K. (1998). Gender stereotypes: Data needs for ageing research. Aging International, 24(4), 18–38. Victoria, C. G. (2006). The challenge of reducing health inequalities. American Journal of Public Health, 96(1), 10–11. Virginia Transgender Health Initiative Study (VTHIS). (2007). Available at http://www.vdh. virginia.gov/epidemiology/DiseasePrevention/documents/pdf/THISFINALREPORTVol1. pdf. Accessed on 7 March. Walinder, J. (1971). Incidence and sex ratio of transsexualism in Sweden. The British Journal of Psychiatry: The Journal of Mental Science, 118, 195–196. Walinder, J. (1972). Transsexualism: Definition, prevalence, and sex distribution. Acta Psychiatrica Scandinavica, 44(suppl.), 255–257. Watt, G. (2001). Policies to tackle social exclusion. British Medical Journal, 323, 175–176.
248
TARYNN M. WITTEN
Weitze, C., & Osburg, S. (1996). Transsexualism in Germany: Empirical data on epidemiology and application of the German transsexual’s act during the first ten years. Archives of Sexual Behavior, 25(4), 409–425. Whipple, B., & Scoura, W. K. (1989). HIV and the older adult. Journal Gerontological Nursing, 15, 15–18. Wikan, U. (1991). The Xanith: A third gender role? In: U. Wikan (Ed.), Behind the veil in Arabia: Women in Oman. Chicago: University of Chicago Press. Wilhelm, A. D. (2004). Tritiya-Prakriti: People of the third sex: Understanding homosexuality, transgender identity and intersex conditions through Hinduism. Philadelphia, PA: Xlibris Corporation. Willging, C. E., Salvador, M., & Kano, M. (2006a). Unequal treatment: Mental health care for sexual and gender minority groups in a rural state. Psychiatric Services, 57(6), 867–870. Willging, C. E., Salvador, M., & Kano, M. (2006b). Pragmatic help-seeking: How sexual and gender minority groups access mental health care in a rural state. Psychiatric Services, 57(6), 871–874. Winter, S. (2005). Heterogeneity in transgender: A cluster analysis of a Thai sample. International Journal of Transgender, 8(1), 23–30. Winter, S. (2006). Thai transgenders in focus: Demographics, transitions and identities. Internal Journal of Transgender, 9(1), 15–27. Witten, T. M. (2002). Geriatric care and management issues for the transgender and intersex populations. Geriatric Care and Management Journal, 12(3), 20–24. Witten, T. M. (2003). Transgender aging: An emerging population and an emerging need. Review Sexologies, XII(4), 15–20. Witten, T. M. (2004). Life course analysis: The courage to search for something more: Middle adulthood issues in the transgender and intersex community. Journal of Human Behavior in a Social Environment, 8(3-4), 189–224. Witten, T. M. (2005). Birth sex is not gender. Comments on ‘‘efforts to address gender inequalities must begin at home.’’ The Lancet, 366(9496), 1505. Witten, T.M. (2007a). End-of-life concerns for transgender elders and their caregivers: Graceful exits. American Society on Aging, Outward, to appear. Available as a free download at http://www.transcience.org/ Witten, T.M. (2007b). Gender Identity and the Military: Transgender, Transsexual, and Intersex-identified Individuals in the U.S. Armed Forces. http://www.palmcenter.org/ press/dadt/releases/palm_center_releases_study_on_gender_identity_in_u_s_military Witten, T. M., Benestad, E. E. P., Ekins, R. J. M., Ettner, R., Harima, K., King, D., Lande´n, T. M., Nodin, N., P’yatokha, V., & Sharpe, A. N. (2003). Transgender and transsexuality. In: C. R. Ember & M. Ember (Eds), The encyclopedia of sex and gender: Men and women in the world’s cultures. New York, NY: Kluwer Plenum. Witten, T. M., & Eyler, A. E. (1999). Hate crimes and violence against the transgendered. Peace Review, 11(3), 461–468. Witten, T. M., & Eyler, A. E. (2007a). Transgender gerontology and geriatrics: Towards successful aging. In: R. Ettner & A. E. Eyler (Eds), Principles of transgender medicine and surgery. New York, NY: Haworth press, Inc. (in press). Witten, T. M., & Eyler, A. E. (Eds). (2007b). Handbook of GLBTI aging. Baltimore, MD: Johns Hopkins University Press.
Transgender Bodies, Identities, and Healthcare
249
Witten, T. M., Eyler, A. E., & Weigel, C. (2000). Transsexuals, Transgenders, Cross-Dressers: Issues for Professionals in Aging, OutWord, Winter (downloaded from http:// www.asaging.org/LGAIN/outword-063.htm). Witten, T. M., & Whittle, S. P. (2004). TransPanthers: The graying of transgender and the law. The Deakin Law Review, 9(2), 503–522. Yagoda, L. (2005). Older adults and health disparities: The impact on access to care. InterSections in Practice NASW, 4, 6–9. Zucker, K. J., & Blanchard, R. (2003). Birth order in the Fakafefine. Journal of Sex and Martial Therapy, 29(4), 251–253.
SOCIAL SUFFERING AND GAPS IN ALTERNATIVE HEALTH CARE FOR VULNERABLE WOMEN WORKERS Leah Shumka and Cecilia Benoit ABSTRACT The purpose of this study was to examine the prevalence of social suffering among a non-random sample of Canadian women working in socially and economically marginalized ‘‘frontline’’ service occupations. Participants identified a number of health concerns that they link to the everyday suffering they endure – i.e. feeling inadequate, incompetent, lonely, selfconscious, disenfranchised or dissatisfied. The complex etiology of these women’s suffering bars many from finding appropriate health care. As a result, there are health disparities among our vulnerable populations. While they often articulated a desire for alternative/complementary care, the Canadian health care system does not currently fund these services and many of the women are unable to afford the out-of-pocket costs.
INTRODUCTION The purpose of this study was to document the social suffering of a group of women who occupy marginalized social locations in Canada. By social Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 253–275 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00011-7
253
254
LEAH SHUMKA AND CECILIA BENOIT
suffering we mean the pain and distress that can result from what is done to and by people through their involvement with processes of political, economic and institutional power (Kleinman, Das, & Lock, 1997). Social suffering manifests itself in many ways but, as noted by medical anthropologists and sociologists, it often becomes embodied as physical pain and illness by vulnerable individuals who lack the power to communicate their distress by other, more overt, means (Kleinman et al., 1997; Kleinman & Kleinman, 1991, 1997; Lock & Wakewich-Dunk, 1990; Scheper-Hughes, 1994). This is certainly the case for the women whose voices are documented here. Their words and images of pain and illness communicate a deeper, more complex, set of meanings that are intimately linked to the circumstances of their everyday experiences (Brodwin, 1992; Herzfeld, 1986). An individual’s social location – i.e. the multiple roles or statuses that a person can occupy at any given time depending on their age, gender, ethnicity, socioeconomic and health-statuses and occupation (Shumka, 2006) – can change shape as the circumstances of their lives change. What forms and coalesces the societal location of the women in our study is their engagement in one of three marginalized ‘‘frontline’’ service occupations – two in the formal economy (food and beverage service and hairstyling) and one in the informal/shadow economy (sex work). These women’s involvement in these economic activities is not necessarily the sole cause of their social suffering; however, as shown below, structural aspects of these occupations, including low pay, poor work security and lack of workplace health insurance, do bar them from accessing many non-medical alternative services1 not covered by the public health care system and enjoyed by betteroff Canadians with access to private health insurance and/or the surplus income to pay out-of-pocket costs (Armstrong & Armstrong, 2003). While some of these services are available to very low-income people on Income Assistance or, in a small number of cases, to those receiving Disability Benefits, many people working in marginalized jobs such as the women in this study, tend not to be eligible for these means-tested services and thus fall through the cracks. Below we examine in more depth how social suffering becomes embodied and results in unequal health outcomes among our purposive sample of women working in frontline service jobs. Before presenting our findings, we review the relevant literature and present the research setting and methodology. The final section discusses the relevance of our results for health care policy – upstream approaches that include holistic care for vulnerable populations.
Social Suffering and Gaps in Alternative Health Care
255
RELEVANT LITERATURE The term ‘‘social suffering’’ is defined by anthropologists Kleinman, Das, and Lock (1997) as the pain and distress that can result from what is done to and by people through their involvement with processes of political, economic and institutional power. By this definition, pain and distress refers to all manner of ‘‘wound or injury to the (mind), body and spirit’’ (Kleinman & Kleinman, 1991). This definition is broad but useful because it acknowledges an assemblage of different human experiences – including health, welfare, legal, moral, ethical, political and religious issues – that can shape individual suffering. Traditionally, the term ‘‘social suffering’’ has been applied to three areas of social science interest. There are the ‘‘contingent misfortunes’’ which refer to serious acute diseases such as cancer. This suffering can strike suddenly and affect anybody but is considered traumatic to the afflicted person because of the threat of imminent mortality. Less obvious, but more insidious are ‘‘routinized forms of suffering’’ that are attached to conditions like poverty and the attendant hunger, thirst, homelessness and contagious infections; this suffering tends to strike ‘‘the poor, the vulnerable and the defeated.’’ Finally, is the suffering resulting from ‘‘extreme conditions’’ such as war, famine, dispossession, rape and torture (which are often found in combination with one another) (Kleinman & Kleinman, 1991). As may seem readily apparent, this suffering has a strong political dimension to it and is often tied to the oppression and rejection of economically, politically and socially marginalized groups of individuals. More rarely discussed in this literature are ‘‘everyday’’ forms of suffering. Everyday is defined here as the commonplace events that impact the lives of ‘‘ordinary’’ individuals who do not necessarily fall within the category of oppressed or downtrodden. These events are quotidian, which is to say they are virtually normative. However, everyday suffering is known to be embodied in culturally elaborated ways that have specific moral and political dimensions (Kleinman et al., 1997). Pertinent here is the highly gendered nature of social suffering. As social scientists have pointed out, women are more commonly linked to the expression of pain and illness than are men (Annandale & Hunt, 2000). Reasons for this are complex but, as Das Gupta (1997) suggests, women may be more likely to embody their experiences of suffering because they have less power and autonomy to negotiate their circumstances. The author’s research into the phenomenon of pre-menstrual syndrome (PMS) and other related health concerns, such
256
LEAH SHUMKA AND CECILIA BENOIT
as menopause, anorexia nervosa and bulimia, highlights that these illnesses are indicators of women’s experiences of feeling overwhelmed by a need to fulfill contradictory social roles – mother/nurturer, home caretaker, wife, employee/employer, friend and lover (Davis, 1996). Gender inequality thus leads to health disparities. Gender also interacts with the work people do for a living, resulting in increased health disparities for those who have ‘‘unattractive’’ jobs (Lipscomb, Loomis, McDonald, Argue, & Wing, 2006). A large portion of the work that women do for pay (as well as the work they do inside the home) – often dismissively referred to as ‘‘women’s work’’ – is work of this type (England, Hermsen, & Cotter, 2000). Perhaps it is not surprising then that, according to Bendelow and Williams (1998), women workers in the United States demonstrate a higher incidence of temporary, persistent and chronic pain than men and other research and scholars have found this to be the case for Canadian women (Stephens, Dulberg, & Joubert, 1999). Bendelow and Williams (1998, p. 201) contextualize their findings with a discussion of the cultural appropriateness of expressing pain and point to cross-cultural studies in other Westernized countries. Their research shows that there is an overwhelming perception among both men and women that women are more able to cope with pain than are men. This is partly credited to women’s experience of childbirth, and thus it is hypothesized that they have a higher pain threshold than men. More intriguing is the argument that women are socialized to believe it is culturally acceptable to communicate that they are in pain, whereas men are encouraged to exhibit stoicism and fortitude. Bendelow and Williams (1998, p. 209) also report that there is a perceived hierarchy to pain whereby physical pain is considered more ‘‘real’’ and ‘‘legitimate’’ than emotional pain or anguish and that physical pain is more deserving of sympathy and respect. Numerous studies have also shown that women are more likely to experience work-related stress and or psychological distress than men (Jick & Mitz, 1985; Haw, 1982; Lennon, 1987). It is not clear from these reports, however, why women are vulnerable to work-related psycho-social stress. One suggestion is that women are more likely to work in jobs that are known to be stressful, and thus it is an issue of exposure (Lennon, 1987; Sprout & Yassi, 1995). For example, service-oriented occupations where women predominate are commonly characterized as high-stress jobs with low work control (Benoit, 2000; Statistics Canada, 2005a, 2005b). Women are thus more exposed to repetitive and monotonous work and to stressful conditions and, as a result, have a greater likelihood than men to experience back strain, skin diseases, headaches and eyestrain (Messing, 2000). What this means is that women are not necessarily more vulnerable to distress, but
Social Suffering and Gaps in Alternative Health Care
257
more likely to embody the effects (Lennon, 1987). DelVecchio-Good (1992, p. 51) notes that it is well documented that women have ‘‘higher physical and psychiatric morbidity than do men and are more likely to seek health care for their symptoms.’’ Waldren (1991, p. 20) indicates that while men are more likely to suffer occupational health injuries because they tend to work in ‘‘dangerous’’ or ‘‘hazardous’’ jobs (i.e. logging, mining and construction), women ‘‘report more symptoms, more acute conditions, more days of restricted activity due to illness and more doctor visits than do men.’’ To clearly conceptualize how social suffering becomes embodied as physical pain and illness we need to examine the recent literature on the body. At one time, the body was considered simply a biological/organic organism. More recently, however, social scientists have argued for a more sophisticated perspective of the body, one that considers it simultaneously an individual, social, cultural, historical and political entity (Frank, 1995; Messing, 1998; Moss & Dyck, 2002; Scheper-Hughes & Lock, 1987; Van Wolputte, 2004). This epistemological perspective views the body as three separate, yet overlapping units of analysis. The first body is the individual or ‘‘phenomenological body,’’ where the body is perceived, experienced and sensed in the mind of the individual. The second body is the ‘‘social body,’’ and it links individual experience to more widely shared interpretations and to social relationships. The third and final body is referred to as the ‘‘body politic.’’ This unit of analysis is more concerned with collective bodies and how they are regulated and controlled (Scheper-Hughes & Lock, 1987). Society at large has given scant attention to how social suffering becomes embodied by women workers in service industries across high-income countries. According to Karen Messing (2001), women’s occupationalhealth concerns are rarely recognized, compensated or taken seriously in the workplace. The result is that there exists little research on women’s health problems at work – instead these problems get dismissed as due to women’s reproductive events (menstruation, menopause, etc.), aging or emotional instability. Physicians, the gatekeepers to compensation claims for injuries at work as well as Disabilities Benefits are often blind to the health problems experienced by women service workers. When diagnosis is given at all, their ‘‘complaints’’ are usually categorized as a medical condition falling under the rubrics ‘‘stress,’’ ‘‘anxiety’’ and/or ‘‘depression.’’ However, as shown below, the women in our study, for the most part, do not see themselves as having mental illnesses or mental health problems. Instead they speak about the everyday emotional and social problems that are affecting their overall health and well-being. These include experiences such as a separation from a
258
LEAH SHUMKA AND CECILIA BENOIT
spouse, the inability to gain meaningful employment, feeling dissatisfied with one’s personal achievements, not getting along with family members and/or feeling pulled in too many directions (work, friendships, school and family). The complex, and often inchoate, etiology of this kind of suffering thus makes it difficult for many women service workers to find help for their health problems through the Canadian public health care system. Medical professionals often do not have the comprehensive level of training necessary to diagnose or treat these health problems and the alternative services the women would like to access – ranging from counseling, naturopathy, therapeutic massage or acupuncture – are not currently funded by the public system. Instead these services are included in the 30 percent of health care services in Canada that are paid by residents through other means (i.e. private insurance and/or workplace health benefits), as well as homecare, prescription drugs, dentists and long-term care (Armstrong & Armstrong, 2003). Highly problematic is that women working in low-paying service jobs do not have access to additional health insurance through their job. Some of them are covered under a partner’s health insurance plan but others do not have this recourse (i.e. are single, their partner does not have private health insurance, etc.) or additional funds to pay the out-of-pocket costs. Some very poor women working in the sex industry have access to alternative services though Income Assistance available in Canada but many of the other women workers are deemed to be making too much money to be eligible for this program and thus fall through the cracks of the health care system.
STUDY AND METHODS Parent Study The data presented below are drawn from a qualitative study, Working through the Body: Women, Pain and the Embodiment of Work (henceforth referred to as the Working through the Body sub-project), that builds on an ongoing mixed-methods research project entitled, The Impact of Stigma on Marginalized Populations’ Work, Health and Access to Services, lead by the second author (Benoit, Jansson, McCarthy, & Leadbeater, 2002–2005). This latter study is a longitudinal research project funded by the Canadian Institutes for Health Research (CIHR) that examines the social and health
Social Suffering and Gaps in Alternative Health Care
259
costs associated with providing emotional labor in working environments that are non-unionized, sexualized and, in the case of the sex industry, highly stigmatized. To date, the study has interviewed 306 adults living in the Victoria Metropolitan Area of British Columbia, Canada, who are working in one of three service occupations that are socially and economically marginalized to greater and lesser degrees: hairstylists and barbers (henceforth referred to as hairstylists), food and beverage servers (henceforth referred to as servers) and people working in different sectors of the sex industry (henceforth referred to as PWSI). A parallel study also lead by the second author is underway in Sacramento. These jobs differ in terms of social prestige but they share important structural characteristics that marginalize them to greater and lesser degrees. Firstly, the financial remuneration (i.e. tips and commission which these occupations depend) these women receive is contingent on successfully hiding or disguising their own emotions while managing their customers’ feelings (Hochschild, 1983; Hall, 1993). Thus, women working in these occupations are expected to follow particular ‘‘serving scripts.’’ These scripts involve ‘‘doing gender’’ by giving ‘‘good service’’ so that customers will leave satisfied and return (Hall, 1993). Secondly, all three of these occupations are disproportionately female. According to Statistics Canada, 81 percent of hairstylists and 77 percent of food and beverage servers in the study area are female (Statistics Canada, 2005a, b). It is estimated that between 70 and 80 percent of sex workers in most metropolitan areas of Canada, including Victoria, are female (Benoit & Millar, 2001). Thirdly, these occupations are marked by limited educational requirements. According to Provincial statistics, the food and beverage and hair and beauty industries require a minimum grade 10 high school education or equivalent (Industry, Training & Apprenticeship Commission, 2001). There is no minimum education requirement for PWSI. Fourthly, there is a distinct hierarchy of work settings within and between each of these occupations. Individuals can work across a variety of venues that include high prestige venues (i.e. elite hair salons, expensive fine-dining restaurants and exclusive escort agencies) to low-end locations (i.e. budget hair salons, family-style restaurants with simple food and low prices and, in terms of PWSI, street-level work). The result of this hierarchy of work setting is differing levels of prestige/stigma and significant difference in economic stability vis-a`-vis commissions and tips. Fifthly, all the jobs are marked by unstable employment which translates as frequent moves from one venue to another, high turnover of personnel over short periods of time and lack of access to employment-based health benefits.
260
LEAH SHUMKA AND CECILIA BENOIT
Sixthly, on average, all three occupations are compensated with low pay. The average income for hairstylists and barbers in Victoria in 2001 was approximately $18,000; for food and beverage workers it was even lower, at $12,700 per annum (Statistics Canada, 2005a, b). Tips, however, are not included in these reported earnings, which can be between 10 and 15 percent of the hairstylist or server’s sales per shift (Industry Training & Apprenticeship Commissions, 2001; Industry Training & Apprenticeship Commissions, 2002). The median earnings of PWSI reported in a former study in the same research site was $18,000 (Benoit & Millar, 2001). The parent study comprises four separate interviews or waves that are administered approximately every four months. Each interview is made up of a series of closed and open-ended questions that cover a range of topics including: demographic variables (age, gender, ethnicity); family history; education (early education and ongoing training or education); work history; description of current occupation (hours, schedule, satisfaction and description of working environment); income (monthly, yearly and household income); tipping (reliance on tipping, percentage of take-home wage); enacted and internalized stigma; occupational injuries (including perceived occupational risks); current physical and mental health; utilization of health care services and children’s access to and use of health care services. Data collection for the Victoria and Sacramento projects is scheduled to be completed in 2007. From September 2005 through January 2006 the first author conducted a fifth wave of interviews with a select group of women who had completed all four interview waves of the parent study and reported significant levels of physical pain, stress and fatigue during those interviews. This fifth interview focused specifically on the personal experiences of women workers for two reasons. First, women make up the majority of workers involved in these occupations; of the 306 individuals involved in the parent project, 77 percent of these are women. Second, as mentioned in the literature review above, research suggests that women are more likely than men to embody their suffering as physical pain and illness (Lennon, 1987; DelVecchio-Good, 1992). The purpose of these follow-up qualitative interviews was to better understand how women workers use their bodies and the language of pain and illness to communicate distress. In other words, how and where do women embody their distress and how do they manage the physical and emotional consequences? A second purpose was to discover what these women do to negotiate their social suffering, including what health
Social Suffering and Gaps in Alternative Health Care
261
services – conventional or alternative – they accessed, the main purpose of this chapter.
Sub-Study Methods The methods employed to get at the deep descriptions of women workers’ everyday lives included life history/illness narratives (Kleinman, 1988). Our participants were asked to speak about their personal histories (i.e., the life circumstances that led them to certain occupations and their health). We focused in considerable depth on their current experiences with pain, stress and fatigue and what they believed was the etiology of these experiences. As a result, the life histories did not entail an exhaustive summary of each woman’s life; instead the purpose was to have each woman trace her individual pathway (not necessarily linearly) to pain and illness. We were able to link this fifth wave of data with information from earlier interviews about work, family life and everyday experiences to gain a more complete picture. In the end, this approach did not necessarily produce neatly packaged illness narratives (Bury, 2001; Kleinman, 1988). Instead, the women spoke about what was important to them. Sometimes their responses revealed a clear pathway to physical, emotional and mental pain and illness; other times, answers were ambiguous and contradictory and the pathway was less clear. An additional methodological tool employed to get at women’s embodied experiences was body mapping (Cornwall, n.d.; Guillemin, 2004; Mitchell, 2004–2006). This is a visual technique based on asking participants to map their experiences of pain, stress and fatigue on an outline of their body. This process began with each woman lying on the floor on a large piece of white paper. The first author drew an outline of the woman’s entire body in a black felt marker. From there, each woman was given eight different colored felt tipped markers and asked to personalize her body map by drawing her favorite and least favorite aspects of herself; the women were also asked to draw what she felt was the strongest and weakest areas of her body. The purpose of this was to encourage women to see the drawing as their own or as a space to visually depict their lived experiences and sensations (Mitchell, 2006). Once a woman had personalized her drawing, she was asked to mark all the places where she experienced ‘‘general’’ and ‘‘work-related’’ pain, stress and fatigue. The goal throughout was to move beyond a strictly physical understanding of pain to a wider holistic understanding that might include mental, emotional and even spiritual elements.
262
LEAH SHUMKA AND CECILIA BENOIT
While the body mapping exercise yielded evocative and insightful qualitative data, we will not be presenting those findings here. This is due to space considerations and because body mapping as a methodological approach and the body maps themselves warrants a separate and more indepth discussion than is possible here.
Sub-Study Population Fourteen women were involved in this purposive sub-project: four servers, five hairstylists and five PWSI. Our participants occupy social locations shaped by a number of social factors. While many of the women were born in Canada, three others were born outside Canada – Eastern Europe, England and New Zealand – and immigrated to Canada in their late adolescence or early twenties. Two women self-identify as belonging to an ethnic minority – Aboriginal and African Canadian. The women’s ages range from 22 to 57 with the average age being 41. In terms of education, the typical level of school completion is grade 11. Forty-three percent of the women are married or involved in a common-law relationship and while many have children, only 21 percent have dependant children (i.e., 18 years old or younger). The average yearly income of the women is approximately $22,175 CAN; however, because many of these women were not working at the time of their interview (43 percent of the women are unemployed and 21 percent reported that they work only part-time) the income they report may include the money received from social-service programs such as Income Assistance and/or Disability Benefits. While owning a home in Victoria is not necessarily an index of social security due to the notoriously high cost of housing in the study area, it is, nonetheless, significant to mention that only 29 percent of the women indicate that they own their own home, 57 percent rent a house or apartment and 14 percent indicate that they lack secure housing all together. Also influencing the social location of these women is the set of factors contributing to the occurrence and degree of their suffering. Data from the parent study at the time of the current analysis revealed that a substantial number of the women participants have endured various forms of suffering throughout their lives. For instance, 17 percent of the women interviewed for the parent project indicated that they had lived in a foster home or some other form of government care at some time in their childhood or adolescence. Forty-eight percent of women indicated that they or their family had experienced ‘‘serious financial difficulty’’ in their lifetime. When
Social Suffering and Gaps in Alternative Health Care
263
asked if they had ever been a victim of abuse, 45 percent of women indicated that they had been physically abused, 44 percent said they had been sexually abused and 60 percent reported that they had been emotionally abused. Just under one-third of the women involved in the parent project specified that they had experienced all three forms of abuse. In terms of the 14 women interviewed in the sub-project, 85 percent included tales of abuse narratives. While the women only rarely used the term ‘‘abuse’’ to describe some of the traumatic events in their lives, they spoke about being physically hit or detained, inappropriately touched or sexually molested, feeling emotionally ‘‘put down’’ or ‘‘damaged’’ by close family members including romantic partners. While these high numbers may be attributable to the fact that 40 percent of the women involved in the parent study and 36 percent in the sub-project have been involved in the sex industry, the numbers are, nonetheless, disturbingly high for participants across the three lines of work. On the other hand, while these life experiences are integral to the women’s overall health and well-being, experiences of abuse and trauma were not necessarily the focus of their illness narratives and body maps. As mentioned in the introduction, it was the everyday experiences of suffering that shaped our conversations; experiences were the ones our participants linked to their health. It is significant that when women from the parent project were asked to rate their health, one-quarter said that their physical health was ‘‘fair’’ or ‘‘poor’’2 and 23 percent considered their mental health to be ‘‘fair’’ or ‘‘poor.’’ When asked ‘‘How often do you experience body pain?’’ approximately one-third of the women indicated that they ‘‘very often’’ or ‘‘always/chronically’’ experience body pain. These numbers are obviously much higher among the sub-population as their recruitment into the fifth interview was based on their reporting significant levels of pain, stress and fatigue. It should also be noted here that there was a time delay between the participant’s fourth interview with the parent project and their fifth interview with the Working through the Body study. For instance, although the average time span between the fourth and fifth interviews was four months, in some instances women were interviewed nine months after their fourth interview. This meant that, in a few cases, the women were no longer experiencing significant pain or illness at the time of their fifth interview. While initially this finding was met with concern, it was soon realized that this actually strengthens one of the central tenets of the project: that individuals embody their economic and social locations. Thus, when a woman has made significant, positive changes in her life, her experience with pain and illness is also likely to change. In addition, body mapping the
264
LEAH SHUMKA AND CECILIA BENOIT
experiences of these relatively ‘‘healthy/pain-free’’ women highlighted the potential for this method. Mapping women’s bodies through periods of sickness and health is a rich area for analytical investigation.
FINDINGS Finding Help in the Public Health Care System When we asked women to speak about their health and the circumstances that led to their current experiences with pain and illness, they were apt to tell us about their experiences interacting with health care professionals. Through these discussions, we were able to identify many positive points about Canada’s primary health care system, which are confirmed in the literature (Verhoef, Boon, & Mutasingwa, 2006). We were also able to identify a number of concerns women had in regard to their encounters with physicians. These problems overlap considerably in that they share an underlying concern with issues of communication and power – the ability to take an active role in the diagnosis and treatment of a health concern(s). In many ways, these two themes were inseparable; when women indicated that their physicians ‘‘did not listen’’ to them they were referring to verbal communication problems and a lack of power in the exchange. The problems women articulated are as follows: (a) time constraints (visits with their physicians are too short to speak in-depth); (b) rejection of patient’s etiology (physicians not validating or ‘‘hearing’’ women’s perspectives on the root cause of their health problem); (c) not addressing patient’s concerns (physicians disregarding/ignoring specific health concerns or the request for specialized treatments); one-dimensional care (physicians focusing solely on physical health and bodily ‘‘symptoms’’); inappropriate care (physicians prescribing medications that women consider inappropriate for their problem(s). Below we present the women’s own words and the main concerns articulated by them in regard to seeking the services of physicians. At the time of our fifth interview Coral3 was unemployed. Up until recently she had been working as a hairdresser in a chain of ‘‘budget’’ hair salons. Coral disliked her job and frequently commented on the embarrassment she felt at working in a ‘‘dead end’’ job.4 Coral’s dissatisfaction was not limited to her job; she also describes feeling unfulfilled in her personal life. She spoke about wanting a ‘‘real career,’’ rewarding friendships, enough money to buy a home and a ‘‘healthier’’ relationship
Social Suffering and Gaps in Alternative Health Care
265
with her husband and in-laws. She also describes her adolescence as a particularly difficult time. When speaking about her health, Coral identified a range of health problems including intense pain in her neck, shoulders, wrist, lower back and head, acid reflux, stomach ulcers and discomfort associated with being overweight. Her health concerns were such that, ‘‘I didn’t want to go to the doctor with four hundred things ‘cause I thought they’d put me in a straightjacket [light laughter].’’ When Coral did, eventually, decide to go see her physician, she described the interaction this way: I wrote down a list for my doctor and I wrote down the problems I was having and he kind of chucked [threw] it. I started crying, and he kind of chucked it down and he was like, ‘‘I don’t need this.’’ I felt like a 5 year old kid going, ‘‘Sorry,’’ you know? And he’s like, ‘‘I think you’ve been suffering from depression cause you have this chronic [pain]’’ y like it’s written down that I have chronic pain now. And um, I kept going in cause my shoulder y like I kept going in for these [health concerns] but he’d be just like ‘‘oohhhh.’’ Um, it wasn’t ever dealt with; it was never, so that day I said, I said, can I get my wrist checked? He said ‘‘oh, you don’t need to get your wrist checked, you y ’’ and he prescribed me some anti-depressants.
According to Coral, her doctor had already decided that the pain she was experiencing was ‘‘all in her head’’ and consequently he was unwilling to listen to what she felt might be the root case of some of her body pain: I always thought there was something wrong but my doctor wouldn’t help me. So it’s kind of, really strange for me, to like the doctor not helping me, and wanting to feed me antidepressants. I never took them. I’m glad I didn’t cause I did not feel depressed at all, I feel frustrated sometimes, but I do not feel depressed. I want to put somewhere [on body map] like not feeling believed when I went to the doctor. Not feeling umm y yeah y like the doctor didn’t really believe me. I didn’t feel validated by the doctor.
While Coral’s health had improved between the time of these incidents mentioned above and her fifth interview, these changes were brought about through her commitment to self-improvement and her ability to access some alternative therapies through her husband’s health insurance (she was able to receive both personal and marital counseling and limited therapeutic massage). Coral was not alone in feeling disillusioned by the care she received through primary health care. Sam is a 46-year-old woman working in the sex industry. She took up this line of work in her thirties because it paid more and was less physically demanding than the minimum wage, manual labor jobs she had in the past. Furthermore, Sam felt that she got ‘‘more respect’’ from her clients in the sex industry than she ever did from the
266
LEAH SHUMKA AND CECILIA BENOIT
romantic partners in her life. At the same time, Sam disliked the inconsistent wages and bad weather associated with street work. At the time of our interview she was single and sharing a low-income apartment with a friend. Similar to Coral, Sam had health concerns that ranged from arthritis and gastro-intestinal problems (acid reflux, stomach pain and irritable bowel syndrome (IBS)) to what she describes as ‘‘chronic pain’’ (neck, back and shoulders). When Sam was asked to talk about what she felt might be causing the chronic pain and illness, she responded: [F]rom what I’ve been ‘‘eating.’’ By eat I mean the crap I take and the feelings I put down. I’ve been pushing down for years and, and, and the result is the IBS, the arthritis and definitely the stress headaches.
When asked about her interactions with physicians, Sam became frustrated: I wish he would listen to me a little bit more. I wouldn’t mind giving root therapy a chance. On my exit from that counselor I was seeing, [root therapy] was suggested [as] an alternative to help me with ah, dealing with stress and stuff y [M]y doctor just went over that one and he’s more into ‘‘let’s get you on pills’’ and I’ve always ah, been one who didn’t like to do a lot of pills, I want to try and find a different approach. I’d like to find out what the problem is. You know, what’s the problem?!
While Sam was eventually able to access free acupuncture for her chronic neck and shoulder pain and counseling programs through a local community outreach program for people working in the sex industry, Sam’s describes her access to these services as precarious5: I’ve been going to [outreach program] the last four months and we’ve been doing a lot of step work. I don’t have an alcohol problem but you get all the childhood crap brought up. I’m getting the acupuncture through [outreach program]. [outreach program] end in two and half months, that scares me. The program I’m in will be done in two and half months. And I’m going to y acupuncture’s out the window and that bar [referring to a ‘‘bar’’ of pain across shoulders] could be right back.
Other women, such as Tobi, feel as though their health care requests are ignored and disregarded by their physicians. Tobi is a self-employed hairstylist in her mid-thirties. She has a common-law spouse who is unemployed and together they have three dependant children. While Tobi loves her occupation, she has to be available to work at least 6 days a week to make ends meet. As she puts it, ‘‘y if I don’t work, I don’t get paid.’’ The stress of having to support herself and her family on a single income takes a toll on Tobi. Not only does she not have time for exercise (which has led to significant weight gain) but Tobi is concerned that she is not living up to
Social Suffering and Gaps in Alternative Health Care
267
her idea what a ‘‘good’’ parent should be: I’m not a very good mom. I’m not saying that in a negative way. I’m just not your stayat-home typical, really ‘‘Good Housekeeping,’’ mom. So it’s an effort for me and um, I don’t always look forward to it. And I like being at the shop but I don’t always like being at home. I feel good about what happened in the day but I don’t always feel good about having to go home and deal with dinner and that time, things like that.
These everyday concerns, in turn, have had a negative impact on Tobi’s health. Through body mapping Tobi identified stomach ‘‘issues,’’ chronic headaches and physical discomfort/low self-esteem related to her weight gain. When asked how she ‘‘deals’’ with these problems Tobi responded: I get constant headaches, always have. Um, no one’s ever delved into it enough to y give me any insight into that. If there is a problem, a lot of the time [my doctor] just brushes it off. Like I’ve been getting headaches for years.
Tobi goes on describe another incident when her doctor was seemingly unresponsive to her request for a referral: I have behavior problems with one of my daughters. And I’ve told [doctor] over and over that I want to see a child psychologist or a y professional to deal with her behavior issues and [doctor] just won’t do that. I don’t think he thinks it’s serious. He says ‘‘well what’s the problem.’’ You know. So and I know it’s a huge problem. I know it’s something we can’t deal with at home.
While Tobi identifies the need to find a more responsive physician and communicates the desire to receive alternative health therapies, her circumstances make it difficult: [T]here’s a lot of things that I feel that I could be doing for myself. I could be doing y um, chiropractic and massage um y alternatives but y [brief interruption, while someone comes into her salon] y taking the time off work to go do it and the money situation, um, paying it out [of pocket]. [If] I make an appointment, I don’t know that I’m gonna have money that day. Cause I don’t always.
Seeking Alternative Health Services When women communicated their concerns with the current delivery of health care, they invariably described what kind of health care they want to receive. They did this in two ways: (a) by describing something that they have, as of yet, not received and (b) by describing the alternative services they have actively sought to fill the gaps in the public health care system. The overwhelming description of what women want is non-medical/ non-conventional health care – in other words, holistic health care that
268
LEAH SHUMKA AND CECILIA BENOIT
considers the interconnection of mind, body and the social environment while being both preventative and therapeutic (Saks, 1992, 2001). Alternative or ‘‘complementary’’ care (Sharma, 1992; Verhoef et al., 2006) exists in the shadow of the Canadian public health care system and receives almost no support from the medical establishment and the government (Kelner, Wellman, Pescosolido, & Saks, 2000). Athena is in her late twenties. She was originally recruited into our parent project through her involvement in the food and beverage industry. At the time of our fifth interview Athena was still working part-time as a restaurant manager but she had also started up her own small business. Athena was also one of the few women who had made significant improvements in her health between her fourth and fifth interview and consequently described herself as ‘‘in the best health of my life’’ when we last spoke to her. Athena’s illness narrative included a sustained period of ‘‘poor’’ health, including recurring fatigue and body pain. She linked her poor health to ‘‘not listening’’ to her body and ‘‘not being very happy.’’ As Athena puts it: I look back and think ‘‘jeez I haven’t really been that happy’’ because I have never really gone with what I feel like doing. [Instead] a lot of the time I do what I think I should do or what other people feel I should do.
During the past several years, Athena has gradually been learning about what it takes for her to be healthy. This has involved getting into ‘‘natural medicine and alternative medicine over the last four years, [and] that’s been the greatest catalyst to me feeling better.’’ When asked about why she no longer seeks conventional medical care, Athena states: Unless I’ve lost a limb I don’t go and see a regular doctor anymore y we [herself and naturopath] discuss the work that I do and the hours and sleeping patterns and basically my entire day, which work is always a part of, and um, and so he tries to understand what I do I think more than anyone else has. He pays a lot of attention to that because to me it’s smaller changes that are going to make a difference in my life, or my energy level or how I feel, not major ones.
Athena then goes on to comment on the added expense of going to see a naturopath: I think the money that I spend there [at the naturopath] is y well, well, well worth it. And sometimes I see [naturopaths and acupuncturists] for preventative things and sometimes I see them to actually treat something that I have. But, they do a better job for me than any general practitioner has ever done. So I don’t really use our medical system that we pay for. I wish that part of it [alternative therapies] was paid for.
Athena was not alone in her endorsement of alterative care. For instance, Brandy is a woman in her middle to late twenties. She too has worked in the
Social Suffering and Gaps in Alternative Health Care
269
food and beverage industry but at the time of our fifth interview she had gotten a ‘‘great’’ job with the Provincial government. While Brady, like Athena, had recently been through a difficult and unhappy period of her life, she described herself during the fifth interview as also being in ‘‘the best health’’ of her life as well. She described the unhappy times in her recent past as being brought on by relationship problems, poor self-esteem and a general lack of direction in her life. In turn, these everyday sources of suffering had caused her to experience significant physical pain and illness including chronic pain in her back and shoulders, severe anxiety and stomach pain. When she looks back on the conventional medical care she received at that time, she communicates her frustration and dissatisfaction: I remember I went to a drop-in clinic and um I had broken up with my boyfriend and I was going through a horrible break-up, this was like two years ago. I went to a drop-in clinic, and he [doctor] prescribed me sleeping pills. Like hey, that’s just wrong! Looking back on it, you don’t know anything about me, and you just go prescribing drugs to fix something! Which I think, that’s just horrible.
Shortly after this incident, Brandy started to see a naturopath. When asked why she decided to make this switch she says: medical doctors, I think, should only really be used if you know you broke your leg or you broke, you know, something like that. For the most part, if you take care of your body and you look at the whole picture and you’re not just treating one little symptom or something, you know like, you’re looking at everything, then um, a naturopath is the way to go to overall health.
Brandy was lucky because she had personal resources vis-a`-vis her new job to purchase natural remedies and seek the ongoing advice of a professional counselor. This ability to purchase alternative health services to fill the gaps in the primary health care system was not, however, the case for all the women we spoke to. Evelyn is a single mother in her late twenties. She does not currently have a ‘‘square’’ (i.e. regular, mainstream employment) job but hopes to one day upgrade her skills to get a job she finds fulfilling. At the present time, Evelyn is on Income Assistance and, to supplement her meager income, works sporadically in the sex industry. When asked about the health care she receives, Evelyn is very articulate about what she wants and needs: Well, I think rather than looking on what I’m experiencing more as a symptom, right, like, coming in with a sore back, right, certainly helping me with my sore back would be one way of doing it and that would be cool, but I think you know, helping me learn new ways [to] prevent getting a sore back [and] giving me a handful of condoms right before I leave is preventative measures. Or, you know, help me become aware of the long term
270
LEAH SHUMKA AND CECILIA BENOIT
effects of different ways of coping, even if it’s just emotional coping. Asking me if, asking me questions like if I’m eating properly or becoming anorexic or bulimic. Those sorts of things. But of course that takes longer than 15 minutes, so.
What is concerning about Evelyn’s situation is that while she is able to access some of the alternative services she needs, her situation is precarious and it forces her to make some difficult life decisions: I have been thinking about going back full-time into the [sex] trade. I don’t know if that’s gonna be good for me or not, right? And the problem is that if I get out of the system [social welfare] then all of my supports leave as well. Like it’s very difficult to find a good therapist. I finally found a good therapist and ah I need to stay in the system in order to get the therapy that I need. So even if I were to get a job – which is part of, you know, one of the programs that I’m doing which is about looking forward and train[ing] you to get a job or at least, get you moving – If I get a job I lose my therapy.
Evelyn was not the only woman we spoke to who had to rely on Income Assistance and community outreach programs to provide the alternative health care they need to get well. Sage is a single woman in her late forties. She collects Income Assistance and has only recently been able to get a phone (first time in four years) and a television. Like Evelyn, Sage has worked in the sex industry but left very recently. Her exiting was precipitated by a long stay in the hospital due to a ‘‘horribly painful’’ infection that was brought on by the use of ‘‘dirty coke.’’ While Sage describes herself as ‘‘on the mend,’’ she requires ongoing medical aid to help her deal with her addictions, arthritis, irritable bowel syndrome and ‘‘unhealthy thoughts.’’ Sage is very pleased with the health care she receives through a community outreach program: They’re really helping me pull through. They’ve given me a rent incentive that’s a little bit extra money a month towards my rent. They provide all kinds of self-healing and group therapy, you name it, everything. And you know they’re really helping me get back on my feet.
Sage is fortunate because, at the moment, she is getting exactly the kind of non-conventional alternative health care she wants. She has access to numerous alternative therapies and she received care from a volunteer doctor at the community outreach center who is attentive and responsive to all of her concerns: A doctor there who um volunteers on Monday morning [sent me to] go get some X-rays. My X-ray looked like I had bone spurs on it. ‘‘Oh no!’’ So she sent me for a cat scan and then it was, it was discovered it was [an] infection.
Social Suffering and Gaps in Alternative Health Care
271
While this situation is a vast improvement from the life Sage had recently been living she, like Evelyn and Sam introduced above, is reliant on continued funding through this community outreach program. When she leaves Income Assistance, Sage will not have access to these alternative health benefits she currently relies on.
DISCUSSION AND CONCLUSION We have presented findings from a qualitative study of Canadian women working in three frontline service occupations (food and beverage servers, hairstylists and sex industry workers). Our participants identified a number of health concerns that they link to the everyday suffering they endure – feeling inadequate, incompetent, lonely, self-conscious, disenfranchised and/or dissatisfied – indicating that socioeconomic status, gender and work interact to create health disparities as well as barriers to accessing the health services certain women desire to ease their suffering. First of all, many of these women do not have the economic resources to purchase the services they feel would have positive preventative and therapeutic benefits. This is not to say that none of the women documented here are accessing alternative health resources. Rather, access is precarious and largely dependant on: (a) Income Assistance (for those who are very low income) or (b) spouses/partners who cover the participant under an employment insurance plan. Secondly, many of the women described health problems that are complex and rooted deep within their social, cultural and economic environment. As a result, they deemed physicians not to have the time or expertise to explore and adequately address these particular concerns. As a result, women complained that their physicians were focused on diagnosing only acute illness and taking a one-dimensional view of their health, which, in the long run, does little to improve their overall health and well-being. Not all the women we spoke to were dissatisfied with their physicians. Some women indicated that they ‘‘like’’ their physicians and that their physical health concerns are addressed adequately. In some instances, women narrate stories of doctors who have gone to extraordinary lengths to ensure that they received the best possible care. However, in almost every instance, it was volunteer doctors and nurses at street-based outreach centers who earned this kind of praise. Although the publicly funded, universal health care system in Canada takes us a long way in the equitable access to medical and hospital services
272
LEAH SHUMKA AND CECILIA BENOIT
(Benoit, 2003; Armstrong & Armstrong, 2003), our findings suggest that it falls short in regard to the health care needs of vulnerable women and, as such, other strategies are required to sensitize health services to meet their individual needs. It comes as no surprise that research on the kinds of people who access alternative health care in Canada and other high-income countries tend to be mainly older women who are well-educated, employed in more prestigious professional occupations and earning good incomes (Kelner & Wellman, 1997; Kelner et al., 2000). Unfortunately, women working the frontline are not so lucky. These findings are preliminary due to our small sub-sample size and need for further confirmation via analyses of the larger sample of participants involved in the parent study. Additional thought is also needed to address the health care gaps identified. While the women we spoke to were clear that our publicly funded health care system should expand to include an array of alternative services, how best to integrate these into the current health care system remains a major challenge, as does finding the resources to do so.
NOTES 1. Referred to by some as ‘‘Complementary Alternative Medicine’’ or CAM (Verhoef et al., 2006). 2. The response categories for this question were ‘‘Excellent,’’ ‘‘Very Good,’’ ‘‘Good,’’ ‘‘Fair’’ and ‘‘Poor.’’ 3. We use pseudonyms throughout in order to help protect our participants’ anonymity and confidentiality. 4. It should be noted that many of the individuals in both the parent and subprojects who worked as hairstylists derived personal pleasure and pride from working in what is viewed by many as a skilled trade. 5. The ‘‘outreach program’’ mentioned throughout (which shall remain nameless to ensure the privacy and confidentiality of our participants) is a local social welfare service organization staffed in large part by former sex industry workers. It offers temporary access to education and alternative health services to adults working in the sex industry in the city. Most of these services are funded by the Ministry of Employment and Income Assistance. Their availability varies from year to year depending on monies forthcoming to the community outreach program from the Ministry.
REFERENCES Annandale, E., & Hunt, K. (Eds). (2000). Gender inequalities in health. Philadelphia: Open University Press.
Social Suffering and Gaps in Alternative Health Care
273
Armstrong, P., & Armstrong, H. (2003). Wasting away: The undermining of Canadian health care. Toronto: Oxford University Press. Bendelow, G. A., & Williams, S. J. (1998). Natural for women, abnormal for men: Beliefs about pain and gender. In: S. Nettleton & J. Watson (Eds), The body in everyday life (pp. 199–217). New York: Routledge. Benoit, C. (2000). Women, work and social rights: Canada in historical and comparative perspective. Scarborough, ON: Prentice Hall Canada. Benoit, C. (2003). The politics of health-care policy: The United States in comparative perspective. Perspectives in Biology and Medicine, 46(4), 592–599. Benoit, C., Jansson, M., McCarthy, B., & Leadbeater, B. (2002–2005). The impact of stigma on marginalized populations’ work, health and access to services. Canadian Institutes of Health Research. Benoit, C., & Millar, A. (2001). Dispelling myths and understanding realities: Working conditions, health status, and exiting experiences of sex workers. Sponsored by Prostitutes Empowerment, Education and Resource Society (PEERS). Funded by BC Health Research Foundation, Capital Health District and BC Centre of Excellence on Women’s Health. Brodwin, P. E. (1992). Symptoms and social performance: The case of Diane Reden. In: M. J. DelVecchio-Good, P. E. Brodwin, B. J. Good & A. Kleinman (Eds), Pain as human experience: An anthropological perspective (pp. 77–99). Berkeley: University of California Press. Bury, M. (2001). Illness narratives: Fact or fiction? Sociology of Health and Illness, 23(3), 263–285. Cornwall, A. (n.d.). Body mapping: Bridging the gap between medical messages, popular knowledge and lived experience. In: Body Mapping in RRA/PRA, RRA Notes, 16, pp. 69–76. Das Gupta, M. (1997). Life course perspectives on women’s autonomy and health outcomes. American Anthropologist, 97(3), 481–491. Davis, D. (1996). The cultural constructions of the premenstrual and menopause syndrome. In: C. F. Sargent & C. B. Brettell (Eds), Gender and health: An international perspective (pp. 57–86). New Jersey: Prentice Hall. DelVecchio Good, M. J. (1992). Work as haven from pain. In: M. J. DelVecchio-Good, P. E. Brodwin, B. J. Good & A. Kleinman (Eds), Pain as human experience: An anthropological perspective (pp. 49–76). Berkeley: University of California Press. England, P., Hermsen, J., & Cotter, D. (2000). The devaluation of women’s work: A comment on Tam. The American Journal of Sociology, 105(6), 1741–1751. Frank, A. (1995). The wounded storyteller: Body, illness and ethics. Chicago: The University of Chicago Press. Guillemin, M. (2004). Understanding illness: Using drawings as a research method. Qualitative Health Research, 14(2), 272–289. Hall, S. (1993). Smiling, deferring and flirting: Doing gender by giving good service. Work and Occupations, 20(4), 452–471. Haw, M. A. (1982). Women, work and stress: A review and agenda for the future. Journal of Health and Social Behavior, 23(2), 132–144. Herzfeld, M. (1986). Closure as cure: Tropes in the exploration of bodily and social disorder. Current Anthropology, 27(2), 107–119. Hochschild, A. R. (1983). The managed heart: Commercialization of human feelings. Berkeley: The University of California Press.
274
LEAH SHUMKA AND CECILIA BENOIT
Industry, Training and Apprenticeship Commission. (2001). Cosmetology, work from work futures BC: Occupational profiles (electronic version) (accessed on April 2006, available at http://www.workfutures.bc.ca/profiles/profile.cfm?noc=6271olang=enosite=grap). Industry, Training and Apprenticeship Commission. (2002). Hospitality, work from work futures BC: Occupational profiles (electronic version) (accessed on April 2006, available at http:// www.workfutures.bc.ca/profiles/profile.cfm?noc=645olang=enosite=graph). Jick, T. B., & Mitz, L. F. (1985). Sex differences in work stress. The Academy of Management Review, 10(3), 408–420. Kelner, M., & Wellman, B. (1997). Health care and consumer choice: Medical and alternative therapies. Social Science and Medicine, 45(2), 203–212. Kelner, M., Wellman, B., Pescosolido, B., & Saks, M. (2000). Complementary and alternative medicine: Challenge and change. Amsterdam: Gordon and Breach/hap. Kleinman, A. (1988). The illness narratives: Suffering, healing, and the human condition. New York: Basic Books. Kleinman, A., Das, V., & Lock, M. (1997). Introduction. In: A. Kleinman, V. Das & M. Lock (Eds), Social suffering (pp. ix–xxvii). Berkeley: University of California Press. Kleinman, A., & Kleinman, J. (1991). Suffering and its professional transformation: Toward an ethnography of interpersonal experience. Culture, Medicine and Psychiatry, 15(3), 275–301. Kleinman, A., & Kleinman, J. (1997). The appeal of experience; the dismay of images: Cultural appropriation of suffering in our times. In: A. Kleinman, V. Das & M. Lock (Eds), Social suffering (pp. 1–24). Berkeley: University of California Press. Lennon, M. C. (1987). Sex differences in distress: The impact of gender and work roles. Journal of Health and Social Behaviour, 28(3), 290–305. Lipscomb, H., Loomis, D., McDonald, M., Argue, R., & Wing, S. (2006). A conceptual model of work and health disparities in the United States. International Journal of Health Services, 36(1), 25–50. Lock, M., & Wakewich-Dunk, P. (1990). Nerves and nostalgia: Expression of loss among Greek immigrants in Montreal. Canadian Family Physician, 36, 253–258. Messing, K. (1998). One-eyed science: Occupational health and women workers. Philadelphia, PA: Temple University Press. Messing, K. (2000). Ergonomic studies provide information about occupational exposure differences between men and women. Journal of the American Medical Women’s Association, 55, 72–75. Mitchell, L. (2006). Child-centered? Thinking critically about children’s drawings as a visual research method. Visual Anthropology Review, 22(1), 60–73. Moss, P., & Dyck, I. (2002). Working through theories of the body. In: P. Moss & I. Dyck (Eds), Women, body, illness: Space and identity in the everyday lives of women with chronic illness (pp. 19–34). Lanham, MD: Rowman & Littlefield. Saks, M. (1992). The Paradox of Incorporation: Acupuncture and the medical profession in modern Britain. In: M. Saks (Ed.), Alternative medicine in Britain (pp. 183–200). Oxford: Clarendon Press. Saks, M. (2001). Alternative medicine and the health care division of labour. Current Sociology, 49, 119–134. Scheper-Hughes, N. (1994). Embodied knowledge: Thinking with the body in critical medical anthropology. In: R. Borofsy (Ed.), Assessing cultural anthropology (pp. 229–239). New York: McGraw-Hill.
Social Suffering and Gaps in Alternative Health Care
275
Scheper-Hughes, N., & Lock, M. M. (1987). The mindful body: A prolegomenon to future work in medical anthropology. In: J. Brown (Ed.), Understanding and applying medical anthropology (pp. 208–225). California: Mayfield Publishing. Sharma, U. (1992). Complementary medicine today: Practitioners and patients. London: Routledge. Shumka, L. (2006). Working through the Body: Women, pain and the embodiment of work. Unpublished Master’s Thesis, University of Victoria. Sprout, J., & Yassi, A. (1995). Occupational health concerns of women who work with the public. In: K. Messing, B. Neis & L. Dumas (Eds), Issues in women’s occupational health: Invisible (p. XX). Charlottetown: Gynergy books. Statistics Canada. (2005a). The 2001 Census (Electronic Data file). Retrieved on April 2006, available at http://www.statcan.ca/start.html Statistics Canada. (2005b). Women in Canada: A gender-based statistical analysis (5th ed.). Ottawa: Statistics Canada. Catalogue no. 89-503-XIE. Stephens, T., Dulberg, C., & Joubert, N. (1999). Mental health of the Canadian population: A comprehensive analysis. Chronic Diseases in Canada 1999, 20(3), 118–126. Van Wolputte, S. (2004). Hang on to your self: Of bodies, embodiment, and selves. Annual Review of Anthropology, 33, 251–269. Verhoef, M. J., Boon, H. S., & Mutasingwa, D. R. (2006). The Scope of naturopathic medicine in Canada: An emerging profession. Social Science and Medicine, 63, 409–417. Waldren, I. (1991). Effects of labor force participation on sex differences in mortality and morbidity. In: M. Frankenhaeuser, U. Lundberg & M. Chesney (Eds), Women, work and health: Stress and opportunities (pp. 17–38). New York: Plenum Press.
AGE, BREAST CANCER AND DISPARITY IN HEALTH KNOWLEDGE AND TREATMENT Lisa Cox Hall ABSTRACT This chapter focuses on the differences that younger, middle-aged, and older women with breast cancer experience, particularly in health knowledge and treatment. These differential experiences, in part, stem from our youth oriented culture. This ideology extends into medicine and can affect day-to-day medical practice. Differential experiences are, therefore, likely to result in inequality and disparity in health and in healthcare. It is argued that older women are less empowered than their younger counterparts to display the same degree of agency. This analysis has important implications for health care professionals in the treatment of older women with breast cancer.
INTRODUCTION Breast cancer is the most common form of cancer among women in the country. Approximately 200,000 women will be diagnosed with breast cancer each year. That translates to one-out-of-eight women being at risk Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 277–306 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00012-9
277
278
LISA COX HALL
for the disease within her lifetime. It is a significant matter that the risk of getting breast cancer increases as a woman ages. Just as risk increases as a woman ages, so does the probability of her dying from the disease (www.cancer.org). Seventy-three percent of breast cancer cases occur in women over age 60, yet that fact is severely under-represented by popular culture. That is, accessible, highly visible media sources focus on the younger women with the disease, who make up only 11% of actual cases. Specifically, 84% of breast cancer feature stories are of women under age 50, and only 2% are of women over age 60 (Lantz & Booth, 1998; Marino & Gerlach, 1999; Burke et al., 2001). Such misrepresentation reveals the ageist ideology in our society, and because there is evidence to indicate that such ideology extends into medical practice, it is likely that women of different ages will have different experiences when dealing with breast cancer. These differences, consequently, may be indicators of, and may result in, disparities in health care.
BACKGROUND There are several reasons to expect that women’s experience with breast cancer would differ in significant ways for women of different ages. First, age is socially constructed. That is, aging is a process that begins with birth and extends over the life-course (Moody, 2002) and is steeped in history and is subject to our dynamic values and perceptions. Second, our culture is youth oriented. Throughout the modern era, our culture has tended to value youth while holding disdain for old age (Cole, 1992). Third, humans are aged by culture, which means that each of us is taught to submit, in a normatively appropriate manner, to society’s prescriptive AGE behaviors. We play the role that society assigns, because it is an institutionalized force in which each of us has participated for a lifetime. Fourth, breast cancer is a sexually charged and highly feminized disease. There are deep cultural meanings associated with the breast (Rosenbaum, 1994). Breasts are, or at least are representations of what is, erotic. Ninety-nine percent of breast cancer cases occur in women. Beyond this literal fact, however, is the realization that the disease has knowingly been framed by society as a feminine one. This is evidenced by the pink, rather than the blue or white ribbons. Such feminized symbols, it is argued, are more easily applied to and associated with younger women. Make no mistake, cancer in other parts of the body such as the stomach or colon, is in no way comparable to the laden, shrouded – even mythologized – essence of cancer in the
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
279
breast. Fifth, researchers suggest that women are not the same when they are young as when they are old. This is due to women’s physical, psychosocial and structural circumstances that change over the life-course (Moen, 1996). These changes contribute to differences in experience for a woman at different stages of her life, for women of different ages, and for women at various historical periods in time (the age-period-cohort effect). There are competing views about which group is likely to have a more stressful experience of illness. Some think older women are more likely to experience difficulty with illness such as breast cancer. Older women are perceived (and studies have confirmed) as having less emotional support from family and friends than younger and middle-aged women have. Thus, it is reasoned that older women have less mastery (Schieman & Turner, 1998), a smaller sense of control and, therefore, less well-being (Smith et al., 2000). Perceived loss of control, reduced sense of competence, and perceived powerlessness can occur in old age, or with disability or illness. Such perceptions or occurrences threaten identity and are considered ‘‘sites for anger’’ and, therefore, reduce well-being (Schieman, 1999). When education is used as a proxy measure for socioeconomic status and when it is linked to morbidity rates and depression, researchers have found that those with less education are at greater risk for more and earlier illnesses. More and earlier illnesses are correlated with depression, or lowered sense of well-being (Meich & Shanahan, 2000). Since younger and middle-aged women have had greater opportunities for education than older women have, older women today, may be at higher risk for illness and depression and reduced sense of quality of life. Others think that older women are less likely to experience difficulty with illness such as breast cancer. One of the reasons is that older women have a reduction in ‘‘sites for anger’’, such as family roles (parenting) and work roles, which cause stress (Schieman, 1999). Research shows that older people actually report lower levels of anger than middle-aged adults do. Contributions to lower anger among older adults include differences in role status, personal and social circumstances, health and perceived control, and socio-emotional outlook (Schieman, 1999). Less anger is one indication of higher well-being. Wyatt and Friedman (1998) found that there were no differences between middle and older-aged women’s descriptions of the quality of their lives or in the demands of their illness, after breast cancer surgery and treatment. Other studies have found older persons reporting more ease in coping with stressful life events than younger and middle-aged persons. It is suspected that expectations change over the life-course, and/or
280
LISA COX HALL
that a lifetime of events has a cumulative effect on the ability to cope (Ryff, Marshall, & Clarke, 1999). Put together, these research findings and related theoretical arguments suggest that older women are likely to have different experiences with breast cancer than younger women. Despite these indications, very little critical investigation has taken place. Though the medical research literature on older women’s health beliefs about breast cancer risk and treatment has increased in the past 12 years, we still know relatively little about the psychosocial dimensions of breast cancer for women over the age of 60 compared to younger women. We know even less about women over age 70. This trend exists in the medical research literature, the social science research literature, and to an even greater extent in the popular culture media. The few medical and social science studies that have addressed older women and their psychosocial experience of breast cancer emphasize the importance of and need for further research of this kind (Yarbrough, 2004). Today, 50% of all new breast cancers occur in women aged 65 years and older (Holmes & Muss, 2003). It is projected that the number of elderly women in the U.S. diagnosed with breast cancer will increase 72% by the year 2025 (Alberg & Singh, 2001) due in part to the fact that the population of older women is increasing. Given that, in the future, our nation will experience an unprecedented increase in the burden of older women with breast cancer (Edwards et al., 2002), it is of critical importance that we know how to manage and provide the best care and services for such persons. In addition, it would be ideal to break down age related myths, ageist practices and cultural youth biases so that older women’s experiences are not disenfranchised.
LITERATURE REVIEW Illness is socially constructed (Friedson, 1970; Mishler, 1981). Breast cancer is certainly socially constructed, which means that cultural assumptions and biases influence our knowledge, perceptions, and experiences of breast cancer in this society. (Kasper & Ferguson, 2000, p. 2)
Society’s construction of breast cancer, which affects women’s experiences, takes place in a climate of politics, economics, and self-interest. The politics, or the value judgments and power relations in society, behind the disease, more fundamentally, serve to construct what and how much we know about breast cancer and in turn influence our actions toward it.
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
281
For instance, pharmaceutical research for the treatment of breast cancer has received much more funding than behavioral research for the cause, prevention, and management of breast cancer (Zones, 2000). This is the case, in part, because drug sales are directly profitable while insight into behavior is not. Society’s construction of meanings of breast cancer also influences which breast cancer organizations and associations emerge as the most prominent, in the context of policy making. Breast cancer policy, as Weissman (2000) describes, is not always or only in the best interest of women, but rather is the outcome of several entities ‘‘fighting’’ for their self interest. In simpler terms, breast cancer is political. For instance, recommendations for mammography screening, despite empirical evidence of the benefits have been, at best, vague – especially for older women. This is due, in part, because insurance companies realized that yearly screenings would not help reduce medical care costs. Medical care providers, on the other hand, while aware of the cost, were also aware of the benefits of detection. They also were aware that early detection may be at the cost of increased exposure to radiation. Women’s health advocates unequivocally supported screening and detection without concern for monetary cost or extra radiation risk. The struggle resulted in an indistinct recommendation that women (especially older women) ‘‘should decide for themselves’’. This recommendation was made amidst the knowledge that, generally, the older the woman, the more she relies on her physician to tell her what to do (Degner et al., 1997). If the physician does not remind her to have a screening, then she very well may never have one. If the physician recommends yearly screening, she may find herself having to pay out of pocket for some of the screenings. Overall, taken into account the range of knowledge and resources among all women, this recommendation is obscure and may actually diminish the faith in the efficacy, as well as the actual efficacy, of mammography especially for older women.
Popular Cultural Representations of Breast Cancer Popular culture represents society’s dominant and preferred ways of thinking about breast cancer. That is, young, successful, empowered survivors of breast cancer are usually featured in magazines marketed for the general public. Popular culture magazines are indeed a formidable producer of information and knowledge. Women’s magazines, in particular, help to shape both a woman’s view of herself, and society’s view of her (Ferguson, 1982). Ironically, though, it is actually common for popular
282
LISA COX HALL
culture magazines to NOT present a true and accurate picture of most women’s lives (Winship, 1987). One reason for this is the patriarchal culture in which women are mere objects of ‘‘the gaze’’. Gamman and Marchment (1989) elaborate on this notion by describing the implications for women as viewers of popular culture in The Female Gaze. Sociological analyses of women’s magazines have concluded that the popular culture sensationally frames breast cancer as affecting only a limited population: young, white, middle class, heterosexual women in their prime of their lives. On a micro-level, such biased representations are bound to misrepresent risk while leaving minority and/or diverse women without models for which to identify. To borrow a term from Tuchman et al. (1978), older women are ‘‘symbolically annihilated’’ in these publications. On a macro-level, such bias and misrepresentation impacts our society’s inquiries, actions, and policies regarding breast cancer (Zones, 2000; Weissman, 2000). The dominant values of Western culture are transmitted and perpetuated in popular culture, so, of course, it tends to resemble only the lives of the advantaged in society. First, the culture is oriented to individualism. Fosket et al. (2000) argue that popular culture magazines represent this individualism especially in the way it transmits, first and foremost, a message of personal responsibility. That is, women are portrayed as being responsible for ‘‘detecting, preventing and surviving breast cancer’’. This can lead to an ideology of blaming the victim. Second, as stated earlier, the culture is patriarchal. Clarke (1999) claims that the media illustrates such patriarchal dominance when it reports about breast cancer in stereotypically gendered ways. She found that women are described as self-absorbed, scared, emotional, and overly-body-conscious in most pop culture breast cancer features. Third, our culture is profit oriented. As a capitalist enterprise, the media publishes articles that appeal to their widest readership and is shocking enough to garner interest from other ages. The age demographics of the readers of most popular women’s magazines are early 20s to approximately 35–40 years old (Davis, 2001); therefore, stories tend to feature women of similar ages. We become used to seeing breast cancer associated with women in the prime of their lives, and therefore, more easily ignore the reality that it is older women who are at the highest risk of getting and dying from the disease (Kimmick & Balducci, 2000). Marketing and profit are not the sole reason for the misrepresentation of breast cancer in the popular press, though. Our culture is youth (and therefore beauty) oriented (Wolf, 1991) while it fears and holds disdain for aging and death (Cole, 1992). This is reflected in many breast cancer articles from popular women’s magazines. Lantz and Booth (1998) analyzed articles
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
283
from 27 popular culture magazines from 1987 to 1995. Though more than 80% of breast cancer diagnoses are in women over the age of 50, these pop culture magazines framed the disease as striking young women. Specifically, they found that 85% of the featured stories were of women under 50, and 80% of the photographs were of women under the age of 50. Of the underage-50 photographs, many were of attractive women – perhaps models – in their 20s. The majority of stories featuring women over the age of 50 were celebrities, rather than ‘‘ordinary’’ older women. Marino and Gerlach (1999) analyzed 46 popular culture magazine articles between the years 1987 and 1995 and found that 59% of the featured stories were about women younger than 50 years of age. Burke et al. (2001) analyzed 389 articles in U.S. magazines from 1993 to 1997 that had circulations of at least 500,000. In 84% of the vignettes, the women had been diagnosed with breast cancer before age 50. In 47% of the vignettes, they were diagnosed before age 40. This compares to age incidence rates of 16% and 3.6%, respectively. Only 2.3% of vignettes featured women in their 60s, and there were no vignettes of women in their 70s, despite that population data show that 65% of breast cancer cases are in women over age 60. The messages of these pop culture texts is clear: our culture is more interested, frightened, and concerned about young and middle-aged women suffering from breast cancer than older women. Furthermore, the magazines’ portrayal of breast cancer as striking mostly younger women has created, at least, a misconception, and at most, a myth, about whom is at risk for, who gets, and who dies from this disease. It is likely that this inadequate portrayal has dampened sociological interest in older women’s experiences with breast cancer and that it affects health care professionals’ perceptions about their patients. No studies, to my knowledge, have focused on how such representation throughout the life-course serves to construct variously aged women’s lived experiences of breast cancer, and their corresponding health care and treatment.
Medical Literature on Breast Cancer The disease, does not, by itself, determine breast cancer experience for women. Social aspects of the disease, such as interactions with the institution of medicine, including screening recommendations, physician– patient relationships, and treatment protocols do have an incredible impact. These social aspects, to a large extent, construct a woman’s experience and shape her identity. For instance, they can influence whether a woman sees
284
LISA COX HALL
herself as a victim, as a survivor, as disfigured, or as a warrior (Lorde, 1980). Research on the current state of breast cancer care for women of various ages and the influences and effects of that care indicate that differential, if not ageist, care exists. The following will discuss clinical research and screening, women’s knowledge and perceptions of breast cancer risk and effective treatment, women’s willingness to participate in treatment decisions, the tendencies and biases of physicians, the rates of women – by age – who undergo specific treatments, and differences in recommendations for middle-aged and older women. To begin with, medical literature reports that elderly women were usually left out of clinical breast cancer research in regards to screening, treatment, and psychosocial experiences. Once studies were conducted that did include elderly women, it was found that women 66–79 years of age do benefit from screening (Smith-Bindeman et al., 2000). At this point, researchers became interested in assessing how the small benefit compared to cost, and to potential radiation harm. Screening for women aged 69 and over results in a small gain in life expectancy and is moderately cost effective in those with high bone mineral density; therefore, it was recommended that older women make the screening decision for themselves (Kerlikowske et al., 1999). The American Geriatrics Society encourages women aged 65 and older to be screened every year; however, Medicare only covers the cost of a mammogram every other year (Du & Goodwin, 2001). It has been found that significant numbers of women are not knowledgeable about their risk for breast cancer, the screening recommendations for their age, or effectiveness of treatment strategies. Friedman et al. (1998) interviewed women in three age groups about their perception of risk. The three age groups were very similar in their perception of risk despite that risk does indeed vary between groups. It is likely, they concluded, that women are not being screened as often as they should be due to these misperceptions. Lazovich et al. (2000) found that older women were less likely than younger women to know that outcome was virtually the same for breast conservation therapy (lumpectomy) as for mastectomy. Possible explanations for lack of knowledge include (1) a significant number of women do not report the desire to independently seek information or make treatment decisions, and (2) physician characteristics are associated with certain practice philosophies that guide information dissemination, screening and treatment recommendations. Each is discussed later. In a study that investigated women’s desire to obtain information and make decisions about breast cancer, only 22% of middle-aged and older women wished to select their own treatments, while 44% wanted to
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
285
collaborate with their physician, and 34% preferred their physician to decide entirely (Degner et al., 1997). Petrisek et al. (1997) found that older women are less likely than younger women to wish to participate in the selection of their treatment for breast cancer. Older women did not seek out medical information as proactively as did younger women, nor did they report having considered the possibility of recurrence when selecting treatment. Women’s lack of knowledge and involvement in decision-making has been linked to their strong reliance on physicians. Literature has repeatedly discussed the subjective nature of medical practice. For instance, certain physician characteristics are associated with information dissemination, screening recommendations, and treatment options. Marwill et al. (1996) found that physicians are more likely to recommend mammograms for women aged 65–74 than for women aged 75–84, for women without dementia than for women with dementia, and for women living at home with their daughters than for women living in institutions. Silliman et al. (1999) found that ‘‘women with early stage breast cancer cared for by female surgeons are more likely to receive standard therapies’’ than are women cared for by male surgeons. Furthermore, lumpectomies are more likely to be the treatment of choice for women over 75 when the physician is female (Cyran et al., 2001). Such studies lead to the suspicion that there is institutional ageism within medicine in regard to breast cancer screening and treatment. Wanebo et al. (1997) strongly claim that breast cancer management is compromised for the elderly. Madan et al. (2001) found that residents have biases against older women in their rates of recommendation for mastectomy and reconstruction. In their study, modified radical mastectomy was recommended to 38% of women aged 59 and older, compared to 11% of women aged 31 and younger. Reconstruction was recommended only 70% of the time for older women, while it was recommended 96% of the time to their younger counterparts. Cyran et al. (2001) found that women aged 65 and over undergo lumpectomy less often than younger women. Similarly, use of chemotherapy declines with age. Du and Goodwin (2001) found that younger women used chemotherapy treatment more than older women. While Guadagnoli et al. (1997) claim that use of such treatments merely reflect available information about treatment efficacy; Lash and Silliman (2001) concluded that older women (aged 75–90) are least likely to receive guideline surveillance in all aspects of breast cancer care, screening, diagnosis, prognostic evaluation and treatment. (p. 945)
286
LISA COX HALL
In sum, studies indicate that older women do not receive definitive care as younger women do. In addition to ageism in medical practice, other studies suggest that there is ageism in health care financing, as well. Feldstein et al. (2001) found Medicare patients to be at higher risk for poor outcomes than for patients with HMOs and fee-for-service systems. As was alluded to earlier, cost and quality are often pitted against each other when it comes to health and health care. If one believes that all women, regardless of age, should be treated equally, then we see ageism occurring. If, however, one believes in a world of limited resources, care should be rationed and limits should be set (Callahan, 1987) then the reality is that the aged do not get the care and resources. To summarize, over one million women each year battle breast cancer in this nation. The disease is shaped by cultural, social, and political forces. The institution that treats breast cancer – medicine – is also influenced by cultural, social, and political forces. Both popular culture and medical culture, it has been found, are age biased when dealing with breast cancer. On the basis of the high prevalence of this disease and the misrepresentation and mistreatment of age by popular and medical cultures, it is important to investigate whether women of various ages do indeed experience the disease and its treatment differently, and if so, in what ways.
METHODS I specifically chose the qualitative method of in-depth interviewing because it allows unbiased discovery, and avoids a priori assumptions. Qualitative interviews ask respondents to explain their perspectives and experiences using their own words and categories, as opposed to structured surveys that impose predetermined variables and response categories, which may not always capture the aspects of experience respondents consider most important (Sudman & Bradburn, 1982) and may limit discovery and data analysis to already recognized analytic concepts. According to Gubrium and Holstein (1997), a qualitative approach offers an in-depth exploration of taken-for-granted processes in everyday life, close examination of social action and the creation of meaning, sensitivity to the processes of interaction, and ‘‘tolerance of complexity’’. Standard intensive interviewing methodology (Lofland & Lofland, 1995) was followed in this study. Specifically, each interview was planned to be face-to-face and to last from 1 to 2 hours in duration, at one point in time.
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
287
Part of that standardized procedure involved the use of a loosely structured interview guide, which was used with each participant in order to ensure that similar questions were asked in similar ways in order to elicit similar categories of information. The interview guide acted as a series of prompts, rather than as a narrowing, or controlling tool. The guide was comprised of a few broad questions in order to allow participants to start where they were, rather than with my assumptions (Lofland & Lofland, 1995). To ensure that this study could address possible age variation, the participants were selected with attention to age. Seven women from each age decade (30s, 40s, 50s, 60s, 70s, and 80s) were recruited for a total of 42 participants. Previous studies have varied widely in how they have addressed age in breast cancer research. Many studies with attention to age have either been limited to a single broad age group such as 65+ or have used two or three basic groups (i.e. younger, middle, older) with varying cut points for these distinctions (Baider et al., 2003; Cimprich et al., 2002; Jones et al., 2003; Woodard et al., 2003; Watters et al., 2003). Because age is the central concern of this study, I opted for a more detailed age grading of the interview population. I extended the four categories used in studies by DeMichele et al. (2003) and Matthew et al. (2004) to six, thus enabling an even coverage of the age spectrum from 30 to 88. The eligibility parameters included the following criteria: FIRST breast cancer diagnosis, stage II breast cancer diagnosis, receipt of chemotherapy treatment, and currently in treatment for the FIRST breast cancer diagnosis.1 The participants were recruited from two sites: a private oncology office in the state of Colorado and a radiation oncology department at a teaching hospital in Kansas. See Table 1 for a summary of participant characteristics. None of the names used in this written report are real names. The names of interview participants, doctors, and friends and family members (when mentioned by the interviewees in their narratives) have been altered to protect said identities and to ensure confidentiality of everyone involved. All interviews were transcribed by only three transcriptionists in order to ensure consistency and minimize the impurity that can come from transforming the format of stories. All interviews were also transcribed verbatim in order to maintain the integrity of the data and ensure that nuances, such as pauses, crying, laughing, and sighing, were retained. The transcriptionists were made aware of the absolute necessity of confidentiality. The data were read and analyzed according to the constant comparative method, so that trends, similarities, and differences could be identified
288
LISA COX HALL
Table 1. Characteristics Race African American White Religious affiliation Jewish Catholic Protestant Non-Denom./New Age None Missing Educational level 8th grade 10th grade 12th grade Some college BS/BA MA Doctorate Missing Relationship status Widowed Married Partner Divorced Engaged Single
Characteristics of Participants. N
%
5 37
12 88
2 3 23 6 7 1
5 7 55 14 17 2
1 3 10 13 7 6 1 1
2 7 24 31 17 14 2 2
7 22 3 7 2 1
17 52 7 17 5 2
among the cases and between the age groups. The 1,000 plus pages of interview data were analyzed using both standard, manual techniques and qualitative data analysis software. Organization and coding of the data and the formation of preliminary conclusions took place concurrently and were ongoing throughout the analysis stage.
FINDINGS I found, in general, that younger women displayed more agency than older women in health knowledge and practice and in treatment planning. Agency refers to a state of being capable of movement or representation. A contemporary display of agency includes being aggressive, in control,
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
289
responsible, direct, and open. When a woman has agency in matters of health and healthcare, she can act on her own behalf in her best interests. Specifically, younger women in this study more often found their own cancers, and more regularly had routine mammograms (except for those under 40, for whom mammograms are not recommended), and more thoroughly discussed and better understood mammography effectiveness and biopsy details than the older women. Younger women were offered more options by physicians than older women were. Finally, younger women were more willing to have multiple surgeries and procedures (such as incremental removal of breast tissue and reconstruction) than older women were. Each is discussed later.
Variation in Health Knowledge and Practice I found in this study that younger women discover their cancers, themselves, more often than older women do. Twenty-three of 42 women (55%) in this study reported noticing something unusual about their breast, on their own, before having a breast physical exam by a physician or before having a mammogram.2 Fifteen of those 23 women (65%) were under the age of 49. In the majority of older women’s cases in this research, a mammogram found the cancers. In the few cases where an older woman told me she found the lump, herself, she did not describe how or why she found it herself, nor did she attempt to explain the reasons for her behavior, as younger women did. One reason for this age significant difference may be that younger women are ‘‘taskmasters’’, as Jill, 32, suggested. ‘‘We are very knowledgeable. We try to find all the information. We try to research everything we can. Often times we are a resource pool for each other.’’ She voiced a principle that was begun long ago by the Women’s Health Movement. ‘‘The best thing you can do is be your best health care advocate.’’ Another reason for this age-related difference may be due to the disease features of breast cancer and corresponding medical practice (i.e. screening recommendations), which do differ for women of different ages. Specifically, mammograms are not recommended for women under the age of 40;3 therefore, most women in their 30s are not having mammograms, and if a lump is to be detected, it would first be detected through palpation, and then it could be detected visually through an ultrasound. On the basis of odds alone, a woman is more likely to feel a lump than her doctor is to feel one, as she may see her doctor less than once a year. That young women find their
290
LISA COX HALL
cancers, themselves, more often than older women, then, may be because older women rely on mammograms because they believe them to be sufficient and effective. By the same token, it may be that the youngest women know they, themselves, are their only source of detection and are, therefore, more proactive. I found that women in the younger age group are more knowledgeable about the strengths and weaknesses of various breast cancer screening procedures than women in the older age group are. There is variation in knowledge regarding the accuracy of mammograms and there is variation in the amount of faith that women have in the effectiveness of mammograms alone. There has been, in the research literature, an active and public debate about the efficacy of mammograms (Love, 2005). Such debate has also been featured in some popular culture magazines, as well. Several women in this study, however, reported not having been aware that there were limitations of mammography. The women in the youngest and middle age groups said they were especially surprised about their lack of awareness and voiced concern that not enough information is available to educate women. Brooke, age 37, recounted how mammograms did not detect her cancer due to her younger breast tissue being a similar density as the cancer. She explained ‘‘In all [my] mammograms y you couldn’t see anything because my breast [tissue] was light and the cancer was light, and I even had the DCIS,4 which should have [appeared as] little particles floating around. You can’t see any of it.’’ Meisha, age 30, raised another limitation of mammography, which is its inability to find cancers outside the main breast area. ‘‘My relative’s cancer showed up in her chest wall, so the mammogram was not very helpful’’. Lindsey, age 47, experienced a different limitation, yet: ‘‘my lump was actually very small but the cancer fingered out. And those are called lobular features. And those are extremely difficult to see on a mammogram.’’ Contrary to the younger women in the study, the older women would simply report that cancer showed up on their mammogram. They did not discuss or criticize the details about mammography. Only one of the 14 oldest women (Francis, age 71) had to have a follow-up magnetic resonance imaging (MRI) to detect her cancer. This extra knowledge of young women may be due not only to increased agency or intelligence, but also to features of the disease and to medical practice. That is, because mammograms are least effective on younger women, additional methods must be used, such as ultrasound or MRI. During the course of multiple screening techniques, younger women end up
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
291
learning more about the benefits and limitations of various screening mechanisms. I found that there are age differences in mammography practices and in the amount of discussion of those practices. Women of older ages are less likely to discuss their mammography practices in detail than women in the middle age groups are; and women of older ages are less likely to have had baseline mammograms than women in the middle age groups. Mammogram screening practices have been routinely studied by medical and social science researchers, alike. Mammography screening has increased from approximately 24% of women in 1987 to 70% in 2000 (www.cdc.gov). Consistently, it has been found that women who do not screen have lower incomes and/or no insurance coverage. Furthermore, white women have mammograms more than women of other ethnicities and races. Women aged 50–64 have the highest screening rates, at 78.6%, while women aged 65–74 have the next highest screening rates, at 74%. Women aged 40–49 screen at a rate of 64.2%, which is higher than the 75+ age group, which screens at 61.3% (NCHS, Health, U.S., 2004). Women in their 40s in this study had regular mammograms at an even higher rate than the national average. Only two of the seven women in their 40s had higher risk factors for breast cancer, so this does not significantly account for the study group’s higher rate of screening. Women 50+ in this study had regular mammograms at a slightly lower rate than the national average. Four women in the study (three in their 80s, one in her 60s) gave no information about whether and when they had mammograms prior to their diagnoses. The women in this study had different habits for keeping regular gynecological appointments, and having annual mammograms. Some women, with family histories of breast cancer or with frequent benign fibrocystic lumps were having screenings at least every six months. Two of the women in their 80s, had never had a mammogram. Ella, age 82, having lived in a small town in Indiana her whole life, never inquired about nor was ever ordered to have a mammogram. It was not until she found a lump, and came to Colorado at her son’s request, that she had her first mammogram at the age of 82. ‘‘I’d never had a mammogram. Well, I knew that I should have been getting mammograms. But our doctor, he didn’t put his self in, in, ah y I guess he took it for granted we should know to go and get them.’’ She later added ‘‘if I’d have went – went ahead and had my mammogram, but then it might not have made any difference but I – I could’ve felt better about it maybe.’’ Grace, 80, had never had a mammogram, either. It was only at the persuasion of a young doctor, after she had suffered a stroke that she agreed to one.
292
LISA COX HALL
Most women in this study had been having routine mammograms before their cancers were diagnosed. Lindsey, age 47, had a baseline at 35, and then had yearly mammograms after age 40. Leslie, 47, had consistent screening practices: ‘‘I started doing my mammograms the minute I turned 40, and always did them in June.’’ Hazel, 56, was just as tenacious: ‘‘I do my yearly, every year. I’m really careful about that.’’ Only two of the oldest women reported being vigilant about routine screenings. They also happened to be highly educated. Bertram, 80, however, had to request and even demand that her physician perform a proper breast examination during one of her well-woman checks. ‘‘I had a doctor that gave me a physical – what he called a physical breast exam – it wasn’t what I call a physical breast exam, so I had to ask him – I had to ASK him for an exam!’’ At least three women were considered high risk and were being watched closely for cancer: Mary, 48; Patty, 56; and Tamar, 63. These women had already experienced the gamut of procedures: doctor exams, mammograms, ultrasounds, biopsies, and lump removals. Two women in their 70s, due to prior diagnoses of breast cancer were likely being monitored, though they did not directly say so: Lynn, 71; and Mutzia, 75. Five women in the study had had mammograms, but not routinely, and they were spread across the age groups: Emily, 52; Joy, 56; Irene, 63; Nancy, 62; Elsie; 66; and Dorothy, 70. Though there was variability in behavior across the age groups, I would argue that the middle-aged women were more purposeful and certainly more talkative about their habits. I found that women in the younger and middle age groups are more likely to discuss the details of their biopsies than women in the older age groups. If women discussed biopsy, it was in response to my first general question (‘‘Can you please tell me when you first noticed something unusual about your breast?’’) because I did not specifically ask them about their biopsies. Upon finding a lump or suspicious sign about the breast, and upon scheduling an appointment with a Primary Care Physician, or Gynecologist, or in high-risk cases, an Oncologist, or in cases of referral, a surgeon, the physician usually palpates the woman’s breast. If the physician agrees that there is suspicion, then a mammogram, ultrasound and/or biopsy is ordered. There are four common ways to biopsy the suspicious tissue: fine needle aspiration, core needle biopsy, surgical incisional biopsy, and surgical excisional biopsy. The first two use local anesthesia and the last two use general anesthesia. The women in this study varied significantly in (1) whether they discussed the biopsy stage, and (2) the amount of details they reported about the biopsy. Six women (14%) did not, throughout the entire interview, ever
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
293
discuss having a biopsy. Twenty-one women (50%) made a general mention of having a biopsy. Nine women (22%), while mentioning their biopsy experience, distinguished between biopsy types. Six women (14%) discussed the details of their biopsies, and appeared to have accurate knowledge of biopsy, in general. There are differences in age regarding discussion of and knowledge about biopsy. Women in their 80s either did not discuss biopsy at all (2) or made only a brief mention of biopsy (5). These women had either little or no knowledge about biopsy, or they chose not to discuss it. Women in their 70s were slightly more open to discuss, or more knowledgeable about biopsy: only one woman made no mention; five women made a brief mention; and one woman, Francis (age 71) discussed her biopsy in detail, knowledgably. Women in their 60s were most variable of all the age groups in discussion and knowledge of biopsy. Only one woman made no mention; three made brief mention; two acknowledged types of biopsies; and one (Nora, age 61) discussed her biopsy in detail, knowledgeably. Women in their 50s, overall, had more knowledge and spoke in more detail than any other group. One woman made no mention of biopsy; only two made brief mention; two distinguished between biopsy type; and two women (Patty, age 56; Emily, age 52) spoke in great detail and with knowledge about biopsy. The women in their 40s were next most knowledgeable and open to discuss biopsy. This is the only group in which everyone at least mentioned biopsy (4). Two women described types of biopsy; and one woman (Lou, age 45) spoke with knowledgeable detail about her biopsy. The women in their 30s were equally knowledgeable and willing to discuss biopsy. One made no mention; two made brief mention; three displayed knowledge of biopsy types; and one (Jill, age 32) spoke at length and with considerable knowledge about her biopsy. In sum, the women in their 30–60s were similar in whether and how they discussed the biopsy part of their experience. They were different from, however, the women in their 70s and 80s. The women in their 80s discussed biopsy much less, and demonstrated little or no knowledge about the procedures, as illustrated by Connie, 83, when she recalled the events, ‘‘I honestly can’t tell you how it proceeded y guess I had a biopsy’’. Once again, younger women may talk more about health knowledge and may be more vigilant about health practices because they are new to the world of medicine and illness. It may be that older women have experienced health procedures and dilemmas for a longer amount of time, and so they just do not talk about them, anymore. (Compare this to a woman who gives birth for the first time and tells everyone every detail of the labor and
294
LISA COX HALL
delivery; however, when she gives birth the second or third time, she tells much less about the experience to even fewer people.) Also, once again, these age significant findings may be a result of the effects of the Women’s Health Movement and the Consumer Movement manifesting more strongly on the youngest women. They, therefore, voice and display much more self-responsibility, knowledge, and agency in their health behaviors.
Variation in Treatment Planning Younger women are more willing to have multiple surgeries and procedures (such as incremental removal of breast tissue and reconstruction) than older women are. Seventy-five percent of women under age 48 in this study had reconstruction. Only one woman over age 49 had reconstruction. Despite that the oldest women usually spent the shortest time in treatment for breast cancer, because they did not receive chemotherapy and radiation as often as women younger than them, they still did not want to have additional surgeries, after the mastectomy. The youngest women in the study, often spent the longest time in treatment, and they were willing to spend even more time having reconstructive procedures. Jill had the TRAMflap;5 Meisha, Karen, Lindsey, and Dorothy (age 70) had the tissue expanders and implants.6 Leslie was having a procedure where fat tissue from the buttocks is transplanted into the breast. She was waiting until all treatment was completed before having the reconstructive surgery. It required her to go to Louisiana or to California. Mary was also waiting to have a reconstructive procedure, but she did not specify which one. Because reconstruction is done in stages, it requires several appointments and multiple surgeries over several months. Meisha, 30, had her tissue expanders for several weeks and had just had her permanent implants placed, the day before I interviewed her. She explained: ‘‘The second surgery, here, I’m in pain. The surgery went well, but I’m in pain and I still have two more procedures to go. They have to do the nipple reconstruction and then the tattooing, within the next four months.’’ She was aware that that would not be the end of surgeries, though. ‘‘It feels like an ongoing battle that never ends. The implants only last 10–15 years and you have to have them replaced anyway. So, I’ll have to put up with that for the rest of my life and y [I’m] only 30 years old.’’ Lindsey, 47, said ‘‘I saw myself, you know, flat as a board after the bilateral mastectomy. They told me y the tissue might not expand properly after
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
295
radiation.’’ It took about six weeks of periodically injecting saline into the implant in order to get close to the size Lindsey wanted. ‘‘By the time I got to mid February I had my breasts back, actually I have a little more than what I had before. And so I was ready to start radiation but we had to do it quickly because I wanted to get to radiation soon.’’ Lindsey would have to wait for the tissue to then heal from radiation before continuing with the reconstruction process. ‘‘I’m going to have a little uhm, revision because this tissue expanded lower on this side than it did on this side. They started out even but then, uhm, this one stretched more than the other. So I’m going to have a little revision to uhm, to move this one down so it will match the other.’’ It would then be another three months or more before she would have her permanent implants placed and then another four months to have the nipples made. ‘‘Hopefully I will be able to have the nipple reconstruction, the areola, soon, but [the surgeon] said you don’t want to do that too soon after radiation or you don’t get a good result.’’ Lindsey was dealing with discomfort, too, when lying down at night. The real discomfort, she said, came right after the expansion and would last about two weeks. It would ease up for a week or so, then it was time to get expanded again. It was also painful to have the expanders during radiation. ‘‘I would be lying on that radiation table and I’d think, ‘oh dear Lord’, because it hurt, you know, it was just searing through my back.’’ Jill did not have tissue expanders because she opted for the immediate TRAMflap. She did not have radiation, either; therefore, she did not have to delay nipple reconstruction in order for her skin to heal. She talked about the extra abdominal scars that she incurred for the reconstruction of breasts, though. ‘‘Under all this it’s like I’m a surgical roadmap, you know, I got, you know, I’ve got cuts, which will heal and that’s fine. Uhm, but I wanted it all done.’’ She acknowledged the ongoing steps of the procedure: ‘‘I’m not done with surgeries. I’ll have to go in and do the nipple reconstruction and do the areola complex, which means I’ll have two different tattoos.’’ Mary, 48, did not know she wanted reconstruction immediately after her mastectomies, like Jill did, and it was only through complaining to her husband and having him suggest that she do reconstruction that she seriously considered it. ‘‘I went to the gynecologist to get a second opinion y and she said, [Mary], I think you want it.’’ Mary was a graduate student and had to schedule reconstruction into her school calendar. ‘‘I think this summer I’ll do reconstruction. And reconstruction isn’t, the plastic surgeon explained, isn’t just like one thing. It can be a series of procedures.’’ Mary said she was willing to go through it.
296
LISA COX HALL
Only one woman over age 49 decided to have reconstruction: Dorothy, age 70. Her husband had died a few years before she got breast cancer and she was currently engaged to a high school sweetheart, whose first wife died from breast cancer ten years earlier. ‘‘I knew what he had been through – that his wife had had a radical, and I guess terrible scaring and things like that. I’m, you know, I’ve got a mentality that, you know, I’m not gonna go that way (laughing)!’’ Her fear of the stories about the scars, and her concern about her fiance drove her to ask about reconstruction. She also said that, to her, mastectomy meant being ‘‘disfigured’’. She was in the middle of having her tissue expanded when we talked. She pointed to the area and said ‘‘I look a little weird. This has been filled with saline, I guess, or something. I was all for it. I could have waited, but I didn’t want to wait.’’ It is significant to note, here, that Dorothy’s circumstance of being engaged may be driving her decision more than age. She has more in common with younger women because she is experiencing the life-course norm (i.e. engagement) that is associated with younger ages. Lynn, 71, however, is an older woman with typical life-course norms who wishes she would have had reconstruction. She blames herself for being too shy to express herself to her physician when he made an assumption and said ‘‘at your age, you probably don’t want reconstruction’’. Joy, 56, did wait to decide, taking one step of treatment at a time. Then, she said ‘‘I made the decision not to do reconstruction y At the age I am now, you kind of care, but I thought, ‘I’ve had two surgeries, so’ I thought ‘I’m not doing this again’.’’ Not wanting more surgeries and not wanting to drag out the management of the disease and its effects was a common theme for many of the older half of the women in the study. Connie, 83, said ‘‘I did not [want reconstruction]. I was not interested in that. No more surgery! I’d had enough.’’ Beverly, 70, also shared this opinion: ‘‘all I could think of was I didn’t want any more surgeries than this one, and so that’s why I didn’t want reconstruction or anything y I was 65 and ah, I just, one surgery seemed like enough.’’ Grace, 80, agreed: ‘‘I just don’t think it is necessary. I don’t think it is an option for me. I mean, I don’t feel that I would need that [extra surgery].’’ Some of the other older women (Esther, 88, Bertram, 80, Cass, 70, and Ivy, 72) do not wear their prostheses. In sum, they claim there is no reason to go through extra effort, especially due to their experience that prostheses are uncomfortable. This is a contrast to the younger women who are willing to go through a lot of time, pain and effort for a variety of reasons. While age is not the only determinant of reconstruction choice, age is correlated with many factors that encourage younger women to have
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
297
reconstruction and older women not to. As reviewed, physicians and surgeons have tended to offer reconstruction to younger women more readily than to older women. Having a relationship with an intimate partner is a correlate that increases a woman’s likelihood of having reconstruction. It is more normative for the oldest women’s partners to be deceased; therefore, they are less likely to have reconstruction. Familiarity with technology, in general, and with medical technology and elective plastic surgery, specifically, are correlates that increase a woman’s likelihood of having reconstruction, as well. Older cohorts of women are least comfortable with the notion of elective surgery, while younger cohorts of women tend to take it for granted. Three decades ago ‘‘elective surgery’’ was little more than an oxymoron. The older women in this study admitted that voluntary, unnecessary surgery was hard to comprehend. Today, ‘‘elective surgery’’ is considered common practice. The media is responsible for this in two ways. Media inundate us with images of beautiful, flawless, thin people, leading us to believe we can, or should, look that way, too. Simultaneously, media reveal the thousands of celebrities who have had plastic surgery, and then advertise how easy it can be for us to have it, too. These constant messages result in particularly the younger women thinking that reconstructive surgery after mastectomy (as opposed to breast augmentation sans any disease) seems less elective and more necessary. Clearly, some women in this study said that cancer, mastectomy, and reconstructive surgery provided the opportunity to have the breast surgery or tummy tuck they had always wanted but were unable to justify having.
DISCUSSION AND CONCLUSION This research provides needed information on women with breast cancer over the age of 65. In general, it suggests that older women display less agency, as it is contemporarily defined, than younger women in health knowledge and treatment planning. More specifically, it shows that older women talk less and give fewer details about their health and health care experiences and that their treatment planning is less complex than younger women’s. In the midst of the consumer culture and the major changes in the organization and financing of health care, to be proactive and to have agency, today, means to be aggressive, direct, and not private or too modest. Younger women in this study more quickly volunteered to be interviewed than older women. Younger women were also more likely to be aggressive
298
LISA COX HALL
and direct, while older women were more private and modest. Younger women talked more and in more detail than older women about their experiences. One should not, however, necessarily conclude that younger women are more resilient, optimistic, determined and in control than older women; only that younger women seem to talk more about these qualities. I would, therefore, argue that older women do not lack agency, altogether. This research indicates that older women have strengths that, though subtle and private, equip them with mastery and control, thereby enabling them to cope with breast cancer, perhaps even better than younger women. For instance, older women in this study complained less often about illness, pain, and health care providers; they exhibited grace amidst loss; and they expressed thankfulness for each day they were alive. Perhaps privacy and modesty is a form of control, and when they declined to be interviewed, they were expressing a version of agency. It is possible that our culture of individualism and self-responsibility, however, preclude us from recognizing and working with older women’s forms of agency. Since these forms of agency are not likely to garner attention and rally change, one possible consequence is that older women’s perspectives are not recognized. The result may be that they miss out on the most definitive care possible. Due to the rising numbers of elderly women with breast cancer, though, it is important to better understand older women and their disease experience and to strive to provide them with better care. To better understand older women and their plight and preferences regarding breast cancer and treatment, more psychosocial studies are necessary. This study indicates that recruiting older women can be challenging for a variety of reasons. Many older women are dependent upon their adult children for transportation, for example. Many older women in this study simply said they did not want to talk about their cancers and, therefore, declined participation. There simply are fewer numbers of women over the age of 70 in the population from which to recruit. For these reasons, researchers need to be prepared to recruit older participants from many sites and to allow plenty of time to do so. They also need to be prepared to accommodate older women with issues of transportation, interview location preference, and duration of interview, to name a few. Devices to aid elderly participants in recalling dates, order of events and basic medical information, without overly controlling their stories, should be considered, as well. Such aid may provoke more discussion and yield richer understanding. Steps toward providing better care for older women with breast cancer begins with making health care providers more aware of the subtle ways in
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
299
which our ageist culture affects patient–physician interaction and breast cancer treatment. Older women appear, to physicians, to be less concerned than younger women about the loss of a breast. It is true that some women, of various ages, are more comfortable without reconstruction or prosthesis. However, these data show that sometimes older women are concerned about losing their breast. Furthermore, this analysis suggests that older women are taught by an ageist culture to regard their breasts as unimportant. Even if they lament the loss of their breast, they learn to say it does not matter. Popular culture magazines have consistently relayed the message that older women are not desirable to look at or to read about. Our youth oriented society decrees through jokes, television, and print media that older women are sexually revolting. We receive such messages all along the life-course, and by the time women are older, they have internalized these perceptions and may, therefore, be less apt to express that their femininity or sexuality matters. Recall how Lynn, 71, abandoned reconstruction after her physician’s ageist remark. She aptly represents all older women who cower under demeaning views. Lynn’s sense of self and the worth of her body have been shaped by a culture that avoids or depreciates older women. Physicians are affected by negative cultural messages, as well, and some therefore may be less ready to discuss treatment and reconstruction options as well as how an older woman will feel about losing a breast. Since many older women will not broach the subject on her own behalf, it may be up to health care professionals to initiate such conversation. Wanzel et al. (2002) found, unfortunately, that over 30% of surgeons, oncologists, and primary care physicians do not offer to refer their patients for breast reconstruction because they felt they did not have adequate knowledge of the procedure and believed that reconstruction could interfere with cancer treatment and detection. There is no scientific evidence, however, that support these beliefs. When Alderman et al. (2003) found that women aged 55–64 were least likely to have reconstruction (they did not even include women over 65 in their study) they presumed that ‘‘older women may have different priorities and different perceptions of mortality’’ (p. 700). The evidence, however, is that physicians are less likely to offer reconstruction to patients over age 49. This may better explain older women’s low rates of reconstruction. Again, the presumption was that physicians think older patients do not consider a reconstructed breast to be important (Wanzel et al., 2000). Physicians also stated that older patients were thought to be ‘‘unsuitable candidates’’, which likely refers to several variables, but undeniably involves the view that the benefits of a new breast (i.e. self esteem, convenience, esthetics, maintenance
300
LISA COX HALL
of sexuality) do not outweigh the costs (medical resources, risk of infection, health coverage arrangements) for older women who have fewer years ahead of them than younger women. It is, therefore, this particular view of some physicians, which result from our ageist society that may best explain older women’s low rates of reconstruction. If older women are defined as ‘‘unsuitable’’ for reconstruction, they may very well feel unsuitable and therefore may avoid requesting information or discussing reconstruction options. General disinterest may be older women’s attempts to respond the way society expects them to respond. In order to genuinely care for older women, then, health care professionals should not mistake older women’s reluctance or silence as lack of concern. All options should be given to older women, just as they are sympathetically given to younger and middle-aged women. Just as reluctance and silence should not impede the presentation of options and information, neither should cost or disease features, alone, determine medical practice; especially if it means that medical practice will be inferior for older women. Full disclosure regarding screening efficacy, detailed explanation of biopsy results, and complete offers of all available treatments should be given to all women, no matter her age or apparent knowledge – because this study suggests that when it comes to older women, appearances may be deceiving! A physician may see an ‘‘old lady’’ with breast cancer who has not discovered her own cancer, asks few questions, respectfully takes advice, offers little resistance, and insists on few explanations. This is the stereotype, and was the norm, for women born at a time when women were submissive and less educated, and when medicine was unsophisticated and the practice of it was arcane (Starr, 1982). The analysis from these data, however, suggests that there may be more about older women than meets the eye. The most important point for health care professionals to be aware of is this: for several reasons, many older women will not speak openly about breasts and breast cancer, nor will they broach the subject of reconstruction on their own; therefore, it is up to health care professionals to initiate such conversation, provide all options and information, and then follow up with patience and care. Silence should not be mistaken for understanding, agreement, and satisfaction. On the contrary, this research reveals that older women comply not necessarily out of complete understanding of the recommendations or out of total faith in doctors; rather, they tend to opt for the quickest response. They want to get the operation or treatment over as soon as possible. What this means is that an older women is likely to choose a radical mastectomy tomorrow – before the
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
301
surgeon goes on vacation for two weeks, even though she may not be completely informed of her options, risks, and prognosis – rather than wait. In the spirit of full disclosure and full consent, then, health care providers should perhaps offer all options, and discuss all risks and implications for prognosis to older women BEFORE they tell them the surgery timetable. In addition, because older women are unfamiliar with the notion of ‘‘elective procedures’’ and are therefore reluctant to consider reconstructive surgery, health care professionals may consider presenting this option as feasible, manageable and increasingly more common. It is important to interject, here, that some physicians and surgeons may be reluctant to introduce breast reconstruction because, currently, the reconstructive options are not tailored to older women’s bodies. That is, a reconstructed breast will be much more firm than her real breast. This usually results in asymmetry. If medical suppliers would prioritize the creation of an implant or other product that could more closely resemble an aged woman’s relaxed, less dense breast, then older women could have one surgery on only one breast, instead of both. Currently, the only way to achieve symmetry is to have the other breast operated on, which increases the surgery site and may necessitate additional surgeries, as well.7 A less dense implant or similar product, then, perhaps, would result in physicians and surgeons offering older women reconstruction more often and may increase the number of older women accepting it. In conclusion, in the challenging, busy world of medicine a quiet and compliant patient is a pleasant relief and an ‘‘easy case’’ for a physician. After all, a patient who listens, follows orders, does not complain and is thankful is more rewarding for a physician than a patient who asks a million questions, demands certain tests and treatments, and criticizes. In the cases of older women and breast cancer, though, it may be that they are quiet and compliant not due to total satisfaction but because of a stifling culture that defines older women who have questions, concerns, opinions, and needs as ‘‘crabby’’ (Esther, 88). Their compliance may come from years of hearing that old women are worth less than young women and that old breasts are not as important as young breasts. For the many health care professionals who truly want to provide genuine, tailored care for all their patients, especially the older ones, they will have to take note of the pervading ageist culture. They will have to try to go above and beyond the ‘‘easy, pleasant case’’ to ensure that older women feel worthwhile and legitimate, and to see that they get the treatment and care they truly desire.
302
LISA COX HALL
NOTES 1. Due to challenges in recruiting older participants not all older participants received chemotherapy and some were in treatment for a second breast cancer diagnosis at the time of the interview. 2. Mammography began in the 1950s. Mammo means breast and gram means picture. A mammogram is an X-ray of the soft tissue in the breast. The breast is pressed between metal plates, in a variety of poses or views. While awkward and perhaps embarrassing, it does not always hurt. It does change the round shaped breast into a flattened shape, though. The X-ray, of course, is a two-dimensional view, the images of which appear in shades of gray. When the cancer can be seen, a lesion as small as .5 cm (or .2 in.) can be detected. Lesions normally have to be 1 cm (or .4 in.) before they can be palpated (felt by the touch). Recommendations for screening have varied and are still wrought with myths and political pressure and are not based solely on a biological understanding of breast cancer or on mammography’s effectiveness and affects (Love, 2005). 3. The reasons mammograms are not recommended for women under 40 are two-fold. First, women under 40 are at less risk for the disease, and second, mammograms cannot detect cancer in dense breast tissue as easily as in fatty tissue. Young women have more breast tissue, while older women have more fatty tissue. 4. Ductal Carcinoma In Situ (DCIS) is an early stage cancer, sometimes referred to as a precancer, and is non-invasive, meaning it stays in one place and does not break through normal tissue (www.breastcancer.org). 5. The TRAMflap is a reconstructive procedure whereby fat and muscle from the abdomen and/or the buttocks in transferred to the breast area. This occurs immediately after the mastectomy but requires several weeks of healing and may limit a woman’s arm movement and lifting capacity. 6. These mechanisms are placed under the chest skin after a mastectomy and are enlarged periodically over a four to eight week period. They literally stretch the chest skin to make room for a full breast sized implant. 7. As is, a federal law entitled The Women’s Health and Cancer Rights Act (WHCRA) of 1998, requires insurance companies and health maintenance organizations, which cover the cost of mastectomy, to also cover augmentation or reduction mammoplasty of the other breast in order to achieve symmetry. This law may not apply to church or government health plans, and though it is a federal law, only 36 states plus Washington DC mandate the benefits (Kansas does; Colorado does not) (www.statehealthfacts.org). Insurance providers who do cover surgery on the other breast are required to disclose these benefits at time of enrollment and yearly, thereafter. WHCRA does not apply to Medicare (www.cancer.org). Medicare covers breast reconstruction when a woman has had a mastectomy due to breast cancer, and Medicare may pay for reconstruction of the contralateral breast (www.cms.hhs.gov). As always, in network provider requirements, pre-authorization requirements, pre-existing condition limitations, and waiting periods may apply, which may result in no coverage or delayed coverage.
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
303
REFERENCES Alberg, A. J., & Singh, S. (2001). Epidemiology of breast cancer in older women: Implications for future healthcare. Drugs Aging, 18(10), 761–772. Alderman, A. K., McMahon, L., Jr.., & Wilkins, E. G. (2003). The national utilization of immediate and early delayed breast reconstruction and the effects of sociodemographic factors. Plastic and Reconstructive Surgery, 111(2), 695–703. Baider, L., Andritsch, E., Uziely, B., Goldzweig, G., Ever-Hadani, P., Hofman, G., Krenn, G., & Samonigg, H. (2003). Effects of age on coping and psychological distress in women diagnosed with breast cancer: Review of literature and analysis of two different geographical settings. Critical Reviews in Oncology/Hematology, 46(1), 5–16. Burke, W., Olsen, A. H., Pinksy, L. E., Reynolds, S. E., & Press, N. A. (2001). Misleading presentation of breast cancer in popular magazines. Effective Clinical Practice, 4(2), 58–64. Callahan, D. (1987). Setting limits: Medical goals in an aging society. Washington DC: Georgetown University Press. Cimprich, B., Ronis, D. L., & Martinez-Ramos, G. (2002). Age at diagnosis and quality of life in breast cancer survivors. Cancer Practice, 10(2), 85–93. Clarke, J. N. (1999). Breast cancer in mass circulating magazines in the U.S.A. and Canada, 1974–1995. Women’s Health, 28(4), 113–130. Cole, T. (1992). The journey of life: A cultural history of aging in America. New York: Cambridge University Press. Cyran, E. M., Crane, L. A., & Palmer, L. (2001). Physician sex and other factors associated with type of breast cancer surgery in older women. Archives of Surgery, 136(2), 185–191. Davis, J. (2001). Women’s magazines misrepresent breast cancer risk. Most stories overstate risk for the young, ignore those over 60. WebMD Medical News, May 3. Degner, L. F., Kristjanson, L. J., Bowman, D., Sloan, J. A., Carriere, K. C., OuNeil, J., Bilodeau, B., Watson, P., & Mueller, B. (1997). Information needs and decisional preferences in women with breast cancer. JAMA, 277(18), 1485–1492. DeMichele, A., Putt, M., Zhang, Y., Glick, J. H., & Norman, S. (2003). Older age predicts a decline in adjuvant chemotherapy recommendations for patients with breast carcinoma. Cancer, 97(9), 2150–2159. Du, X., & Goodwin, J. S. (2001). Patterns of use of chemo for breast cancer in older women: Findings from Medicare claims data. Journal of Clinical Oncology, 19(5), 1455–1466. Edwards, B. K., Howe, H. L., Ries, L. A. G., Thun, M. J., Rosenberg, H. M., Yancik, R., Wingo, P. A., Jemal, A., & Feigal, E. G. (2002). Annual report to the nation on the status of cancer, 1973–1999, featuring implications of age and aging on U.S. cancer burden. Cancer, 94(10), 2766–2792. Feldstein, A., Feldstein, P. J., Buchmueller, T., & Katterhagen, G. (2001). Breast cancer outcomes among older women: HMO, fee-for-service, and delivery system comparisons. Journal of General Internal Medicine, 16(3), 189–199. Ferguson, M. (1982). Forever feminine: Women’s magazines and the cult of femininity. London: Heimemann Educational Books Ltd. Fosket, J. R., Karran, A., & LaFia, C. (2000). Breast cancer in popular women’s magazines from 1913–1996. In: A. S. Kasper & S. J. Ferguson (Eds), Breast cancer: Society shapes an epidemic. New York: Palgrave.
304
LISA COX HALL
Friedman, L. C., Neff, N. E., Webb, J. A., & Laham, C. K. (1998). Age-related differences in mammography use and in breast cancer knowledge, attitudes and behaviors. Journal of Cancer Education, 13(1), 26–30. Friedson, E. (1970). Profession of medicine: A study of the sociology of applied Knowledge. Chicago: University of Chicago Press. Gamman, L., & Marchment, M. (1989). The female gaze: Women as viewers of popular culture. Seattle: The Real Comet Press. Guadagnoli, E., Shapiro, C., Gurwitz, J. H., Silliman, R. A., Weeks, J. C., Borbas, C., & Soumerai, S. B. (1997). Age-related patterns of care: Evidence against ageism in the treatment of early-stage breast cancer. Journal of Clinical Oncology, 15(6), 2338–2344. Gubrium, J. F., & Holstein, J. A. (1997). Method talk. The new language of qualitative method. New York: Oxford University Press. Holmes, C. E., & Muss, H. B. (2003). Diagnosis and treatment of breast cancer in the elderly. CA: A Cancer Journal for Clinicians, 53, 227–224. Jones, A. R., Thompson, C. J., Oster, R. A., Samadi, A., Davis, M. K., Mayberry, R. M., & Caplan, L. S. (2003). Breast cancer knowledge, beliefs, and screening behaviors among low-income, elderly black women. Journal of the National Medical Association, 95(9), 791–797, 802–805. Kasper, A. S., & Freguson, S. J. (2000). Breast cancer: Society shapes an epidemic. New York: Palgrave. Kerlikowske, K., Salzmann, P., Phillips, K. A., Cauley, J. A., & Cummings, S. R. (1999). Continuing screening mammography in women aged 70 to 79 years: Impact on life expectancy and cost-effectiveness. JAMA, 282(22), 2156–2163. Kimmick, G. G., & Balducci, L. (2000). Breast cancer and aging. Hemotology-Oncology Clinics of North America, 14(1), 213–234. Lantz, P. M., & Booth, K. M. (1998). The social construction of the breast cancer epidemic. Social Science and Medicine, 46(7), 907–918. Lash, T. L., & Silliman, R. A. (2001). Medical surveillance after breast cancer diagnosis. Medical Care, 39(9), 945–955. Lazovich, D., Raab, K. K., Gurney, J. G., & Chen, H. (2000). Knowledge and preferences for breast conservation therapy among women without breast cancer. Women’s Health Issues, 10(4), 210–216. Lofland, J., & Lofland, L. H. (1995). Analyzing social settings: A guide to qualitative observation and analysis. New York: Wadsworth Publishing Co. Lorde, A. (1980). The cancer journals. London: Sheba Feminist Publishers. Love, S. (2005). Dr. Susan Love’s breast book (4th ed.). Cambridge, MA: Da Capo Press. Marino, C., & Gerlach, K. K. (1999). An analysis of breast cancer coverage in selected women’s magazines, 1987–1995. American Journal of Health Promotion, 13(3), 163–170. Madan, A. K., Aliabadi-Wahle, S., & Beech, D. J. (2001). Age bias: A cause of underutilization of breast conservation treatment. Journal of Cancer Education, 16(1), 29–32. Marwill, S. L., Freund, K. M., & Barry, P. P. (1996). Patient factors associated with breast cancer screening among older women. Journal of American Geriatrics Society, 44(10), 1210–1214. Matthew, A., Pandey, M., & Rajan, B. (2004). Do younger women with non-metastatic and non-inflammatory breast carcinoma have poor prognosis? World Journal of Surgical Oncology, 2(1), 2.
Age, Breast Cancer and Disparity in Health Knowledge and Treatment
305
Meich, R. A., & Shanahan, M. J. (2000). Socioeconomic status and depression over the life course. Journal of Health and Social Behavior, 41(June), 162–176. Mishler, E. G. (1981). Social contexts of health, illness, and patient care. Cambridge: Cambridge University Press. Moen, P. (1996). Gender, age and the life course. In: R. H. Binstock & L. K. George (Eds), Handbook of aging and the social sciences. New York: Academic Press. Moody, H. R. (2002). Aging: Concepts and controversies. New York: Brookdale Center. Petrisek, A. C., Laliberte, L. L., Allen, S. M., & Mor, V. (1997). The treatment decision-making process: Age differences in a sample of women recently diagnosed with nonrecurrent early-stage breast cancer. Gerontologist, 37(5), 598–608. Rosenbaum, M. (1994). Cultural analysis of women, breast cancer surgery and the meaning of breasts. Lexington: University of Kentucky, Anthropology. Ryff, C., Marshall, V., & Clarke (1999). Linking the self and society in social gerontology: Crossing new territory via old questions. In: C. Ryff & V. Marshall (Eds), The self and society in aging processes. New York: Springer Publishing Co. Schieman, S. (1999). Age and anger. Journal of Health and Social Behavior, 40(September), 273–289. Schieman, S., & Turner, H. A. (1998). Age, disability, and the sense of mastery. Journal of Health and Social Behavior, 39(September), 169–186. Silliman, R. A., Demissie, S., & Troyan, S. L. (1999). The care of older women with early-stage breast cancer: What is the role of surgeon gender? Medical Care, 37(10), 1057–1067. Smith, G. C., Kohn, S. J., Savage-Stevens, S. E., Finch, J. J., Ingate, R., & Lim, Y. (2000). The effects of interpersonal and personal agency on perceived control and psychological wellbeing in adulthood. The Gerontologist, 40(4), 458–468. Smith-Bindman, R., Kerlikowske, K., Gebretsadik, T., & Newman, J. (2000). Is screening mammography effective in elderly women? American Journal of Medicine, 108(2), 112–119. Starr, P. (1982). The social transformation of American medicine. New York, NY: Basic Books. Sudman, S., & Bradburn, N. M. (1982). Asking questions. San Francisco: Jossey-Bass Publishers. Tuchman, G., Daniels, A. K., & Be-Net, J. (1978). Hearth and home: Images of women in the mass media. New York: Oxford University Press. Wanebo, H. J., Cole, B., Chung, M., Vezeridis, M., Schepps, B., Fulton, J., & Bland, K. (1997). Is surgical management compromised in elderly patients with breast cancer? Annals of Surgery, 225(5), 579–586. Wanzel, K. R., Brown, M. H., Anastakis, D. J., & Regehr, G. (2002). Reconstructive breast surgery: Referring physician knowledge and learning needs. Plastic and Reconstructive Surgery, 110(6), 1441–1450. Watters, J. M., Yau, J. C., O’Rourke, K., Tomiak, E., & Gertler, S. Z. (2003). Functional status is well maintained in older women during adjuvant chemotherapy for breast cancer. Annals of Oncology, 14(12), 1744–1750. Weissman, C. S. (2000). Breast cancer policymaking. In: A. S. Kasper & S. J. Ferguson (Eds), Breast cancer: Society shapes an epidemic. New York: Palgrave. Winship, J. (1987). Inside women’s magazines. London: Pandora. Wolf, N. (1991). The beauty myth: How images of beauty are used against women. New York: Anchor Book, Doubleday.
306
LISA COX HALL
Woodard, S., Nadella, P. C., Kotur, L., Wilson, J., Burak, W. E., & Shapiro, C. L. (2003). Older women with breast carcinoma are less likely to receive adjuvant chemotherapy: Evidence of possible age bias? Cancer, 98(6), 1141–1149. www.breastcancer.org. Narburth, PA. www.cancer.org. American Cancer Society 2006. American Cancer Society, Inc. Surveillance Research, Atlanta, GA. www.cdc.gov. Centers for Disease Control, Atlanta, GA. www.cms.hhs.gov. Centers for Medicare and Medicaid Services. U.S. Department of Health & Human Services, Baltimore, MD. www.statehealthfacts.org. Kaiser Family Foundation, Menlo Park, CA. Wyatt, G. K., & Friedman, L. L. (1998). Physical and psychosocial outcomes of midlife and older women following surgery and adjuvant therapy for breast cancer. Oncology Nursing Forum, 25(4), 761–768. Yarbrough, S. S. (2004). Online exclusive. Older women and breast cancer screening: Research synthesis. Oncology Nursing Forum, 31(1), E9–E15. Zones, J. S. (2000). Profits from pain: The political economy of breast cancer. In: A. S. Kasper & S. J. Ferguson (Eds), Breast cancer: Society shapes an epidemic. New York: Palgrave.
COLORECTAL CANCER PATIENTS DUALLY ELIGIBLE FOR VA AND MEDICARE HEALTH BENEFITS: IMPROVED HEALTH OUTCOMES OR INCREASED HEALTH DISPARITIES? Katherine S. Virgo, Mary P. Valentine, Lucille C. Dauz, Lan H. Marietta, Brandie S. Adams, Sangita Devarajan, Walter E. Longo and Frank E. Johnson ABSTRACT Many individuals are concurrently eligible for multiple sources of government-reimbursed health care services (e.g. Department of Veterans Affairs (VA) and Medicare). Unclear is whether combined eligibility translates into increased access to care and/or improved outcomes of care. Alternatively continuity of care may suffer, promoting health
Inequalities and Disparities in Health Care and Health: Concerns of Patients, Providers and Insurers Research in the Sociology of Health Care, Volume 25, 307–328 Copyright r 2008 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0275-4959/doi:10.1016/S0275-4959(07)00013-0
307
308
KATHERINE S. VIRGO ET AL.
inequalities when patients receive health services from multiple unrelated sources of care. The current study examines the impact of dual eligibility for government-reimbursed care on long-term outcomes of care for a population of veterans diagnosed with colorectal cancer and initially treated surgically at Department of Veterans Affairs Medical Centers.
INTRODUCTION The historical role of the U.S. government has been to provide for the needy, to furnish social and public goods, to regulate the marketplace, and to introduce accountability. Ten more specific roles have been proposed as better suited to the role of government as pertains to health care: (1) purchase health care services; (2) provide health care services; (3) ensure access to quality care for vulnerable populations; (4) regulate health care markets; (5) support acquisition of new knowledge; (6) develop and evaluate health technologies and practices; (7) monitor health care quality; (8) inform health care decision makers; (9) develop the health care workforce; and (10) convene stakeholders from across the health care system (Tang, Eisenberg, & Meyer, 2004). This expanded list of roles is more appropriate due to the patchwork-quilt nature of the U.S. health care system. Rather than a single universal health plan to cover all Americans, the U.S. system of health care is comprised of a series of federal and state government-sponsored entitlement programs (Medicare, Medicaid, Department of Defense, Department of Veterans Affairs, State Children’s Health Insurance Program) each providing differing levels of benefits for specific segments of the population. Those not qualifying for government programs must purchase some form of health insurance, pay for health services on a fee-for-service basis, rely on charitable assistance, or forego medical care completely. Currently, 46.6 million Americans (15.9% of the U.S. population) are uninsured (U.S. Census Bureau, 2006). Approximately 18.6 percent of the U.S. population had no insurance for at least part of 2006 (Cohen & Martinez, 2007). An additional 17.1 million insured individuals less than 65 years of age were underinsured, lacking adequate financial protection from high out-of-pocket costs (Banthin & Bernard, 2006). Conversely, many other Americans are eligible to receive health benefits through multiple programs concurrently, such as VA and Medicare.
Dual Eligibility for VA and Medicare Health Benefits
309
The reason that the U.S. system of health care is so piecemeal and complicated to navigate, compared to the universal coverage common to other countries, is often debated. Differences in ideology have been postulated as a major contributing factor. Four specifically American characteristics that have shaped the development of health policy over time are: (1) individualism, voluntarism, and a passion for liberty; (2) distrust of authority, especially governmental authority; (3) inconsistent demands for social equality; and (4) belief in the fairness of the market mechanism in distributing goods and services (Lee, Oliver, Benjamin, & Lee, 2006). These deeply rooted beliefs served as motivating factors as the various stakeholder groups that comprise the current health care arena developed and acquired power. National health insurance proposals were defeated repeatedly in 1935, 1945, and 1948–1949, by the strong influence of the American Medical Association (AMA) in concert with business, organized labor, and insurers. By the 1960s, groups such as the insurers and organized labor had taken more neutral positions and sweeping social legislation was passed in the form of Medicare and Medicaid (Quadagno, 2004). Though some believed these programs would be eventually broadened into universal health coverage, further attempts at passing national health legislation failed again under Nixon in 1974, Carter in 1979, and Clinton in 1994. Once again, stakeholder interests could not be appeased. Incrementalism in health policy making was once again demonstrated in 2003 with the passage of the Medicare Prescription Drug Improvement and Modernization Act of 2003. Representing the most significant reform in health care financing since Medicare, providing much needed prescription drug coverage for the elderly, the legislation provided no benefits for the 63.7 million uninsured and underinsured in the U.S. As the battle to achieve national health insurance continues, access, disparities, and continuity of care issues within the existing piecemeal health care system need to be addressed. A recent systematic review of disparities in the Veterans Affairs healthcare system grouped the review by disease categories (Saha, Freeman, Toure, Tippens, & Weeks, 2007). For patients with some cancers, blacks were less likely than whites to undergo potentially curative surgical resection, but equally likely to undergo such procedures as chemotherapy and radiation. It is postulated in the literature that physicians provide less information to black as opposed to white veterans, thus black veterans are less trusting of physicians. For those who are concurrently eligible for multiple sources of government-reimbursed health care services (e.g. Department of Veterans Affairs (VA) and Medicare), it is unclear whether combined eligibility
310
KATHERINE S. VIRGO ET AL.
translates into increased access and improved outcomes of care or decreased continuity of care and negative outcomes such as health inequalities. The current study examines the impact of dual eligibility for governmentreimbursed care on long-term outcomes of care for a population of veterans diagnosed with colorectal cancer and initially treated surgically at Department of Veterans Affairs Medical Centers.
BACKGROUND Each year approximately 153,800 new cases of colorectal cancer are diagnosed in the U.S. (Jemal et al., 2007). Thirty-nine percent of patients are diagnosed with localized cancers, 36 percent have regionalized cancers, and 19 percent have distant metastases (Ries et al., 2007). Colorectal cancer causes the third highest number of deaths related to cancer in the U.S. (Jemal et al., 2007). It has been estimated that the average number of years of life lost per person dying of colorectal cancer is 14.1 (National Institutes of Health, 2005). If colorectal cancer is detected in an early stage (localized), the five-year relative survival rates are 90.8 percent for colon cancer and 87.8 percent for rectal cancer (Ries et al., 2007). Once the cancer has spread to adjacent organs or lymph nodes (regionalized), survival rates drop markedly. Patients with distant metastases (advanced stage) have a five-year relative survival rate of only 10.3 percent. Five- and ten-year relative survival rates for all colorectal cancer patients in the U.S. (irrespective of stage) are 65.7 and 55.9 percent, respectively. The extensive resources that are devoted to cancer care nationwide, estimated at $78.2 billion in direct medical costs and $128.1 billion in indirect costs due to morbidity and premature mortality (National Institutes of Health, 2006), were the prime rationale for conducting the current research in a cancer population. If dual eligibility for health care services causes unnecessary duplication of services and/or negative outcomes, interventions can be designed to reduce inefficiency, thereby decreasing costs and improving patient satisfaction and outcomes. The study of dual use/eligibility arose as a result of ever-tightening restrictions on reimbursement. Initially restrictions were imposed in 1983 by the Medicare prospective payment system that tied hospital reimbursement to a fixed amount based on diagnosis-related groups. This was closely followed by the implementation in 1985 of the Medicare resource-based relative value scale that restricted physician fees according to set schedules.
Dual Eligibility for VA and Medicare Health Benefits
311
Subsequently, the Medicare requirement for utilization review facilitated the growth of managed care plans. Plans profit by competing for enrollment of the healthiest patients to maximize profit from the Medicare capitation payment. Managed care plans can also profit by enrolling certain categories of high-risk patients, such as Medicare beneficiaries who are also eligible for VA benefits. Profitability could be enhanced if these patients could be encouraged to enroll in a Medicare managed care plan while continuing to obtain at least some of their health care services through the VA. One of the earliest published studies of patients dually eligible for VA and Medicare benefits examined inpatient users of VA hospitals who underwent any of ten surgical procedures or were diagnosed with either acute myocardial infarction (MI) or hip fracture in the New England region or in New York state over a three-year period (Fleming, Fisher, Chang, Bubolz, & Malenka, 1992). Depending upon the diagnosis or type of surgery performed, from 17.6 to 37.4 percent of all 10,264 index admissions occurred outside the VA system. For inpatient users of VA hospitals undergoing colon resections, 19.7 percent of 534 index hospitalizations occurred outside the VA system. Readmissions within six months were then examined for the ten procedures and two diagnoses. Readmissions under Medicare among patients initially treated at the VA accounted for 12.4 percent of all 2,494 readmissions. As the purpose of readmissions within six months is generally to treat complications of treatment, this rate was higher than expected. Procedure- and diagnosis-specific readmission rates ranged from 6.2 to 17.2 percent (11.3% for colon resections). With the patient as the unit of analysis, 23.7–39.8 percent were readmitted to the VA within six months and 2.2–7.0 percent were readmitted to a non-VA hospital. For VA patients undergoing colon resections, 32.9 percent were readmitted to the VA within six months and 4.2 percent were readmitted to a non-VA hospital. Other more recent studies have examined dual-use patterns for specific diagnoses or procedures, but no study to date in the English-language literature has focused on VA/Medicare dual use among patients with colorectal cancer on a nationwide basis (Jia et al., 2007; Halanych et al., 2006; Rosen, Gardner, Montez, Loveland, & Hendricks, 2005; Stroupe et al., 2005; Wright, Daley, Fisher, & Thibault, 1997). Similarly, studies have examined dual-use rates and predictors of dual use for both primary care as a whole (Bean-Mayberry, Chang, McNeil, & Hayes, 2004; Borowsky & Cowper, 1999) and mental health care (Desai & Rosenheck, 2002). Dual-use rates for primary care based on patient self-report were
312
KATHERINE S. VIRGO ET AL.
38.7 and 28 percent, respectively. For mental health, defining non-VA use as use of any state mental health service, the dual-use rate was 7.7 percent. In these studies, dissatisfaction with VA care, health insurance, education beyond high school, income greater than $20,000, younger age, high VA mental health utilization levels, and travel time were identified as predictors of dual use. Though these studies make an important contribution to the literature, they may not be generalizable at a nationwide level. One study at the nationwide level examined the relationship between dual use and three groups of factors (patient, environment, and geographic) known to be related to access to care (Hynes et al., 2007). The dual-use rate for the 416,455 patients who used any inpatient services in 1999 was 6.4 percent. Dual use was divided into three categories for outpatient care: mostly VA (12.0%), equally dual (15.9%), and mostly Medicare (18.1%). ‘‘Mostly’’ was defined as 75–99 percent. ‘‘Equally’’ was defined as 26–74 percent. Combining the three dual-use categories, results in an overall dualuse rate for the 1,474,417 patients who used any outpatient services in 1999 of 46 percent. Patients with higher severity of illness scores were most likely to be dual users. In addition, even at the nationwide level, few studies in the Englishlanguage literature have attempted to examine the relationship between dual use/eligibility and outcomes. The two studies that examine the relationship between dual use and outcomes both focus only on mortality (Wolinsky et al., 2006; Wright et al., 1997). The more recent article uses a highly questionable proxy for VA dual use, defined as over-reporting of Medicare use (based on a comparison of Medicare claims data with self-reported inpatient utilization on the Survey of Assets and Health Dynamics among the Oldest Old), rather than actual VA data. Wolinsky et al. (2006) estimate that 11 percent of veterans 70 years of age and older are dual users. Mortality rates are estimated as 15.6 percent higher (64.9% vs. 49.3%) for dual users after controlling for sociodemographic characteristics, activities of daily living, self-rated health, five chronic diseases, cognitive ability, and depressive symptoms. Wright et al. (1997) obtained quite different results in their study of the subpopulation of patients with acute MI. After merging VA and Medicare data and adjusting for patient characteristics, Wright et al. (1997) found that veterans 65 and over with acute MI had similar 30-day, 90-day, and one-year mortality rates irrespective of whether they were initially admitted to a VA or Medicare hospital (OR: 1.0 vs. 1.04, CI: 0.97–1.11; OR: 1.0 vs. 1.04, CI: 0.97–1.10; OR: 1.0 vs. 1.02, CI: 0.96–1.08, respectively). The dual use rate in the Wright et al. (1997) study was 54 percent.
Dual Eligibility for VA and Medicare Health Benefits
313
METHODS A retrospective cohort study was conducted of patterns of health services utilization and outcomes of care post-treatment of all patients diagnosed with colorectal cancer (ICD-9-CM codes 153.0–154.1) and surgically treated for cure nationwide over the five-year period 1989–1993 in the VA system. A minimum of five years of utilization data were available for all patients after diagnosis, though for some patients as many as nine years of data were available. Death data were obtained through August 2006. Data for the three-year period prior to diagnosis were examined to verify that the presence of a colorectal cancer ICD-9-CM code in the data for fiscal years 1989–1993 was in fact documentation of new disease and not recurrence. The purpose of the study was to identify patients who were users of both VA and Medicare services and compare outcomes of care between dual users and VA only users. All patients surgically treated during fiscal years 1989–1993 were included in the study with the exception of those excluded according to the following criteria: (1) patient not eligible for Medicare (age o65 unless patient has end stage renal disease and/or qualifies for disability insurance benefits) three years prior to diagnosis; (2) diagnosis prior to fiscal year 1989; (3) earliest available date of treatment (date of diagnosis proxy) is for treatment of a recurrence of colorectal cancer rather than the index colorectal tumor; (4) index surgical resection not curative, i.e. gross tumor left behind; (5) patient died during the index admission; (6) surgery was a resection of metastasis of another cancer to the colon or rectum; (7) cancer was unresectable; (8) tumor staging data were unavailable; (9) cancer was not found at surgery, or colorectal mass was not malignant; and (10) index surgical resection conducted in a non-VA facility and not reimbursed by the VA. Although the inclusion of non-VA index surgical resections might be appealing, this was not possible for two reasons. First, the cost of obtaining the additional Medicare data was prohibitive since data would be required for all patients with colorectal cancer nationwide who were surgically treated to facilitate a match with VA data to select only those who are veterans and use the VA system. At the time this study began, the VA did not have a sharing agreement with the Centers for Medicare and Medicaid. Thus, all Medicare data had to be purchased. Second, the percentage return was expected to be low according to data available at that time (Fleming et al., 1992). Patients included in the study underwent at least one of the following procedures: (1) partial excision of the large intestine (ICD-9-CM codes 45.7–45.79), (2) total intra-abdominal colectomy (45.8), (3) local excision or
314
KATHERINE S. VIRGO ET AL.
destruction of lesion or tissue of rectum (48.3–48.35), (4) pull-through resection of the rectum (48.4–48.49), (5) abdominoperineal resection of rectum (48.5), or (6) other resection of rectum (48.6–48.69). Patients were identified as having a diagnosis of colorectal cancer and as having undergone one of the above six procedures through a search of the Patient Treatment File (PTF) for fiscal years 1989–1993. Date of treatment was used as a proxy for date of diagnosis because date of diagnosis is not available in the VA datasets. Prior experience with VA data for cancer patients has shown that date of treatment is a reasonable proxy (Wade, Virgo, Li, Callander, Longo, & Johnson, 1996; Wade, Radford, Virgo, Johnson, 1994). The earliest date on which the patient underwent any of the above procedures for treatment of colorectal cancer during the period 1989–1993 was considered the tentative date of diagnosis. Date of diagnosis determined in this fashion from the VA databases was confirmed as the date of diagnosis of the index tumor through a search of both VA and Medicare computerized records for the three years prior (Table 1). If no earlier record of treatment for colorectal cancer was uncovered in either database for the three-year period, date of diagnosis was considered confirmed. If a date of diagnosis prior to 1989 was discovered in either database, the patient was excluded from the study. Annually, slightly over 2000 patients nationwide are treated for colorectal cancer in the VA. Selecting only those patients who are Medicare-eligible reduces this number to approximately 1320 patients per year. For the fiveyear period (1989–1993) used to select the study sample, data on 6612 patients were available to the study. The overall sample size of 6612 is more than sufficient to permit comparison of the patterns of health services utilization post-treatment and outcomes of treatment between dual users and VA only users. Table 1.
Years of Data Required for the Proposed Analyses.
Fiscal Year in which Patient was Diagnosed
1989 1990 1991 1992 1993
Years Searched to Confirm Tumor is not a Recurrence
1986–1988 1987–1989 1988–1990 1989–1991 1990–1992
Years Searched Post-Diagnosis to Measure Utilization & Identify Date of Death 1990–1998 1991–1998 1992–1998 1993–1998 1994–1998
(9 (8 (7 (6 (5
years) years) years) years) years)
Dual Eligibility for VA and Medicare Health Benefits
315
Sources of Data The majority of the data used in this analysis were obtained from secondary data sources. The VA sources of data used for this study were the PTF, Outpatient Care File (OPC), Beneficiary Identification and Records Location System (BIRLS), tumor registry data, and medical records. The Austin Data Processing Center (DPC) maintains the PTF, OPC, and BIRLS databases for all VA’s nationwide. The databases were accessed via modem. The non-VA sources of data were the Medicare Enrollment Database, the Medicare Provider Analysis and Review (MEDPAR) File, the Medicare Outpatient Standard Analytical File, the Physician/Supplier Part B Standard Analytical File, and for the years 1986–1990, the Medicare Automated Data Retrieval System (MADRS) File (HCFA, 1997a, 1997b). Unlike the VA databases that use social security number as the patient identifier, Medicare databases use a beneficiary identification number that may change over time. To obtain all beneficiary identification numbers for each veteran for the time period of interest, the investigator provided a disk of names and social security numbers of all veterans in the study to be matched against the Medicare Enrollment Database (previously the Medicare Health Insurance Master File). This database contains both Social Security numbers as well as beneficiary identification numbers. The file resulting from this match procedure was used to select the requisite population from all other requested Medicare files. Once extracts of these datasets were received, the files were stored on the VA’s mainframe computer system in Austin, Texas, to facilitate matching Medicare data with the numerous VA datasets. Another unusual feature of the Medicare databases is that claims for services provided to a spouse, child, or parent are also identified under the beneficiary identification number of the primary beneficiary. Upon receipt of the claims files from Medicare, all claims unrelated to the care of the patients of interest were eliminated from the database using gender and date of birth to facilitate matching. The Medicare Enrollment Database was also used to identify which veterans were enrolled in Medicare health maintenance organizations (HMOs). Unfortunately, for the time period in which a veteran is enrolled in a Medicare HMO, detailed Medicare claims data were not available. The implication of this limitation is that estimates of non-VA health services utilization are conservative. The degree of conservativeness of these estimates depends on the percentage of the population who were enrolled in HMOs and the length of time enrolled in HMOs. According to Riley,
316
KATHERINE S. VIRGO ET AL.
Potosky, Lubitz, and Kessler (1995) approximately 8 percent of patients in their database, spanning the period 1978–1989, were enrolled in HMOs at some time during their study. In 1997, approximately 13 percent of Medicare beneficiaries were enrolled in Medicare HMOs (HCFA, 1997a, 1997b). The HMO enrollment percentage was expected to approximate the average of these two figures. An admission was considered colorectal cancer-related if colorectal cancer was coded as either the primary or any one of the secondary diagnoses. An outpatient visit was identified as colorectal cancer-related in one of four ways depending on the data source: (1) By the diagnosis code (available in all years of the Medicare institutional physician visit data and the Medicare Physician/Supplier File data (non-institutional physician visit data for 1991–1998), but only in fiscal years 1997 and 1998 for the VA OPC data); (2) By the outpatient procedure code (Current Procedural Terminology code) or clinic type for fiscal years 1991–1996 for the VA OPC data. Relevant procedures included colonoscopy, sigmoidoscopy, proctosigmoidoscopy, computed tomography of the abdomen, rectal ultrasound, abdominal ultrasound, barium enema, abdominal magnetic resonance imaging, and carcinoembryonic antigen test. Relevant clinic types included General Surgery, Radiation Therapy, Chemotherapy, Oncology, Colorectal Screen, FOBT-Guaiac Screen, Gastroenterology, GI Endoscopy, and Oncology Nuclear Medicine. Though primary care physicians managed the care of some patients, the care of the majority of patients was managed by either the surgeon who performed the initial tumor resection, a medical oncologist, or a gastroenterologist. Including all primary care and general internal medicine visits would have risked including visits unrelated to the treatment and follow-up of colorectal cancer; (3) By clinic type only for fiscal years 1986–1996 for the VA OPC data because diagnosis and procedure data are unavailable for these years; and (4) By diagnosis code and procedure code for the MADRS File data (non-institutional physician visit data for 1986–1990). In the MADRS File, medical care provided in an office setting is rolled up monthly as is ‘‘other than medical care’’ provided in an office setting. Thus, only a limited number of diagnosis and procedure codes are recorded, necessitating the use of both diagnosis code and procedure code to ensure identification of the maximum number of relevant non-institutional physician visits. As mentioned earlier, with the exception of the time that a veteran is enrolled in a Medicare HMO, all follow-up visits are generally recorded in either the VA or Medicare datasets. One additional factor that impacts the recording of visits slightly is the use of global fees or bundling in the
Dual Eligibility for VA and Medicare Health Benefits
317
Medicare data. A global fee is the single price charged for surgical services and the corresponding outpatient post-surgical follow-up care provided over the next 90 days. The importance of bundling of services is that there is no requirement that post-surgical follow-up visits be coded since there is no additional reimbursement to be gained. Such visits can be coded for documentation purposes only, though the frequency with which this is done was expected to be rare. For the purposes of this study, the use of global fees under Medicare was not anticipated to pose a major problem, since only those patients who were surgically treated at the VA for their primary colorectal cancer were included in the study. All follow-up visits conducted at the VA were coded in the VA datasets and all follow-up visits conducted at non-VA facilities were coded in the Medicare datasets (since the original surgery was not conducted as a Medicare patient). The only exception was for those visits conducted at a non-VA facility after surgery for a recurrence or second primary. Visits within the first 90 days were bundled with surgery. It is anticipated that the impact of this limitation will be small (1–2 visits for a small percentage of the population). The tumor registrar at each Department of Veterans Affairs Medical Center was initially contacted to obtain the primary tumor, regional lymph nodes, distinct metastasis (TNM) stage of the primary tumor at diagnosis, histopathology, treatment type (curative vs. palliative), and date of recurrence or new primary, if any. Copies of operative reports, pathology reports, and discharge summaries were then obtained from each facility on an as-needed basis only. Specifically, these reports became necessary when a tumor registrar’s data were incomplete or the data from two or more computerized sources conflicted. Though multiple staging systems are available for classifying colorectal cancer patients by severity of illness, the American Joint Committee on Cancer/International Union Against Cancer TNM system was used for the purposes of this study (Table 2). The TNM system is based on the depth of tumor penetration into the wall of the intestine, the number and site of regional lymph nodes involved with the tumor, and the presence or absence of distant metastases. Though the classification applies to both clinical and pathologic staging, most cancers of the colorectum are staged after pathologic examination of the resected specimen. Death data were obtained from the VA PTF, the VA BIRLS file, the MEDPAR file, the Social Security Administration (SSA) death file, and the Medicare Enrollment Database (Page, Braun, & Caporaso, 1995; Mahan, Page, & Kang, 1995; Fleming et al., 1992; Page, 1992; Boyle & Decoufle, 1990). The BIRLS database contains records of all veterans whose
318
KATHERINE S. VIRGO ET AL.
Table 2. TNM Classification System. Primary Tumor (T) TX T0 Tis T1 T2 T3
T4 Regional Lymph Nodes (N) NX N0 N1 N2 N3
Distant Metastasis (M) MX M0 M1 Histopathologic Grade (G) GX G1 G2 G3 G4
Primary tumor cannot be assessed No evidence of primary tumor Carcinoma in situ: intraepithelial or invasion of the lamina propria Tumor invades the submucosa Tumor invades the muscularis propria Tumor invades through the muscularis propria into the subserosa or into nonperitonealized pericolic or perirectal tissues Tumor directly invades other organs or structures and/or perforates the visceral peritoneum Regional lymph nodes No regional lymph node metastasis Metastasis in one to three pericolic or perirectal lymph nodes Metastasis in four or more pericolic or perirectal lymph nodes Metastasis in any lymph node along the course of a named vascular trunk and/or metastasis to apical node(s) (when marked by the surgeon). Presence of distant metastasis cannot be assessed No distant metastasis Distant metastasis Grade cannot be assessed Well differentiated Moderately differentiated Poorly differentiated Undifferentiated
Source: Fleming et al. (1997).
beneficiaries applied for death benefits at the time of the veteran’s death from 1960 forward. The BIRLS database records both hospital mortality as well as deaths outside the hospital. The PTF and MEDPAR files record deaths at discharge only. The SSA records deaths that lead to a cessation of Social Security benefits. The Medicare Enrollment Database combines death data obtained from intermediary carriers who process claims for Medicare as well as data from a variety of miscellaneous sources. Date of death was extracted from these files for use in conjunction with date of
Dual Eligibility for VA and Medicare Health Benefits
319
diagnosis in computing survival statistics. According to Fleming et al. (1992) total available death data were increased 2 percent by accessing both VA and Medicare sources of death data rather than using only one agency’s sources. Data obtained from tumor registrars were coded into a database maintained on a personal computer with ranges set for minimum and maximum acceptable values. A customized data cleaning routine was used to further reduce the potential for data entry error. Analysis was conducted using the Statistical Package for the Social Sciences (SPSS, 2006) and SAS for the mainframe (SAS, 1990a, 1990b). The coding sheets were pre-tested on a small sample of surgically treated colorectal cancer patients identified by a search of the inpatient database in Austin. The final version of the coding sheet was revised several times in the course of its development. This final version was reliable and practical for our purposes.
Statistical Analysis Three research questions were examined: (1) Is dual eligibility related to earlier TNM stage tumors at initial diagnosis? (2) Is dual eligibility related to lower total length of stay (LOS) for colorectal cancer admissions? (3) Does dual eligibility improve survival after treatment? Riley’s (1995) model of the four phases of disease progression guided the analyses: (1) initial (one month prior to diagnosis through six months after diagnosis); (2) continuing care (period between initial and pre-final, assuming survival of longer than two years from diagnosis); (3) pre-final (7–18 months prior to death through six months prior to death); and (4) final (last six months of life) (Fig. 1). Two categories of patients required special attention: those who survived less than 13 months and those who survived from 13 to 24 months. The health services utilization of patients who survived less than 13 months was categorized as into a single time period rather than into individual phases of disease progression. It was assumed that patients who survived from 13 to 24 months had no continuing care phase and moved directly into the prefinal phase. Chi-square analysis was used to examine categorical variables (such as the relationship between TNM stage and dual eligibility) and t-tests were used to examine continuous variables (such as LOS and dual eligibility). Kaplan Meier survival analysis was to determine whether dual eligibility improved survival after treatment.
320
KATHERINE S. VIRGO ET AL.
Patients who survive > 24 months 1 mo. prior
6 mos from
7-18 mos prior
6 mos prior
to diagnosis
diagnosis
to death
to death
Death
_____\________________\_____________________\_______________\__________________\ Initial Phase
Continuing Care Phase
Pre-Final
Final Phase
Patients who survive from 13-24 months Same as above without the continuing care phase. Patients who survive < 13 months Analyze as one period rather than as phases.
Fig. 1.
Four Phases of Disease Progression from Diagnosis to Death.
RESULTS Of the 6612 patients with diagnosis codes for colorectal cancer and procedures codes for curative-intent surgery, 4924 (74%) could be staged. After applying all the exclusion criteria, 4550 (92%) were eligible for analysis. The sociodemographic characteristics of these patients are displayed in Table 3 and clinical characteristics are displayed in Table 4. Of these, 77% and 23% were diagnosed with colon and rectal cancer, respectively. The distribution of TNM stage was 2.7% Stage 0, 25.9% Stage I, 33.2% Stage II, 24.7% Stage III, and 13.4% Stage IV. Histopathologic tumor grade was primarily either moderately differentiated (54%) or well differentiated (19%). Using a very strict definition of colorectal services, 13.9 percent of inpatients were dual users and 19.7 percent of outpatients were dual users. Combining inpatient and outpatient use, 5.8 percent of all patients were dual users of both inpatient and outpatient services and 27.9 percent were dual users of either inpatient or outpatient services. Using a much broader definition of cancerrelated services, 15.6 percent of inpatients and 31.5 percent of outpatients were dual users. Combining inpatient and outpatient use, 8.5 percent of all patients were dual users of both inpatient and outpatient services and 38.6 percent were dual users of either inpatient or outpatient services. For purposes of comparing dual users to VA only users in Tables 3 and 4, the conservative definition of colorectal services and the ‘‘either inpatient or outpatient services’’ definition of dual use was employed. Dual users of
Dual Eligibility for VA and Medicare Health Benefits
Table 3.
Age (N=4548) Non-white (N=4533) Female (N=4549) Marital status (N=4529) Married Widowed Divorced Never married Separated Region (N=4548) Southeastern Northeastern Mid-Atlantic Southwestern Midwest West Great lakes Metropolitan statistical area (N=4547)
321
Patient Sociodemographic Characteristicsa. All Patients
Dual Users of Inpatient or Outpatient Servicesb
Users of VA Services Only
(72.4, 5.5) 20.1 1.6
(72.1, 4.9) 16.0 2.4
(72.6, 5.6)** 21.7*** 1.4*
61.1 13.8 13.5 7.6 4.0
63.9 13.7 12.5 5.5 4.5
60.0** 13.8 13.9 8.4 3.9
27.8 21.1 19.7 14.5 6.9 5.7 4.4 69.3
25.6 23.9 23.2 13.2 5.4 5.3 3.4 62.4
28.6*** 20.0 18.4 14.9 7.5 5.8 4.8 72.0***
a Ns vary slightly from 4550 due to the presence of missing values. Categorical data are displayed as percentages. Continuous data are displayed in parenthesis as means and standard deviations. b The comparisons are based on the conservative definition of dual use. * po.05. ** po.01. *** po.001.
either inpatient or outpatient services were more likely to be younger ( po.01), female ( po.05), married or separated ( po.01), of white race ( po.001), living in the Northeast or Mid-Atlantic U.S. Census regions ( po.001), and from small cities and towns not considered part of a larger metropolitan statistical area (MSA) ( po.001). Dual users were more likely to be diagnosed with rectal cancer (26.6% vs. 21.2%) and less likely to be diagnosed with colon cancer ( po.001). Early diagnosis (TNM stages 0–II; po.001) and shorter length of hospital stay ( po.05) also favored the dual users. The majority of colon cancers were in the sigmoid colon or cecum. Colon cancer was most commonly treated with hemicolectomy or sigmoidectomy. Rectosigmoid junction was the most frequent site of rectal cancer. Abdominoperineal resection of the rectum was the treatment of choice for
322
KATHERINE S. VIRGO ET AL.
Table 4.
Colon cancer diagnosis (N=4540) Cancer site (N=4548) Sigmoid colon Cecum Ascending colon Transverse colon Descending colon Hepatic flexure Splenic flexure Other site of large intestine Rectosigmoid junction Rectum Surgery type (N=4548) Right hemicolectomy Sigmoidectomy Left hemicolectomy Abdominoperineal resection of rectum Other partial excision of large intestine Other anterior resection of rectum Resection of transverse colon Anterior resection of rectum with synchronous colostomy Other resection of rectum Total intra-abdominal colectomy Cecectomy Multiple segmental resection of rectum TNM stage at diagnosis 0 (Tis,N0M0) I (T1-2,N0,M0) II (T3-4,N0,M0) III (Any T,N1,M0) IV (Any T,Any N,M1)
Patient Clinical Characteristicsa. All Patients
Dual Users of Inpatient or Outpatient Servicesb
Users of VA Services Only
77.3
73.4
78.8**
24.5 11.8 9.4 5.5 4.3 2.5 2.0 9.9 18.1 11.8
23.9 11.6 7.4 5.7 4.3 2.0 2.0 9.7 19.0 14.4
24.7* 11.9 10.2 5.4 4.4 2.7 2.1 10.0 17.8 10.8
30.3 24.3 12.2 11.1
27.8 23.9 11.7 12.8
31.3 24.5 12.4 10.5
6.0
6.4
5.9
4.5
4.7
4.5
3.4
3.5
3.4
2.7
3.8
2.3
2.1 1.6
2.3 1.5
2.1 1.7
0.9 0.5
1.0 0.6
0.9 0.5
2.7 25.9 33.1 24.7 13.4
2.8 28.8 34.0 24.4 9.9
2.7** 24.8 32.8 24.9 14.8
Dual Eligibility for VA and Medicare Health Benefits
323
Table 4. (Continued )
Histopathologic type (N=4532) Adenocarcinoma Mucinous adenocarcinoma Adenocarcinoma in situ Other type Histopathologic grade (N=3999) Well differentiated Moderately to well differentiated Moderately differentiated Poorly to moderately differentiated Poorly differentiated or undifferentiated Pre-existing cancer (not colorectal or squamous cell skin cancer) (N=2591)c Length of stay, winsorized
All Patients
Dual Users of Inpatient or Outpatient Servicesb
Users of VA Services Only
88.5 9.6 0.7 1.2
88.2 9.8 0.9 1.1
88.5 9.6 0.6 1.3
21.0 6.7
20.5 7.1
21.2 6.6
60.9 2.4
61.8 2.4
60.5 2.4
9.0
8.1
9.3
12.7
12.5
12.7
(26.3, 25.2)
(24.8, 22.5)
(26.8, 26.2)*
a
N ¼ 4550 unless otherwise indicated. Variation from 4550 is due to the presence of missing values. Categorical data are displayed as percentages. Continuous data are displayed in parenthesis as means and standard deviations. b Comparisons are based on the conservative definition of dual use. c Based on the 2,591 patients for whom chart review data definitively confirmed whether preexisting cancer was present or not. * po.05. ** po.001
rectal cancer. As expected, the histopathologic type of most colorectal tumors was adenocarcinoma and the histopathologic grade was moderately differentiated. As displayed in Table 4, only a small number of patients had preexisting cancers (not colorectal or squamous cell skin cancer) identified by chart review at the time of their admission for treatment of the colorectal cancer. Not all patients were good historians and some charts had no records pre-dating the colorectal treatment admission, thus a history of preexisting cancer could not be ruled out entirely in some instances (N=1959). If these cases are assumed to have no preexisting cancer, the frequency of preexisting cancers drops to 7.3% due to the increased size of
324
KATHERINE S. VIRGO ET AL.
the denominator, but the results are still the same. Research is currently underway to augment the chart review data with history of preexisting cancer data available in the merged VA-Medicare datasets. Though there were no disparities in winsorized LOS according to gender, there were disparities according to race and MSA. For this analysis, irrespective of whether all six original race categories (Hispanic white, Hispanic black, American Indian, black, Asian, white) or only the three categories of white, black, and other were used, significant differences ( po.01) existed in winsorized LOS. Whites and blacks had the longest LOS (26.4 and 26.2, respectively). Asians and American Indians had the shortest LOS (19.6 and 20.6, respectively). In the analysis of winsorized LOS by MSA, patients from rural areas had significantly shorter ( po.01) LOS than patients in urban areas. By August 2006, 78.4% of patients had died. Average survival in months after treatment of the initial primary was 82.2 (s.d=67.1). Survivorship differed significantly by stage at diagnosis ( po.001), histopathologic grade ( po.01), tumor site ( po.05), region ( po.001), marital status ( po.001), age ( po.001), gender ( po.05), and dual use ( po.01). Dual users had a reduced hazard of death (0.874) compared to those who used VA-reimbursed services only, controlling for stage, histopathologic grade, region, marital status, age, gender, and tumor site.
DISCUSSION Disparities in access to care have been widely demonstrated. Such disparities have been demonstrated in the VA system of care as well, even though financial barriers to care are essentially minimized. In the current analysis, we examined whether disparities exist among patients dually eligible to receive health care benefits from both the VA and Medicare. Two important benefits of dual eligibility for government-reimbursed health care services were identified, significantly shorter LOS and significantly increased likelihood of survival. First limitation of the study is its use of secondary data that restricts the analysis to the variables and coding schemes in the selected datasets. Datasets such as the VA and Medicare datasets are primarily designed for administrative rather than research purposes and often lack such variables as current health status, patient satisfaction, functional limitations, and (for cancer patients) TNM stage at diagnosis. By combining datasets from
Dual Eligibility for VA and Medicare Health Benefits
325
multiple sources, the shortcomings of any one dataset can be readily overcome. A second limitation of the study is that patients are not surveyed to obtain data on personal assessments of current health status, limitations in physical activity, convenience to sources of health care, and perceived quality of health care facilities accessed. Work is currently underway to obtain proxy measures for each. A third limitation of the study is the underreporting of some procedures in the Medicare datasets prior to 1993 due to the ability to list only five procedures. However, it is doubtful that resections of colorectal cancer recurrence would not have been coded. It is most likely that diagnostic tests that individually generate little reimbursement for the respective hospital, such as blood tests, would be underreported. The significance of the study to the VA is threefold. First, expanded knowledge of patterns of health services utilization and the impact of newly established diagnoses on utilization should translate into improved ability to predict future demand for VA services for colorectal cancer patients. The ability to accurately predict future demand and plan services based on predicted needs is particularly important as the traditional VA patient base changes. The VA will need to become increasingly more competitive to attract patients, functioning more like hospitals in the private sector. The VA already has a strong focus on improving patient satisfaction, increasing coordination of patient care services, reducing waiting times, regionalizing administrative functions, increasing operational efficiency, and reducing operating costs. Second, increased data on the extent of care received outside the VA by such a clinically important group of patients should be of particular interest to medical center administrators. As the size of the World War II veteran population rapidly declines and the Operation Enduring Freedom and Operation Iraqi Freedom users of the VA health care system increase, it is of great financial importance to identify which services veterans obtain from the private sector and which services they obtain from the VA as well as trends in dual usage of services. For example, for acute MI it is already known that 27.6 percent of VA index patients cross over to non-VA hospitals to obtain bypass surgery and 12.1 percent cross over to undergo coronary angioplasty (Wright et al., 1997). It is unknown for which specific services VA cancer patients cross over to non-VA hospitals. Further, since individual VA facilities are now able to retain the Medical Care Cost Recovery dollars collected from third party payors such as private insurance plans and Medicare supplemental insurance plans, real health care dollars are lost when patients cross over to another health care system.
326
KATHERINE S. VIRGO ET AL.
Thus, identification of those services sought from non-VA hospitals rather than the VA should be a high priority item for the VA, particularly during this era of flatlined budgets. The finding that a high percentage of colorectal cancer patients are dual users suggests that continuity of care may be lacking. Third, combined VA/Medicare datasets provide an opportunity to determine whether healthcare services are needlessly duplicated, thus highlighting areas for improved efficiency. With the ongoing addition of VA and Medicare health services utilization data for the period 1999 to the present, the potential uses of this unique database include long-term survivorship studies. Such studies will permit identification of long-term complications of therapy that previously were not detected as patients did not survive long enough for such complications to develop (Feuerstein, 2007). Recurrence and second primary detection and treatment rates are also currently under analysis.
REFERENCES Banthin, J. S., & Bernard, D. M. (2006). Changes in financial burdens for health care: National estimates for the population younger than 65 years, 1996 to 2003. Journal of the American Medical Association, 296, 2712–2719. Bean-Mayberry, B., Chang, C., McNeil, M., & Hayes, P. (2004). Comprehensive care for women veterans: Indicators of dual use of VA and non-VA providers. Journal of the American Medical Women’s Association, 59, 192–197. Borowsky, S. J., & Cowper, D. C. (1999). Dual use of VA and non-VA primary care. Journal of General Internal Medicine, 14, 274–280. Boyle, C. A., & Decoufle, P. (1990). National sources of vital status information: Extent of coverage and possible selectivity in reporting. American Journal of Epidemiology, 131, 160–168. Cohen, R.A., & Martinez, M.E. (2007). Health insurance coverage: Early release of estimates from the National Health Interview Survey, 2006. Available from: http://www.cdc.gov/ nchs/nhis.htm Desai, R. A., & Rosenheck, R. A. (2002). The impact of managed care on cross-system use of mental health services by veterans in Colorado. Psychiatric Services, 53, 1599–1604. Feuerstein, M. (2007). Cancer survivorship: A bird’s eye view from an insider. In: Handbook of Cancer Survivorship (pp. 3–6). New York: Springer Science+Business Media, LLC. Fleming, I. D., Cooper, J. S., Henson, D. E., Hutter, R. V. P., Kennedy, B. J., Murphy, G. P., O’Sullivan, B., Sobin, L. H., & Yarbro, J. W. (Eds). (1997). American Joint Committee on Cancer: Cancer Staging Manual (5th ed., pp. 83–90). PA: Lippincott-Raven Publishers.
Dual Eligibility for VA and Medicare Health Benefits
327
Fleming, C., Fisher, E. S., Chang, C., Bubolz, T. A., & Malenka, D. J. (1992). Studying outcomes and hospital utilization in the elderly: The advantages of a merged database for Medicare and Veterans Affairs hospitals. Medical Care, 30, 377–391. Halanych, J. H., Wang, F., Miller, D. R., Pogach, L. M., Lin, H., Berlowitz, D. R., & Frayne, S. M. (2006). Racial/ethnic differences in diabetes care for older veterans: Accounting for dual health system use changes conclusions. Medical Care, 44, 439–445. Health Care Financing Administration. (1997a). Data compendium. HCFA Publication no. 03395. Bethesda, MD: Health Care Financing Administration. Health Care Financing Administration. (1997b). HCFA data users reference guide. Baltimore, MD: Health Care Financing Administration. Hynes, D. M., Koelling, K., Stroupe, K., Arnold, N., Mallin, K., Sohn, M., Weaver, F. M., Manheim, L., & Kok, L. (2007). Veterans’ access to and use of medicare and veterans affairs health care. Medical Care, 45, 214–223. Jemal, A., Siegel, R., Ward, E., Murray, T., Xu, J., & Thun, M. J. (2007). Cancer statistics, 2007. CA: A Cancer Journal for Clinicians, 57, 43–66. Jia, H., Zheng, Y., Reker, D. M., Cowper, D. C., Wu, S. S., Vogel, B., Young, G. C., & Duncan, P. W. (2007). Multiple system utilization and mortality for veterans with stroke. Stroke, 38, 355–360. Lee, P., Oliver, T., Benjamin, A. E., & Lee, D. (2006). Politics, health policy, and the American character. Stanford Law & Policy Review, 17, 7–32. Mahan, C. M., Page, W. F., & Kang, H. K. (1995). Vital status ascertainment through the files of the department of veterans affairs and the social security administration. American Journal of Epidemiology, 141, S58. National Institutes of Health. (2005). 2005 Fact Book: National Cancer Institute, Bethesda, MD, http://fmb.cancer.gov/financial/attachments/FY-2005-FACT-BOOK-FINAL.pdf National Institutes of Health. (2006). Fact Book Fiscal Year 2005: National Heart, Lung, and Blood Institute, Bethesda, MD, http://www.nhlbi.nih.gov/about/factbook/toc.htm Page, W. F. (1992). VA mortality reporting for World War II army veterans (letter). American Journal of Public Health, 82, 124–125. Page, W. F., Braun, M. M., & Caporaso, N. E. (1995). Ascertainment of mortality in the U.S. veteran population: World War II veteran twins. Military Medicine, 160, 351–355. Quadagno, J. (2004). Physician sovereignty and the purchasers’ revolt. Journal of Health Politics, Policy and Law, 29, 815–834. Ries, L. A. G., Melbert, D., Krapcho, M., Mariotto, A., Miller, B. A., Feuer, E. J., Clegg, L., Horner, M. J., Howlader, N., Eisner, M. P., Reichman, M., & Edwards B. K. (Eds). (2007). SEER Cancer Statistics Review, 1975–2004, National Cancer Institute, Bethesda, MD, http://seer.cancer.gov/csr/1975_2004/, based on November 2006 SEER data submission posted to the SEER web site. Riley, G. F., Potosky, A. L., Lubitz, J. D., & Kessler, L. G. (1995). Medicare payments from diagnosis to death for elderly cancer patients by stage at diagnosis. Medical Care, 33, 828–841. Rosen, A. K., Gardner, J., Montez, M., Loveland, S., & Hendricks, A. (2005). Dual-system use: Are there implications for risk adjustment and quality assessment? American Journal of Medical Quality, 20, 182–194. Saha, S., Freeman, M., Toure, J., Tippens, K., & Weeks, C. (2007). Racial and ethnic disparities in the VA healthcare system: A systematic review. Prepared for: Department of Veterans
328
KATHERINE S. VIRGO ET AL.
Affairs Health Services Research & Development Service Evidence Synthesis Pilot Program. SAS Institute Inc. (1990a). SAS/STAT User’s Guide, Version 6 (Vol. 1, 4th ed.). Cary, NC: SAS Institute Inc. SAS Institute Inc. (1990b). SAS/STAT User’s Guide, Version 6 (Vol. 2, 4th ed.). Cary, NC: SAS Institute Inc. SPSS Inc. (2006). SPSS 15.0. Chicago, IL: SPSS Inc. Stroupe, K. T., Hynes, D. M., Giobbie-Hurder, A., Oddone, E. Z., Weinberger, M., Reda, D. J., & Henderson, W. G. (2005). Patient satisfaction and use of veterans affairs versus non-veterans affairs healthcare services by veterans. Medical Care, 43, 453–460. Tang, N., Eisenberg, J. M., & Meyer, G. S. (2004). The roles of government in improving health care quality and safety. Joint Commission Journal on Quality and Safety, 30, 47–54. U.S. Census Bureau. (2006). Current population survey 2006 annual social and economic supplements. Income, poverty, and health insurance coverage in the United States: 2005, pp. 20–26. Wade, T. P., Radford, D. M., Virgo, K. S., & Johnson, F. E. (1994). Treatment of pancreatic cancer in the U.S. veteran: Complications and outcome. Journal of the American College of Surgeons, 179, 38–48. Wade, T. P., Virgo, K. S., Li, M. J., Callander, P. W., Longo, W. E., & Johnson, F. E. (1996). Outcomes after detection of metastatic carcinoma of the colon and rectum in a national hospital system. Journal of the American College of Surgeons, 182, 353–361. Wolinsky, F. D., Miller, T. R., An, H., Brezinski, P. R., Vaughn, T. E., & Rosenthal, G. E. (2006). Dual use of medicare and the veterans health administration: Are there adverse health outcomes? BMC Health Services Research, 6, 131–141. Wright, S. M., Daley, J., Fisher, E. S., & Thibault, G. E. (1997). Where do elderly veterans obtain care for acute myocardial infarction: Department of veterans affairs or medicare? Health Services Research, 31, 739–754.