Linda G. Martin and Beth J. Soldo
Although the survival advantage of whites over blacks has generally declined over the past 30 years, there is continuing debate about differential mortality patterns by age, as well as concern about broader health disadvantages of blacks relative to whites. There is some evidence that the black mortality curve crosses over that of whites in old age—that is, that blacks have an advantage at the oldest ages—but this finding is subject to argument. Moreover, this mortality advantage, if it exists, is not necessarily associated with lower rates of morbidity and disability.
At the same time, we need to better understand the health situations of older Hispanics and Asian/Pacific Islanders, and the extent to which their immigration experiences and cultural heritages are positively or negatively related to their health outcomes. The elderly population in the United States has become more racially and ethnically diverse in recent years, and this trend is expected to continue. Projections based on recent trends in life expectancy and immigration show that the Hispanic origin and ''other race" (Asian/Pacific Islanders; and American Indian, Eskimo, and Aleut) populations aged 65 and over in the United States will each increase elevenfold by the middle of the next century. The black elderly population is expected to more than triple, while the white non-Hispanic population will just double. White non-Hispanic persons were 87 percent of the population aged 65 and over in 1990, but they will be 67 percent of the much larger population aged 65 and over in the year 2050 (Hobbs, 1996).
Of course, making predictions on the basis of such projections is perilous. The Census Bureau figures cited above are middle-range projections that are based on current trends in immigration and life expectancy. Projections must
also contain, if only implicitly, forecasts of how today's preteenagers, teenagers, and young adults (who will constitute the 65-and-older age group in the year 2050), and the future immigrants who will join them, will identify their race and ethnicity when they are asked these questions in the year 2050. It is likely that the race and ethnicity categories used in census and official statistics will change several times by then.
This potential for change underscores an important point for the analyses discussed here: race and ethnicity are fluid categories, whose meanings vary and are to be understood in a particular social and historical context. They are not biological taxa. In this volume we have, wherever possible, used the racial and ethnic classification adopted for reporting purposes in federal government publications (Office of Federal Statistical Policy and Standards, 1978).1 Many of the large data sets analyzed in the chapters that follow used these categories, and vital rates and population figures with which the smaller studies are compared tend to use them. But within these categories there is much diversity (cultural, socioeconomic, behavioral, genetic) that is relevant to health outcomes. Several of the chapters address some of the challenges associated with identifying membership in particular groups.
It is common for researchers concerned with health outcomes (including mortality) to control for race and ethnicity in their analyses. Such procedures statistically adjust for a range of factors that are known to be related to health and that vary across racial and ethnic groups. Of late, however, there has been renewed interest in understanding these racial and ethnic differences and their potential implications for the mix and distribution of health states within the population. Socioeconomic arguments cite the consequences of lifelong poverty. Relevant factors include both early-life differences, such as birth weight and childhood nutrition, and midlife variables, such as access to employer-provided health insurance, the strain of physically demanding work, and exposure to a broad range of toxins, both behavioral (e.g., smoking) and environmental (e.g., workplace exposures). Over the life cycle, these factors combine to increase the demand for health care, while potentially limiting consumption of necessary health services. In late life, these factors may affect the age of onset of both morbidity and disability, the severity of symptoms, and ultimately the age at, and cause of, death. Recent research also highlights the enduring effects of education. Increased education appears to lower the risks for some chronic diseases—most notably, coronary heart disease and, perhaps most intriguingly, organic dementias—while retarding the pace of disease progression for other conditions (Snowdon et al., 1996; Feinstein, 1993).
In contrast to these socioeconomic explanations, cultural theories emphasize differences in norms regarding lifestyle and self-care behaviors, contacts with health care providers, and treatment compliance. Moreover, the experience of racial and ethnic discrimination may have adverse psychological and physiological effects, in addition to limiting the quantity and quality of health care received. Still other research suggests that there are race-related genetic factors both for predisposing conditions, such as hypertension and diabetes mellitus, and for life-threatening conditions, such as aplastic anemia.
Most commonly, competing causal explanations are examined independently of one another, although it is unlikely that any one approach can solely account for differences observed by race or ethnicity. The Committee on Population believed that it would be worthwhile to bring together experts from a variety of disciplines to examine and evaluate alternative models. The overarching goal of the committee's 1994 workshop, which was sponsored by the National Institute on Aging, was to make progress in understanding the extent to which racial and ethnic differences reflect differences in socioeconomic status, health-promoting behaviors, access to health care, genetics, and other factors.
Ideally, we would have had papers that addressed all aspects of health—both total and cause-specific mortality, morbidity, and disability—for each of the major ethnic and racial groups in the United States, as well as providing systematic consideration of the potential causal factors mentioned above. Data limitations, funds, and the current status of the field did not allow us to fill each cell of the multidimensional matrix. Rather, we chose to focus on critical cells in which either research is most advanced or the need for information is greatest—something we hope will itself be a substantial contribution. Accordingly, the nine papers in this volume range from overviews of racial and ethnic differences in the measures of health outcomes to in-depth looks at particular causal factors to investigations of specific diseases or specific ethnic groups.2 The result of our pragmatic approach has been a primary focus on black-white differentials, given that the bulk of available data and analysis have highlighted these. Nevertheless, several of the papers provide insight into the health of the growing proportion of the elderly who are Hispanic or Asian/Pacific Islanders.
The academic disciplines represented include the social and behavioral sciences, demography, epidemiology, genetics, and medicine. The data sources used were even more diverse, including censuses, death registries, administrative records from Social Security and Medicare, national surveys (e.g., the National Longitudinal Mortality Survey, the National Long Term Care Survey, the Health
and Retirement Survey, the Asset and Health Dynamics of the Oldest Old [AHEAD] survey, the National Medical Expenditures Survey), and small area studies (e.g., the New Haven Established Populations for the Epidemiologic Study of the Elderly, the Ni-Hon-San Studies of Japanese and Japanese Americans, the North Manhattan Aging Project).
The volume begins with papers by Irma Elo and Samuel Preston and by Kenneth Manton and Eric Stallard that assess overall differences in mortality among racial and ethnic groups of older Americans and, in particular, investigate the often-observed crossover of mortality rates of blacks and whites in late life. Both papers recognize problems associated with age reporting that may contribute to the apparent black-white crossover in mortality. Elo and Preston (Chapter 2) bring the keen perspective and tools of demography to the issue. They review and evaluate several studies using different data sets and estimation techniques to address the crossover question. Only one study, which is based on a relatively small sample, fails to detect a crossover, but this study is noteworthy because the data set on which it is based has been so painstakingly constructed. The authors and M. Hill matched death certificates of blacks ages 60 and older in 1980 and 1985 with records from the U.S. census and records from the Social Security Administration. Their findings of age misreporting from these matched data are used to correct the age distribution of deaths among blacks. When the corrected distribution is compared with the distribution for whites, no crossover in mortality at older ages is evident.
Manton and Stallard (Chapter 3) link racial and ethnic patterns of age-specific mortality to age patterns of specific disease processes, namely, osteoporosis and hip fracture, heart diseases and stroke, and cancers. They argue that these processes, in combination with the mortality selection of frail persons, provide evidence of a crossover at later ages. Their analysis of death certificates in particular reveals a crossover, even when allowing for a plausible degree of age misreporting. Manton and Stallard conclude their paper with an analysis of disability and active life expectancy and find that older blacks, despite their survival advantage, experience more disability than whites, who they suggest are more subject to acute disease.
Although these first two papers come to conflicting conclusions on the mortality-crossover issue, we believe that both make important contributions to the debate and that together they highlight the value of different approaches and the need for further work on the question. The two papers agree that there is Hispanic advantage in survival at older ages relative to whites and blacks. Elo and Preston's analysis also highlights the possible contribution of immigrant status to the mortality advantage both for Hispanics and for Asians and Pacific Islanders. They argue that this advantage may be due in part to the selectivity of migration. (Immigrants may be among the healthiest and hardiest individuals in their countries of origin.) But the apparent mortality advantage for immigrants may also be partly an artifact of measurement: some immigrants return to their countries of
origin in old age, and thus their deaths are not reported in the United States. Moreover, as noted in other papers in the volume, older Hispanics, despite their apparent mortality advantage, report more physical and functional health problems than do non-Hispanic whites.
The next set of four papers focuses on possible causal pathways to racial and ethnic differentials in health. James Smith and Raynard Kington (Chapter 4) focus on the complex interactions linking race, socioeconomic status, and health. Their review of the literature indicates that taking socioeconomic differences into account eliminates a significant proportion of observed racial differences in health status. Moreover, most of the difference in mortality is eliminated. Smith and Kington use data from the new Health and Retirement Survey (HRS) and the AHEAD survey, which provide the best information to date on the income and wealth of older Americans. They estimate models that allow for nonlinear effects of economic variables and that include risk factors, such as smoking, drinking, and exercise. They find that the effects on health status of being black or Hispanic relative to being non-Hispanic white are substantially reduced by the inclusion of measures of income and wealth. The risk factors, though in some cases statistically significant, have only modest collective effects on racial and ethnic differences in health status in these models. Finally, Smith and Kington present a preliminary investigation of reverse causation that indicates that the feedbacks from health to current socioeconomic status are probably more important than the effects of socioeconomic status on health in the short run.
Taking a step back from health outcomes, Lisa Berkman and Jewel Mullen (Chapter 5) focus on black-white differences in health-damaging and health-promoting behaviors, controlling where possible for socioeconomic status. They base their observations on a general review of the literature, as well as their own analyses of data from the New Haven Established Populations for the Epidemiologic Study of the Elderly. Despite greater apparent concern on the part of blacks than whites about their health, blacks do not consistently adopt more beneficial behaviors than whites. Older blacks engage in less physical activity and are more likely to be obese (especially women), but they are less likely to consume alcohol than whites. Racial differences in smoking patterns are complex, with older blacks less likely to have ever smoked but, if they have, less likely to have quit. Lack of exercise and obesity are associated with hypertension and diabetes, both of which have been reported to be twice as common among blacks than among whites. Berkman and Mullen also explore the role of social networks and social support in influencing health, but find few racial differences in summary measures of social support.
José Escarce and Frank Puffer (Chapter 6) examine the use of medical care of older blacks and whites, yet another possible factor underlying differences in health outcomes. Using data from the 1987 National Medical Expenditure Survey, they analyze racial differences in total medical care expenditures (irrespective of source of payment), physician visits, and inpatient hospital nights. They
estimate two types of models: one that adjusts for variables that primarily describe need for medical care and one that includes additional demographic and socioeconomic characteristics and represents more of what they call a demand perspective. In the latter, although blacks were slightly less likely than whites to have some medical expenses, there were no significant differences in the level of expenditure, conditional on having at least some. In the former, which included only age, sex, race, health status, and measures of attitudes and beliefs about health care, race had a larger effect: whites were significantly more likely than blacks to have at least some medical expenditures and some physician visits. Moreover, conditional on some utilization, blacks had lower levels of use. Escarce and Puffer conclude that racial differences in the quantity of medical care received by elderly people in the United States have largely disappeared, but that the care that blacks receive may not fully reflect their differential health needs. Thus, older blacks may still be underserved by health care services.
James Neel (Chapter 7) also focuses on black-white differences in health, but examines possible genetic bases for these differences. Despite the "avalanche" in the discovery of genetic variation since World War II, Neel argues that most variation has no effect on survival. Among the polymorphisms that have been related to health are those in the code for human leukocyte antigens, which are associated with autoimmune disorders in middle or late life. Although these polymorphisms are present in both blacks and whites, the establishment of an association between an allele and a disease in one group does not necessarily imply that it exists in another, and in this case, the association has been established only in white populations. Another association currently receiving great attention is that of Alzheimer's disease and the type 4 allele of the apolipoprotein E system, but once again research on the link among whites is more advanced.
Next Neel highlights two diseases—hypertension and diabetes—that, depending on the definitions of disease used, are relatively more prevalent among blacks than whites. He notes the familial nature of both, but reminds us that familial patterns do not necessarily reflect genetic differences per se. Both diseases are heterogeneous and are caused by multiple factors, but a number of rare subtypes with strong genetic linkages have been identified.
In sum, knowledge about the genetic basis of racial differentials in adult and late-life diseases is meager. Neel concludes by noting that blacks and whites are generally very similar genetically and that differences in environment play a strong role in determining how inherited susceptibilities are expressed. Until researchers become more adept at measuring and controlling for these factors, it will be difficult to reach conclusions regarding racial differences in susceptibility to complex diseases.
The final three papers of the volume focus on either specific diseases or specific ethnic groups. Barry Gurland and his associates (Chapter 8) highlight a disease not discussed at length in the earlier papers, but one with important implications for the quality of life and the need for long-term care. They report
on a study in North Manhattan of dementia among Hispanics, blacks, and non-Hispanic whites, in which particular care is taken to minimize ethnic and racial biases in diagnosis, some of which may be based in differences in educational attainment. They find that Hispanics and blacks have a higher age-adjusted prevalence of dementia than non-Hispanic whites. However, in a multivariate analysis that controls for education and sex, ethnic and racial differences are not statistically significant. They propose three hypotheses for the relation of education to dementia: (1) that education is curtailed relatively early owing to precursors of dementia, and thus there is reverse causation; (2) that low educational attainment is associated with other deprivations that are related to dementia; and (3) that education builds and maintains a robust neurobiological structure.
To understand the implications of dementia for quality of life and to further test the possibility of ethnic or racial bias in diagnosis, the Gurland team examines the prevalence of memory complaints, functional impairments, and depression among those with advanced dementia and those with border-zone dementia in the three racial/ethnic groups, and they find similar patterns. There are, however, striking differences in service utilization by dementias, with non-Hispanic whites much more likely to be in nursing homes than the other two groups and with blacks likely to be relatively greater users of home care and hospitals. Hispanics and blacks are also more likely to use emergency clinics than whites.
The penultimate paper provides insight into the role of culture by studying coronary heart disease among Japanese Americans and Japanese in Japan. Dwayne Reed and Katsukiko Yano (Chapter 9) compare three groups of men, all of whose grandparents were native Japanese and all of whom were born between 1900 and 1919, but who were distinguished by their residence in the mid- to late-1960s in three different locations: California, Hawaii, and Japan. At baseline examination, age-adjusted prevalence rates indicated generally greatest risk factors for coronary heart disease among the Californians and lowest among those in Japan. The only exception was for smoking, which was highest in Japan. A similar pattern, with Hawaiians intermediate, is found for myocardial infarction and coronary heart disease over various follow-up periods.
Further analysis by birthplace of the group residing in Hawaii found that birth in Japan or extended residence there when younger was generally associated with lower risk factors for coronary heart disease. Various measures of retention of Japanese culture were also predictive of lower risk factors, but in multivariate models of coronary heart disease that included these risk factors, the cultural measures did not have significant effects. Nor were measures of psychosocial stresses statistically significant. The study suggests the important role that modifying high-risk behavior can play in reducing coronary heart disease, highlights cultural influences on those risk factors, and suggests the potential for alterations in risk through community-level interventions.
The final paper, by Kyriakos Markides and colleagues (Chapter 10), reviews the health of older Hispanics, who currently make up only a very small propor-
tion of older Americans but whose numbers are likely to increase rapidly in the future. The authors note the overall advantage of Hispanics in comparison with non-Hispanic whites and blacks in mortality, but caution that there is significant intragroup heterogeneity, with Cuban Americans most advantaged, Puerto Ricans least, and Mexican Americans intermediate. They also note that the Hispanic mortality advantage holds for most major causes of death; only diabetes and liver disease/cirrhosis are exceptions. Given Hispanics' generally lower socioeconomic status as well as their relatively more vulnerable risk profiles, this advantage is difficult to explain. Speculation has focused on protective cultural effects, as well as selective immigration, but the authors cite conflicting evidence of effects of immigrant status on broader indicators of health and reinforce the fact noted earlier that the Hispanic advantage in mortality is not accompanied by a Hispanic advantage in disability.
As a group, these nine papers increase our sophistication in thinking about racial and ethnic differences among older Americans. Besides providing the best estimates to date of differences in mortality and other aspects of health, they underscore the critical interactions of socioeconomic status and environment with race and ethnicity and provide perspective on the roles of culture and immigration experience.
Many of the issues that motivated this volume remain unresolved. Gaps in nationally representative data sets are partially at fault. Several studies provide detailed measures of adult health transitions and late-life socioeconomic status, but no study provides a full range of comparable life-history measures for the individual, no less the family. We lack, of course, survey data linked with genetic markers, and the ethical and legal obstacles to building such an integrated data set are formidable. Finally, there is a continuing need to clarify analytically what it is we mean by racial and ethnic differences. Such research moves us into more refined discussions of unobserved heterogeneity. It also requires that we generate testable hypotheses of how culturally distinct differences emerge and the extent to which such differences are diluted through intermarriage or socioeconomic integration. These are topics that have not commanded much thoughtful reflection in the demography of aging but that are clearly an important part of any emerging research agenda in the field.
Anderson, N.B., and C. McManus 1996 Hypertension in Blacks Across the Life Course: A Biopsychosocial Analysis. New York: Springer.
Edmonston, B., J. Goldstein, and J.T. Lott, eds. 1996 Spotlight on Heterogeneity: The Federal Standards for Racial and Ethnic Classification. Washington, DC: National Academy Press.
Feinstein, J.S. 1993 The relationship between socioeconomic status and health: A review of the literature. Milbank Quarterly 71(2):279-322.
Freeman, H.P. 1991 Race, poverty, and cancer. Journal of the National Cancer Institute 83(8):526-527.
Hobbs, F.B., with B.L. Damon 1996 Sixty-five plus in the United States. Current Population Reports, Special Studies, P23-190. Washington, DC: Bureau of the Census.
Office of Federal Statistical Policy and Standards 1978 Federal Statistical Policy Directive No. 15: Race and Ethnic Standards for Federal Statistics and Administrative Reporting. Washington, DC: U.S. Department of Commerce.
Snowdon, D.A., S.J. Kemper, J.A. Mortimer, L.H. Greiner, D. Wekstein, and W.R. Merkesbery 1996 Linguistic ability in early life and cognitive function and Alzheimer's disease in late life: Findings from the Nun Study. Journal of the American Medical Association 275(7):528-532.