NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Research Council (US) Panel on Race, Ethnicity, and Health in Later Life; Anderson NB, Bulatao RA, Cohen B, editors. Critical Perspectives on Racial and Ethnic Differences in Health in Late Life. Washington (DC): National Academies Press (US); 2004.

Cover of Critical Perspectives on Racial and Ethnic Differences in Health in Late Life

Critical Perspectives on Racial and Ethnic Differences in Health in Late Life.

Show details


Barney Cohen

Life expectancy at birth in the United States has improved dramatically over the past century—from about 46.3 years for men and 48.3 years for women in 1900 to about 74.1 years for men and 79.5 years for women in 2000 (National Center for Health Statistics [NCHS], 2003). But the story behind this piece of good news is a story in two parts. During the first half of the century, life expectancy at birth rose largely because of improvements in nutrition, housing, hygiene, and medical care, as well as the prevention and control of major childhood infectious diseases (NCHS, 2003). But throughout the second half of the century, advances in medicine—particularly in relation to the treatment of heart disease and stroke—along with healthier lifestyles, improvements in access to health care, and better general overall health before reaching age 65 combined to result in continued improvements in life expectancy (Fried, 2000). As a result, in the first half of the 20th century, life expectancy at birth rose dramatically while gains in life expectancy above age 65 were relatively modest. Between 1900 and 1950, life expectancy at birth rose 42 percent for men and 47 percent for women, while life expectancy at 65 only rose 11 percent for men and 23 percent for women. But between 1950 and 2000, the pattern was completely different. In the second half of the century, demographers recorded small gains in life expectancy at birth and large gains in life expectancy above age 65. Life expectancy at birth over this latter period rose 13 percent for men and 12 percent for women, while life expectancy at age 65 rose 31 percent for men and 28 percent for women.

Although universally lauded as a great success story, the increase in life expectancy at older ages over the second half of the century has renewed important concerns about the existence of significant racial and ethnic differences in health. This country's progress on race-related issues is often measured by trends in major indicators of economic and social well-being, such as income or percentage of people in poverty, but few indicators offer more dramatic social commentary than the existence of large racial and ethnic differences in life expectancy. Blacks continue to experience much poorer health than whites, both before and after age 65, even though the black-white gap has narrowed over much of the century. The most recent data available from death certificates indicate that age-adjusted death rates for blacks are 33 percent higher than for whites (NCHS, 2003). On the other hand, age-adjusted death rates for other racial and ethnic minority groups are often lower than the comparable rate for whites, although there is much misunderstanding and misreporting on this point. Age-adjusted death rates for Hispanics (to the extent that they can be considered a discrete and identifiable segment of American society) are 22 percent lower than for non-Hispanic whites (NCHS, 2003), a surprise to many given their far lower average socioeconomic position and their generally poor level of health care coverage. Furthermore, the available data from death certificates suggest that, if taken at face value, both the American Indian and Alaskan Native populations and the Asian and Pacific Islander populations enjoy relatively lower age-adjusted death rates than non-Hispanic whites.

Undoubtedly some fraction of the minority advantage is attributable to measurement errors because death rates for both these minority groups are known to be underestimated (Rosenberg et al., 1999). Even so, when researchers have attempted to adjust the data or reestimate rates using other data sources, the relative ranking described rarely changes (Chapter 3, this volume). Within these various groups, there is much heterogeneity. Among Hispanics, Puerto Ricans generally experience relatively poorer health outcomes than Cubans and Mexican Americans, while among Asians and Pacific Islanders, Samoans and Native Hawaiians generally have worse health than other Asians. Further complicating the picture of relative health is the fact that, although it is generally true that blacks fare worse than other groups, the relative ordering of the other groups is inconsistent. For example, while most studies find that Hispanics fare worse with respect to such health outcomes as diabetes, infectious diseases, and chronic liver disease, they also find a Hispanic advantage for cardiovascular disease, cancer, and pulmonary diseases (Palloni and Arias, 2003).

To date, little research on racial and ethnic differences in health has been directed specifically toward the elderly, and there is still a need for great concern about broad health disadvantages of certain subpopulations, but particularly with regard to the situation of elderly blacks relative to whites. In 1950, there was no survival advantage of white men over black men above age 65. For women, it was only 0.2 of a year (NCHS, 2003). By 1970, it was 0.6 of a year for men and 1.4 years for women, and by 2000, the survival advantage of whites over blacks above age 65 was 2 years for men and 1.9 years for women (NCHS, 2003). This situation changed little over the 1990s.

A further reason for concern over racial and ethnic differences in health is that as a nation, the United States is becoming increasingly diverse. Currently, Hispanics, non-Hispanic blacks, Asians, and American Indians constitute 27 percent of the population, with blacks being the largest ethnic minority group. In 2003, Hispanics overtook non-Hispanic blacks and became the largest minority group. By 2050, half of the U.S. population will be either Hispanic, non-Hispanic black, Asian, or American Indian. In terms of the population aged 65 and over, the changes will be even more dramatic. Current projections suggest that by 2050, the total number of whites aged 65 and over will double. But, over the same period, the number of elderly blacks will more than triple and the number of elderly Hispanic and “other race” populations will increase 11-fold.

Responding to all these concerns, the National Institute on Aging (NIA) requested the National Research Council (NRC) to help guide its efforts to eliminate racial and ethnic differences in the health of older Americans. Specifically, the NIA asked the NRC to review research in this area and to identify research priorities that it might fund. Given that the debate about racial disparities in health has become highly political (see Vedantam, 2004), the case for clear and dispassionate scientific research could hardly be greater.

In addressing its charge, the NRC was forced to confront a large and burgeoning theoretical and empirical literature involving researchers in public health, medicine, and virtually all of the social and behavioral sciences. To make sense of this rapidly growing field, the NRC appointed an ad hoc panel, chaired by Norman Anderson, to prepare a report summarizing the main research lessons learned and offering recommendations to NIA for future directions for policy research and data collection in this important area. To help guide the panel, a series of background papers was commissioned. Revised versions of these papers are included as chapters in this volume.


Taken collectively, the chapters map out many of the major themes of interest to researchers and policy-makers concerned with racial and ethnic inequalities in health. The authors, all people at the forefront of research in their particular field, provide state-of-the-art assessments of the research in their area and identify major gaps in data, theory, and research design.

The papers represent a broad diversity of scholarly perspectives. Many different disciplines have made theoretical and empirical contributions to the study of health and the current collection is, to some extent, an amalgamation of concepts and insights—both new and old—obtained from various disciplines, each with its own domain of interest and style of analyzing and presenting data. Inevitably, the empirical basis for certain conclusions is stronger than for others. For example, while enough large high-quality longitudinal data sets are now available to be able to link socioeconomic resources confidently to observed racial and ethnic differences in health, few data have been collected on the cumulative effects over time of perceived racism on health (Chapter 9, this volume, Chapter 13, this volume). Thus, while some authors review research fields that employ mature methodologies and standard approaches, others report on new avenues of research that are still very much in their infancy. In these latter cases, concepts, methods, and measures still need to be refined. Nevertheless, each paper conveys important ideas that merit careful consideration.


Complicating our ability to study racial and ethnic differences is the fluid nature of the social construct of race, which needs to be understood in a particular social and historical context. Race is not a very meaningful biological categorization. Although it can be defined as phenotypic differences in skin color, hair texture, and other physical attributes, these are not (as often wrongly perceived) the surface manifestations of deeper underlying differences in attributes such as intelligence, temperament, or physical stature. Nevertheless, as we all know, race remains an extremely powerful predictor of life chances as well as social, psychological, and behavioral group differences.

In Chapter 2, Gary D. Sandefur, Mary E. Campbell, and Jennifer Eggerling-Boeck discuss some of the problems of defining and measuring racial and ethnic groups in the United States, how information on racial and ethnic composition has been collected at different points in time in some of the major health data sources, and how this has had important implications for our understanding of racial and ethnic differences in health among the elderly.

There are numerous examples in the study of race in the United States of racial categories and/or data collection efforts changing over time. For example, consider the case of the American Indian population. Since 1960, this population has grown principally as a result of the increased numbers of persons choosing to claim American Indian as their racial identity, as opposed to choosing white or some other race (Eschbach, 1993; Harris, 1994; Passel, 1976; Passel and Berman, 1986). A large proportion of those changing their reported identity from one Census to the next have tended to be of mixed race, which with high rates of intermarriage among American Indians, is likely to be an increasingly frequent occurrence. Interestingly, intermarriage has not only affected how people identify themselves, but in some cases it has led to tribes reconsidering how they define themselves (Thornton, 1996). These factors make it more problematic to compare results over time and across studies for American Indians than for other groups.

Perhaps the one recent policy change that is likely to have the greatest long-run implications is the recent directive of the Office of Management and Budget that allowed individuals to choose more than one racial category in the 2000 Census. Up until then, Americans could choose only one racial category to describe their race. Sandefur et al. examine the impact of this change. They conclude that allowing multiple responses to the “race” question in the 2000 Census has had only a modest impact on the measured racial composition of the U.S. population (see also Hirschman et al., 2000). However, this could easily change over time because it is quite possible that many blacks of mixed racial descent did not identify themselves as such in the 2000 Census because they never had the option to do so in the past (Korgen, 1999).

With these caveats in mind, Robert A. Hummer, Maureen R. Benjamins, and Richard G. Rogers (Chapter 3) summarize what is known about racial and ethnic differences in older health and mortality from large nationally representative data sets, and they discuss the extent to which the observed health differences correspond with differences in sociodemographic factors across population groups. Not surprisingly, the authors document that racial and ethnic differences in health, activity limitation, and active life persist well into later life, despite overall improvements in the general health of the U.S. population. The basic pattern is remarkably consistent over time and across data sources. Blacks have worse overall health across a number of indicators than white elders, while Hispanics and Asian American elders tend to fare better than whites. A major question revolves around the reliability of data on the health of Native American elderly. Note, however, that there is a great deal of heterogeneity in health outcomes within racial and ethnic categories, particularly among Hispanics (e.g., consider Puerto Ricans versus Cubans or Mexican Americans), Asian Americans, and American Indians. Hummer et al. also find that excess black mortality, relative to whites, is concentrated among the younger elderly population, with negligible differences beyond age 80. The authors show that education and income differences across groups continue to play an important role, explaining the overall worse health of non-Hispanic blacks, Native Americans, and, to a lesser degree, Hispanics in old age.

An ongoing debate remains about whether a true black-white crossover in mortality occurs among the oldest old. For many years analysis of mortality data has suggested that the black mortality curve crosses over the white mortality curve in later life so that, at the oldest ages, blacks have an advantage (see, e.g., Manton and Stallard, 1997; Nam et al., 1978). However, racial differences in death rates at older ages are especially vulnerable to distortion both because age misreporting is unusually common at these ages and because any distortions that occur are amplified by the severe slope of the age distribution itself (Coale and Kisker, 1986; Preston et al., 1996). And recent analysis by Preston and colleagues suggests that if there is a crossover, it is postponed to very high ages where data quality is most suspect (Preston et al., 2003).

Hummer et al. also compare the racial and ethnic differences among the elderly with those exhibited by younger age groups, and they find similar relative differences over the entire life course. For example, despite the fact that infant mortality rates have declined for all racial and ethnic groups over the past 50 years, large racial and ethnic differences remain. For a variety of reasons, including differences in underlying health status, socioeconomic circumstances, and the availability and use of health care, infant, neonatal, and postneonatal mortality rates for blacks are more than twice as high as for whites. Infant mortality for Hispanics is quite similar to whites, although under the Hispanic banner there is a degree of heterogeneity, with infant mortality rates being highest for Puerto Rican mothers and lowest for Cuban mothers. Similarly, although mortality rates among children and young adults also have declined over the past 50 years, large differences remain. Homicide and suicide rates, for example, vary by age, sex, and race. In 2000, homicide rates for black males aged 15 to 24 were 18 times as great as for non-Hispanic white males the same age (NCHS, 2003).

The second overview paper, by Jennifer J. Manly and Richard Mayeux (Chapter 4), examines in depth the ethnic differences in dementia and Alzheimer's disease. The major findings are that blacks and Hispanics have higher prevalence and incidence of cognitive impairment, dementia, and Alzheimer's disease than whites. American Indians have a lower rate of Alzheimer's disease than whites, but equivalent rates of overall cognitive impairment and dementia. The authors argue that because no research has been able to convincingly overcome the overwhelming influence of cultural and educational experience on cognitive test performance, the true extent of cross-cultural differences in cognitive impairment, dementia, or Alzheimer's remains an open question. Not surprisingly, therefore, researchers are still a long way from understanding how, controlling for education, observed differences can be explained by biological risk factors such as cerebrovascular disease, differential exposure to environmental risk factors, or genetic risk factors (Chapter 4, this volume). Some intriguing recent cross-national evidence suggests that racial and ethnic differences in rates of dementia may be the result of complex gene-environmental interactions. It remains to be seen, however, whether researchers will be able to disentangle biological and genetic risk factors from their sociocultural or environment context (Chapter 4, this volume).


The papers by Clyde Hertzman (Chapter 5), Alberto Palloni and Douglas C. Ewbank (Chapter 6), and Guillermina Jasso and colleagues (Chapter 7) highlight important analytical and methodological insights that have emerged from research in this area over the past couple of decades.

In Chapter 5, Hertzman argues the need for a life-course approach to the study of health. Events over one's life can have a long-run and cumulative impact on one's health, affecting a diverse range of outcomes from general well-being to physical functioning and chronic disease. For example, studies have linked fetal and early child nutrition to a wide range of disease outcomes in later life, including heart disease, diabetes, obesity, high blood pressure, age-related memory loss, and schizophrenia (see Barker, 1998). Similarly, findings related to the nature of racial and ethnic differences in health and mortality in older ages need to be interpreted within a life-course framework. Hertzman sets out various mechanisms through which early experiences can affect adult health status, distinguishing among latency, pathway, and cumulative effects. Hertzman's chapter implies an urgent need for much better data in order to understand how various life-course factors (e.g., socioeconomic position over the life cycle, family history, migration history, work history, cumulative stress, child and early adult health experience) exhibit different general patterns across racial and ethnic groups, thereby contributing to differential impacts on older adult health by race or ethnic group.

The next papers, by Palloni and Ewbank (Chapter 6) and Jasso et al. (Chapter 7), focus on the conceptual and methodological challenges that arise in the study of observed racial and ethnic differences in health from the potential existence of certain types of selection processes. Generally speaking, standard regression coefficients that are obtained from data drawn from nonrandom samples are not of direct interest because they may exaggerate or attenuate estimates of effects of membership in a particular race or ethnic group that occur due to the existence of other mechanisms. For example, in the classic case of sample selection bias, standard regression coefficients confound meaningful structural parameters with the parameters of a function that determines the probability that the observation makes its way into the nonrandom sample (Heckman, 1979). Perhaps the best known and simplest example of a problem analogous to that identified by Heckman operating in demography relates to the issue of differential survival among population subgroups, which can produce mortality crossovers. If one subgroup is disadvantaged at an early age, then heterogeneity implies that its members will appear to be less disadvantaged or even advantaged at older ages (Vaupel et al., 1979). Palloni and Ewbank demonstrate that selection effects can, in some cases at least, be quite large, and if ignored could lead to misinterpretation of observable data and to erroneous policy prescriptions.

Palloni and Ewbank's general conclusions regarding the value of controlling for selection are generally reinforced by the paper by Jasso et al., who use the particularly salient example of immigrant health to discuss the dangers of using censored (i.e., nonrandomly selected) samples to estimate behavioral relationships. With immigration a driving force in accounting for the future growth of the American population, scholarly and policy-related interest in immigration and health, and how the two are related, is perhaps greater today than ever before. A recurrent finding in the immigrant health literature is that Mexican and non-Mexican Hispanics experience better health, lower adult mortality rates, and lower infant mortality than African Americans and non-Hispanic whites and that the health of immigrants appears to deteriorate with duration of stay in the United States (Chapter 6, this volume). Various explanations have been put forward as to why this might be the case. These include the possibility that migrants come from a cultural milieu that promotes more favorable behavioral profiles in terms of diet, smoking, and alcohol consumption, and more cohesive social networks and social support mechanisms (see Chapter 6, this volume). However, it is also possible that immigrants are healthier because of positive selection of migration for health (or for other endowments or traits that are correlated with health). In other words, their initially superior health status preceded and facilitated their immigration.

Drawing on their survey of legal immigrants who obtained their green cards in 1996, Jasso et al. discuss the problems of using cross-sectional data to investigate these issues. The authors provide strong empirical support to Palloni and Ewbank's concerns regarding the importance of accounting for selectivity bias in interpreting data on racial and ethnic differences in health. They argue persuasively that controlling for health selection (i.e., the propensity of immigrants to be healthier than a representative person in the sending country) is critical to understanding patterns of immigrant health. The authors also develop a theoretical model that attempts to explain the diversity in health selection among immigrants. This work provides a sounder theoretical grounding for research on immigration and health than has previously been available.


Over the past 20 years, papers on racial and ethnic differences in health have shifted away from straightforward descriptive studies toward more analytical studies that attempt to explain the underlying causes and mechanisms behind these differences. The next nine papers in the volume attempt to summarize scientific knowledge about the main factors that contribute to the existence of racial and ethnic differences in health in later life, as described earlier. As many of the authors point out in their chapters, the search for evidence on causal factors has been a stubborn problem, fraught with methodological and conceptual difficulties.

Each of the nine chapters in this section of the volume focuses on a particular factor or set of factors, which often parallel the interest(s) of a particular scientific discipline. Group differences in health are the product of both biology and individual choice, the former modified in some cases by environmentally sensitive gene expression and the latter strongly influenced by economic, social, and cultural conditions. While the volume itself aspires to be comprehensive in its coverage of the major health risks, individual papers introduce concepts and insights from quite different disciplines, not all of which have been fully integrated. Thus, individual chapters can only provide a general treatment of a particular set of relationships, detailing the research evidence in one particular area without attempting to integrate these observations into a comprehensive theory or framework, which, in any event, is still far beyond the current capacities of the field (National Research Council, 2004).

Genetic Factors

Probably no aspect of the debate about the causes of racial differences in health is potentially more sensitive than the discussion about the extent to which genetic factors are in any way responsible. There are numerous historical examples of scientific mischief in the support of racism. The discovery in 1851 of drapetomania (a mental disease that caused slaves to run away from plantations) is only one of a series of misguided attempts over the 19th and early 20th centuries to distinguish biological differences between the races. Yet recent research on genetic variation within human populations has reopened the debate over the validity of race as a sensible research variable (Sankar and Cho, 2002). Proponents of using race assert that there is a useful degree of association between genetic differences and racial classifications, so that the use of race as a research variable is warranted. Critics, on the other hand, argue that bundling the population into four or five broad groups according to skin color and other physical traits is not a useful way of summarizing genetic variation when we know now that there are at least 15 million genetic polymorphisms in humans, of which an unknown number underlie variation in (normal and) disease traits (Burchard et al., 2003). Hence critics argue that there is as much or more genetic variation within the major population groups than between them.

The gene pools of different racial or ethnic groups may contain different frequencies of alleles at different loci that are pertinent to health status or to disease processes. In Chapter 8, Richard S. Cooper explains that there are two ways that genes may be relevant to the study of health differences. First, genetic differences in disease among racial and ethnic groups are particularly relevant for simple genetic disorders caused by a single gene mutation in populations that descend from a relatively small number of founders and that remain endogamous for a large part of their history. An example is the prevalence of Tay-Sachs disease among Ashkenazi Jews. However, such single-gene disorders are generally rare and scientists are only just beginning to confront the challenge of understanding the genetic basis of complex genetic disorders such as asthma, cancer, and diabetes, which likely involve multiple, potentially interacting, genes and environmental factors (see also Burchard et al., 2003).

This leads to the second way that genes may be relevant to the study of health differences, namely through environmental factors, which may vary by ethnic group, and which might interact with genotype to produce difference outcomes. (Note that environment in this context is defined as all influences not coded in DNA.) Significant resources are now being devoted to elucidating further the importance of gene-environment interactions in a wide range of diseases. Although the volume of research in this area is growing rapidly, it is still in the early stages. Cooper is perhaps more pessimistic than some about the prospect of disentangling the role of genetic factors in explaining racial differences in health in the near future. But, given that most of the major diseases differing in frequency among the standard racial classifications appear to be diseases of complex etiology, involving a complex genetic basis and a strong environmental component influencing how this inherited susceptibility is expressed, and given that the environmental milieu of different racial and ethnic groups is quite different in many important respects, the challenge of disentangling the role of genetic factors from other possible explanatory variables appears hard to overstate (Neel, 1997). Nevertheless, the potential for further study is increasing greatly, with growing numbers of large population-based social and behavioral studies also collecting clinical and genetic data (Kington and Nickens, 2001).

Economic and Social Risk Factors

Chapters 9, 10, and 11 review the extent to which racial and ethnic differences in economic, social, and personal resources contribute to racial and ethnic differences in the health and well-being of older adults.

In Chapter 9, Eileen M. Crimmins, Mark D. Hayward, and Teresa E. Seeman review the complex interactions linking race, socioeconomic status, and health. The United States ranks first among industrialized countries with respect to measures of inequality of both income and wealth (Wilkinson, 1996; Wolff, 1996). This is important not only from a social justice perspective, but also because studies in recent decades have demonstrated a strong and persistent inverse gradient between mortality and socioeconomic status, even among those near the top of the socioeconomic distribution (Marmot et al., 1991).

Controlling for socioeconomic status eliminates a significant proportion—but not all—of the observed racial differences in chronic health between blacks and non-Hispanic whites, but not between Hispanics and non-Hispanic whites (Hayward et al., 2000). Nevertheless, for a number of reasons, it is extremely difficult to quantify succinctly the nature of the relationship between socioeconomic resources and health outcomes. First, the meaning and measurement of socioeconomic status has varied a fair amount across studies, and findings from different studies are not always directly comparable. Second, some authors have found the relationship between socioeconomic status and health to be decidedly nonlinear, while others have observed an essentially linear relationship. Third, no data set has truly come to grips with the cumulative and dynamic nature of the relationship over the entire lifecycle. Furthermore, there is growing evidence that conditions in early life or even in utero can profoundly influence one's risk for certain chronic conditions much later in life (Barker, 1998; Elo and Preston, 1992). Finally, although it has been well established that variations in socioeconomic status produce health differences, there is now an increasing appreciation of the need to recognize the alternative casual pathway, namely that among the elderly poor health can also lead to significantly reduced wealth (Smith, 1999). Recognizing the bidirectional nature of the relationship makes it difficult to interpret results from cross-sectional data. Finally, there is still a great deal of ambiguity surrounding the exact mechanisms by which socioeconomic inequality influences health (Deaton, 2001).

Using data from a number of major health surveys of the U.S. population, Crimmins et al. investigate how socioeconomic status—as indicated by educational attainment, family income, and wealth—varies across race and ethnicity. The authors find that the prevalence of diseases, functioning loss, and disability are all negatively related to socioeconomic status (i.e., lower socioeconomic status implies more health problems). The authors also examined the variability in disease and disability prevalence by socioeconomic status within racial and ethnic groups to determine whether the relationships were the same for all ethnic groups. They found that the strongest socioeconomic effects are within the white group.

Undoubtedly, the effects of economic resources on health and well-being are determined to some extent by the sociocultural environment. Building on the chapter by Crimmins et al., Carlos Mendes de Leon and Thomas A. Glass (Chapter 10) review the evidence on the effects of social and other personal resources on health and well-being. Evidence is accumulating to suggest that social and personal resources also play an important role in understanding differences in health and well-being among older adults. Social resources—which can include factors such as the strength of social and community networks, the level of social engagement, and the degree of participation in formal and informal organizations—have been linked with prolonged survival and decreased risk of age-related physical and cognitive disabilities (Chapter 10, this volume). But there is little evidence to suggest that there are substantial differences by race and ethnicity in either social or personal resources among the elderly. While older blacks tend to have similarly sized or slightly smaller social networks than older whites, these networks are more likely to include extended family members (Chapter 10, this volume). Similarly, older whites tend to be more engaged than older blacks with volunteer activities through formal and informal organizations, while older blacks tend to be more involved than older whites in faith-based organizations. Hence the authors conclude that there is little evidence to suggest that differences in social resources play a major role in producing the observed racial and ethnic differences in health in later life. The authors speculate that there may be more important racial and ethnic differences in community networks, given the stark differences in neighborhood characteristics in which different racial and ethnic groups live.

The challenge of reviewing the literature on the effects of neighborhoods and residential contexts on health among the elderly is taken up in Chapter 11 by Jeffrey D. Morenoff and John W. Lynch. Communities and neighborhoods are potentially important levels of aggregation for designing interventions aimed at improving health, yet few studies exist that focus specifically on the linkages between neighborhoods and elderly health. Consequently, the authors are forced to extend their review far beyond direct consideration of the effects of neighborhood influences on the elderly. A considerable body of research on the neighborhood context of health now exists and shows that context matters. Multilevel studies consistently show that poor neighborhoods with concentrated poverty are associated with significantly elevated risks of poor health and overall mortality, even after controlling for individual differences in household income. Unfortunately, however, different investigators have operationalized the concept of neighborhood in different ways, making it difficult to reach definite conclusions about the magnitude and importance of these kinds of effects (Chapter 11, this volume). Studies on neighborhood effects now need to move away from the descriptive to focus more on the question of why context matters for health. With this in mind, Morenoff and Lynch advance the notion of a life-course perspective that emphasizes the importance of cumulative effects. This implies that cross-sectional data would likely underestimate the true strength of neighborhood effects and that more attention needs to be focused on a multilevel longitudinal approach that can incorporate information on neighborhood characteristics over a protracted period.

Behavioral Risk Factors

An enormous amount of research over the past 20 years or so has confirmed the link between certain diseases and health outcomes and various health-damaging and health-promoting behaviors. Smoking, for example, is now known to be a major risk factor for several forms of cancer, chronic bronchitis, emphysema, and cardiovascular disease, while alcoholism is an important risk factor for numerous health outcomes, including cirrhosis of the liver and pancreatitis (U.S. Department of Health and Human Services, 1991). Alcohol is also a factor in approximately half of all homicides, suicides, and motor vehicle fatalities. Similarly, regular physical activity and correct nutrition have been shown to lower one's overall risk of mortality as well as being linked to the risk of certain diseases such as cardiovascular disease, non-insulin-dependent diabetes, osteoarthritis, and depression (U.S. Department of Health and Human Services, 1991).

The paper by Marilyn A. Winkleby and Catherine Cubbin (Chapter 12) explores the extent to which these and other health-damaging or health-promoting behaviors vary by racial and ethnic group. Few studies have examined racial and ethnic differences in health behavior among the elderly, especially using large-scale, nationally representative samples. Drawing on data from the 2000 Behavioral Risk Factor Surveillance System, the authors compare rates of smoking, obesity, leisure-time physical activity, diet, alcohol consumption, and cancer screening practices by racial and ethnic group, age, sex, and sociodemographic status. The differences between racial and ethnic groups are far from consistent. By some indicators, non-Hispanic whites behave in a more healthy manner than non-Hispanic blacks and/or Hispanics, but for other indicators, the opposite it true. For example, for women aged 65 to 74, non-Hispanic whites tend to smoke more than blacks or Hispanics. But for comparably aged men, blacks are the heaviest smokers. Similarly, elderly black women are less likely to engage in leisure-time physical activity and more likely to be obese than non-Hispanic whites or Hispanics, while elderly Hispanic men report significantly lower rates of leisure-time physical activity than non-Hispanic whites or blacks. Significantly, the authors find that these and similar differences in health-promoting or health-damaging behavior exist within each racial and ethnic group by important sociodemographic indicators, including age, sex, educational attainment, household income, and, for Mexican Americans, country of birth and language spoken. These differences clearly have important implications for prevention and public health (see Chapter 17, this volume).

Biobehavioral Factors

The next three chapters, by Hector F. Myers and Wei-Chin Hwang (Chapter 13), Rodney Clark (Chapter 14), and Julian F. Thayer and Bruce H. Friedman (Chapter 15), all explore various aspects of the relationships among stress, psychosocial risk and resilience, and racial and ethnic differences in health. Recent conceptual advances in health research have stressed the importance of an integrative approach that incorporates the interplay among genetic, behavioral, psychosocial, and environment factors over time (National Research Council, 2001). For many years, the study of stress has been considered a potentially important pathway connecting a person's experience, living and working conditions, interpersonal relations, and other behavioral variables to biological factors that more directly influence health. Stress is a known risk factor for hypertension, and the significantly higher prevalence of hypertension in blacks has led some researchers to theorize that there may be an important link between a negative psychological environment, cumulative stress, and hypertension and some of the observed racial differences in health.

Important new attempts to understand the relationship between environmental and behavioral challenges and stressors, health, and disease have introduced the concepts of allostasis and allostatic load (see Chapter 13, this volume). Stress is actually just part of the body's normal regulatory system. Stress causes and modulates a diversity of physiological effects that can either directly enhance resistance to disease or cause damage, thereby making the body more susceptible to disease (Institute of Medicine, 2001). But it is the repeated wear and tear on the body as a function of repeated insults that can lead to negative health effects and which is referred to as a body's allostatic load. This concept has received a lot of attention in medical circles (see, e.g., McEwen, 1998). In Chapter 13, Myers and Hwang outline a biopsychosocial model of cumulative psychological and physical vulnerability and resilience in later life, in which chronic stress burden and psychosocial resources for coping are hypothesized as playing a significant role in accounting for ethnic differences in mental health.

The central hypothesis in these biobehavioral models is that the differential burden of lifetime stress contributes to ethnic differences in health. Chapters 13 and 14 review some of the major potential sources of stress, including race, social class, immigrant acculturation, and age. Other researchers have emphasized the importance of looking at other potential sources of stress such as the degree of job control in the workplace (see, e.g., Marmot et al., 1997). These chapters are necessarily somewhat speculative because very little research has attempted to model or estimate the cumulative effects of various sources of stress. This is partly because of problems of measurement, concepts, and definition that need to be overcome. It is also partly because modeling and estimating dynamic processes over time are extremely complex and data-hungry endeavors. One of the methodological challenges of the work in this area is that much of it is based on subjective measures of stress (i.e., respondents reporting that they are feeling or felt stressed) rather than on measuring objective stressors such as particular life events (e.g., job loss, divorce). In addition, there is always the possibility that the causality runs in the other direction, namely that poor health status leads to higher perceived stress (Kington and Nickens, 2001).

Recently, researchers have turned their attention to investigating the impact of repeated encounters with racism or discrimination on physiologic activity (see, e.g., February 2003 special issue of the American Journal of Public Health as well as Chapter 14, this volume). Studies that attempt to relate physical health status to self-reported encounters with racism have been inconsistent (Chapter 14, this volume), but there is a growing body of research from community studies that discrimination is associated with multiple indicators of poorer physical and, especially, mental health status (Williams et al., 2003). Nevertheless, research on the health impact of racism and discrimination is still very much in its infancy, and difficult methodological challenges and controversies need to be overcome (Krieger, 1999). Racism and discrimination are subjective measures that are arguably far harder to measure than, say, socioeconomic status. The problem is even more complex once one accepts that the effects of discrimination can accumulate over a lifetime and even spill over into subsequent generations (National Research Council, 1989). Furthermore, racism and discrimination are not measured directly in large social and economic surveys. Hence data tend to come from laboratory research in controlled clinical settings. Clark (Chapter 14) suggests several ways to strengthen research in this area.

Another important avenue of research in this area revolves around the observation that people differ widely in their ability to deal with particular situations, including overcoming adversity. A growing body of research indicates that a number of psychosocial factors, including personality, resilience, coping style, early life experience, availability of social supports, and religiosity can serve to moderate the stress-health relationship in adults, including the elderly (Chapter 13, this volume). But, although some researchers have proposed that there may be important racial and ethnic differences in coping styles that relate to health outcomes (see, e.g., James, 1994, on John Henryism), few high-quality studies in this area have been conducted (Chapter 13, this volume).

In the next chapter, Thayer and Friedman collect recent findings that relate psychosocial factors such as stress and perceived racism to the functioning of the autonomic nervous system, which regulates all aspects of cardiovascular function. Our understanding of the multilevel and bidirectional relationships between behavioral and biologic processes has been enriched in recent years by advances in technology and by conceptual advances in the behavioral, biological, and medical sciences (Institute of Medicine, 2001). Recent research has uncovered many previously unknown links among the central nervous system, the endocrine system, and the immune system (Institute of Medicine, 2001). Thayer and Friedman's chapter reemphasizes the need for multilevel studies that address the interplay among physiological, behavioral, affective, cognitive, and social processes and their shared impact on health.

Access to Health Care

The chapter by Amitabh Chandra and Jonathan S. Skinner (Chapter 16) examines the extent to which the spatial distribution of health care and health outcomes across the United States varies by hospital referral region. Differential health care access and quality is a major public policy challenge. The United States is the only developed country in the world that does not have national health coverage, and 40 million Americans do not have any form of health care coverage (Institute of Medicine, 2001). Hispanics in particular are relatively underserved with respect to health coverage, partly because of their relatively low socioeconomic status and partly due to other contributing factors related to their degree of acculturation, language barriers, immigration status, and types of jobs in which they are engaged (Suárez, 2000). How much differential coverage contributes to racial and ethnic differences in health outcomes is unclear, however, because Hispanics do not appear to fare any worse than non-Hispanic whites with respect to a number of health outcomes. Infant mortality rates for Hispanics, for example, are essentially identical to non-Hispanic whites and have been for more than a decade (NCHS, 2003).

Chandra and Skinner argue that the correlation between minority status and health outcomes is confounded by differential access to medical services, specifically by substantial geographic variation in treatment and outcome patterns. Minorities tend to seek care from different hospitals and from different physicians than non-Hispanic whites, in large part a reflection of the general spatial distribution of the United States population with concentrations of minorities in certain hospital referral regions. Chandra and Skinner demonstrate that regional variation in the utilization of health care, and in outcomes, potentially can account for a substantial part of the observed racial and ethnic disparities in health. This implies a different set of policy prescriptions than if the underlying source of racial differences in health were primarily due to differences in treatment within hospitals or communities or differences in the self-management of disease (Goldman and Smith, 2002; Institute of Medicine, 2003).


What can be done about the existence of large and persistent differences in health outcomes by racial and ethnic group that essentially cannot be fully explained by traditional arguments? One starting point is the observation that approximately half of all deaths in the United States can be traced to various health-damaging behaviors, especially tobacco use, poor diet, low activity patterns, and excessive alcohol consumption (McGinnis and Foege, 1993). To the extent that racial differences in health can be attributed to differences in group behavior (see Chapter 12, this volume), then it may be possible to develop programs that can educate and encourage people to modify their behavior. Although there are numerous examples of successful public awareness campaigns, the real challenge has always been to get people to convert new knowledge into sustained behavioral change. Most people, for example, know that if they change certain behaviors (e.g., exercise more, eat and drink more responsibly, avoid cigarettes), they would reduce their risk of certain diseases. But the $30 to $50 billion spent annually on various types of diet products testifies to the self-control problems that many people face (Cutler et al., 2003). In Chapter 17, David M. Cutler reviews the evidence on the effectiveness of large-scale behavioral health interventions that have been implemented at the individual, community, and national levels. Behavioral change is notoriously difficult to achieve and overall there is not very compelling evidence that large-scale interventions have worked. Although there have been studies that show that effective change is possible, by and large interventions have had smaller effects than were originally anticipated (Chapter 17, this volume). There are, however, some important exceptions to this generalization, including the 1964 report of the Surgeon General on the harmful effects of smoking and the Mothers Against Drunk Driving campaign that began in the early 1980s. Cutler notes that in both of these cases, the interventions occurred at the national level, the message conveyed was simple and straightforward, and the behavioral change required was clear. In general, however, more research is needed on why some health interventions succeed while others fail (Chapter 17, this volume).


The United States is not the only country in the world concerned about racial equity in health. In the final two papers in the volume, James Y. Nazroo (Chapter 18) and Debbie Bradshaw and colleagues (Chapter 19) offer international comparative perspectives on these issues from the United Kingdom and South Africa, respectively. Although both papers document slightly different patterns of racial and ethnic disparities in health than occur in the United States, they both suggest the centrality of particular casual factors including socioeconomic status, culture, racism, and, for immigrants, generation (i.e., first versus second or third) and period of migration (i.e., length of stay in country).

In Chapter 18, Nazroo investigates the relationship between age and ethnic inequalities in health in the United Kingdom, in relation to socioeconomic circumstances, migration status, extent of racial harassment, duration of stay, and cohort. Given the complexity of factors involved, the author finds it difficult to come to any firm conclusions. A particularly vexing problem is that although most minorities in the United Kingdom have arrived since the Second World War, different ethnic subgroups have tended to enter the country in waves, making it extremely difficult to untangle the effects of age, period, and cohort on health. As Nazroo points out, this underscores the utility of collecting and analyzing these issues using longitudinal panel data.

In Chapter 19, Bradshaw and colleagues review recent data and analyses on the nature of racial disparities in health in South Africa, a country that has witnessed extraordinary political and social change over the past 10 to 15 years as the government has attempted to overcome the legacy of apartheid. Again the complexity of the issues and the lack of long-term data prevent the authors drawing firm conclusions about how these profound economic and social changes are affecting racial differences in health by age. Nevertheless, the available evidence suggests that socioeconomic and cultural factors as well as health behaviors are important determinants of health in South Africa.

The Way Ahead

The purpose of this chapter is not to outline an agenda for future research in this area. That task has already been undertaken by the NRC's Panel on Race, Ethnicity, and Health in Later Life. Their conclusions and recommendations are presented in a companion volume (see National Research Council, 2004). Nevertheless, taken together, the chapters in this volume highlight both the strengths and weaknesses of the existing knowledge base. Perhaps the main message is that after decades of research in this area, there are still large and persistent differences in health by racial and ethnic groups that cannot be explained fully by traditional arguments such as differences in socioeconomic status, access to health care, or health behaviors.

Clearly, then, there is still a need to continue to monitor changes in the health status of different racial and ethnic groups and to understand the causal factors underlying these differences. Even if there is no way to weigh their relative importance, there is fairly broad agreement among the disciplines that the list of major causes is fairly self-contained: socioeconomic status, education, health risk behavior, psychosocial factors including stress, access to and quality of health care, culture, genetic factors, and environmental and occupational risk factors (Kington and Nickens, 2001). Because of the interaction and the multiple causal pathways between these various factors (e.g., low socioeconomic status leads to poor health, but poor health can also lead to less wealth), the exact contribution that each factor contributes to the observed health differences by race remains unknown. Further research, particularly with emphasis on the integration of data from various disciplines, is therefore essential.

A full understanding and complete consensus on the reasons underlying these observed differences will likely continue to elude us for some time. Nevertheless, the papers in this volume indicate that high-quality research is under way in many disciplines. Increasingly, this research points to the need for a more integrative approach to health that accounts for the interactions among genetic, behavioral, psychosocial, and environmental factors over time. The chapters in this volume represent a useful step forward, but it is only one among many that will ultimately need to be undertaken before sustainable progress in our understanding of the causes of racial and ethnic differences in health in later life can be made. My hope is that these papers will be useful to those charged with making and implementing public policy as well as to scholars from different disciplines wishing to build on this foundation.


I am grateful to Richard S. Cooper, Eileen M. Crimmins, Clyde Hertzman, Charles B. Keely, Faith Mitchell, Alberto Palloni, Jane Ross, James P. Smith, and Marilyn A. Winkleby for their comments on an earlier draft.


  1. Barker DJP. Mothers, babies and health in later life. 2nd ed. Edinburgh, Scotland: Churchill Livingstone; 1998.
  2. Burchard EG, Ziv E, Coyle N, Gomez SL, Tang H, Karter AJ, Mountain JL, Pérez-Stable EJ, Sheppard D, Risch N. The importance of race and ethnic background in biomedical research and clinical practice. New England Journal of Medicine. 2003;348(12):1170–1175. [PubMed: 12646676]
  3. Coale AJ, Kisker EE. Mortality crossovers. Reality or bad data? Population Studies. 1986;40(3):389–401.
  4. Cutler DM, Glaeser EL, Shapiro JM. Why have Americans become more obese? Cambridge, MA: National Bureau of Economic Research; 2003. (NBER Working Paper No. 9446)
  5. Deaton A. Health, inequality, and economic development. Cambridge, MA: National Bureau of Economic Research; 2001.
  6. Elo IT, Preston SH. Effects of early-life conditions on adult mortality: A review. Population Index. 1992;58(2):186–212. [PubMed: 12285320]
  7. Eschbach K. Changing identification among American Indians and Alaska Natives. Demography. 1993;30(4):635–652. [PubMed: 8262285]
  8. Fried LP. Epidemiology of aging. Epidemiologic Reviews. 2000;22(1):95–106. [PubMed: 10939013]
  9. Goldman DP, Smith JP. Can patient self-management help explain the SES health gradient? Proceedings of the National Academy of Sciences. 2002;99(16):10929–10934. [PMC free article: PMC125075] [PubMed: 12140364]
  10. Harris D. The 1990 census count of American Indians: What do the numbers really mean? Social Science Quarterly. 1994;75(3):580–593.
  11. Hayward MD, Crimmins EM, Miles TP, Yu Y. The significance of socioeconomic status in explaining the racial gap in chronic health conditions. American Sociological Reviews. 2000;65(6):910–930.
  12. Heckman JJ. Sample selection bias as a specification error. Econometrica. 1979;47(1):153–161.
  13. Hirschman C, Alba R, Farley R. The meaning and measurement of race in the U.S. Census: Glimpses into the future. Demography. 2000;37(3):381–393. [PubMed: 10953811]
  14. Institute of Medicine. Coverage matters: Insurance and health care. Washington, DC: National Academy Press; 2001. Committee on the Consequences of Uninsurance.
  15. Institute of Medicine. Unequal treatment: Confronting racial and ethnic disparities in health care. Smedley BD, Stith AY, Nelson AR, editors. Washington, DC: The National Academies Press; 2003. Committee on Understanding and Eliminating Racial and Ethnic Disparities in Health Care.
  16. James SA. John Henryism and the health of African Americans. Culture, Medicine, and Psychiatry. 1994;18(2):163–182. [PubMed: 7924399]
  17. Kington RS, Nickens HW. Racial and ethnic differences in health: Recent trends, current patterns, and future directions. In: Smelser NJ, Wilson WJ, Mitchell F, editors. America becoming: Racial trends and their consequences. II. Washington, DC: National Academy Press; 2001. pp. 253–310.
  18. Korgen KO. From black to biracial: Transforming racial identity among Americans. Westport, CT: Praeger; 1999.
  19. Krieger N. Embodying inequality: A review of concepts, measures, and methods for studying health consequences of discrimination. International Journal of Health Services. 1999;29(2):295–352. [PubMed: 10379455]
  20. Manton KG, Stallard E. Health and disability differences among racial and ethnic groups. In: Martin LG, Soldo BJ, editors. Racial and ethnic differences in the health of older Americans. Washington, DC: National Academy Press; 1997. pp. 43–105. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council.
  21. Marmot MG, Davey-Smith G, Stansfeld S, Patel C, North F, Head J, White I, Brunner E, Feeny A. Health inequalities among British civil servants: The Whitehall II Study. Lancet. 1991;337:1387–1393. [PubMed: 1674771]
  22. Marmot MG, Bosma H, Hemingway H, Brunner E, Stansfeld SA. Contribution of job control and other risk factors to social variations in coronary heart disease incidence. Lancet. 1997;350:235–239. [PubMed: 9242799]
  23. McEwen BS. Protective and damaging effects of stress mediators. New England Journal of Medicine. 1998;338(3):171–179. [PubMed: 9428819]
  24. McGinnis JM, Foege WH. Actual causes of death in the United States. Journal of American Medical Association. 1993;270(18):2207–2212. [PubMed: 8411605]
  25. Nam CB, Weatherby NL, Ockay KA. Causes of death which contribute to the mortality crossover effect. Social Biology. 1978;25:306–314. [PubMed: 757307]
  26. National Center for Health Statistics. Health, United States, 2003. Hyattsville, MD: Author; 2003.
  27. National Research Council. A common destiny: Blacks and American society. Jaynes GD, Williams RM, editors. Washington DC: National Academy Press; 1989. Committee on the Status of Black Americans.
  28. National Research Council. New horizons in health: An integrative approach. Singer BH, Ruff CD, editors. Washington, DC: National Academy Press; 2001. Committee on Future Directions for Behavioral and Social Sciences Research at the National Institutes of Health.
  29. National Research Council. Panel on Race, Ethnicity, and Health in Later Life. In: Bulatao RA, Anderson NB, editors. Understanding racial and ethnic differences in health in late life: A research agenda. Washington, DC: The National Academies Press; 2004. Committee on Population, Division of Behavioral and Social Sciences and Education. [PubMed: 20669466]
  30. Neel JV. Are genetic factors involved in racial and ethnic differences in late-life health? In: Martin LG, Soldo BJ, editors. Racial and ethnic differences in the health of older Americans. Washington, DC: National Academy Press; 1997. pp. 210–232. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council.
  31. Palloni A, Arias E. A re-examination of the Hispanic mortality paradox; Paper presented at the meeting of the Population Association of America; Minneapolis, Minnesota. May 1-3.2003.
  32. Passel JS. Provisional evaluation of the 1970 Census count of American Indians. Demography. 1976;13(3):397–409. [PubMed: 955175]
  33. Passel JS, Berman PA. Quality of 1980 Census data for American Indians. Social Biology. 1986;33:163–182. [PubMed: 3563542]
  34. Preston SH, Elo IT, Rosenwaike I, Hill M. African American mortality at older ages: Results of a matching study. Demography. 1996;33(2):193–209. [PubMed: 8827165]
  35. Preston SH, Elo IT, Hill. ME, Rosenwaike I. The demography of African Americans, 1930-1990. Dordrecht, The Netherlands: Kluwer Academic; 2003.
  36. Rosenberg HM, Maurer JD, Sorlie PD, Johnson NJ, MacDorman MF, Hoyert DL, Spitler JF, Scott C. Quality of death rates by race and Hispanic origin: A summary of current research. Hyattsville, MD: U.S. Department of Health and Human Services; 1999. (Vital Health Statistics, Series 2. Data Evaluation and Methods Research. No. 128) [PubMed: 10611854]
  37. Sankar P, Cho MK. Towards a new vocabulary of human genetic variation. Science. 2002;298:1337–1338. [PMC free article: PMC2271140] [PubMed: 12434037]
  38. Smith JP. Healthy bodies and thick wallets: The dual relation between health and economic status. Journal of Economic Perspectives. 1999;13(2):145–166. [PMC free article: PMC3697076] [PubMed: 15179962]
  39. Suárez ZE. Hispanics and health care. In: Cafferty PSJ, Engstrom DW, editors. Hispanics in the United States. New Brunswick, NJ: Transaction; 2000. pp. 195–235.
  40. Thornton R. Tribal membership requirements and the demography of ‘old' and ‘new' Native Americans. In: Sandefur GD, Rindfuss RR, Cohen B, editors. Changing numbers, changing needs: American Indian demography and public health. Washington, DC: National Academy Press; 1996. pp. 103–112.
  41. U.S. Department of Health and Human Services. Healthy people 2000: National health promotion and disease prevention objectives. Washington, DC: Author; 1991.
  42. Vaupel JW, Manton KG, Stallard E. The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography. 1979;16(3):439–454. [PubMed: 510638]
  43. Vedantam S. Racial disparities played down. The Washington Post. 2004 January 14:A17.
  44. Wilkinson RG. Unhealthy societies: The afflictions of inequality. New York: Routledge; 1996.
  45. Williams DR, Neighbors HW, Jackson JS. Racial/ethnic discrimination and health: Findings from community studies. American Journal of Public Health. 2003;93(2):200–208. [PMC free article: PMC1447717] [PubMed: 12554570]
  46. Wolff EN. Top heavy: The increasing inequality of wealth in America and what can be done about it. New York: New Press; 1996.
Copyright © 2004, National Academy of Sciences.
Bookshelf ID: NBK25512


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (9.9M)

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...