The Limited but Insufficiently Understood Role of Medical Care

Several types of evidence point to the limited role of medical care in understanding how and why socioeconomic position affects health. First, there is evidence that modern preventive and therapeutic medical care can account for only a minor fraction of the dramatic improvements in individual and population health over the last 250 years (McKeown, 1976, 1979, 1988; McKinlay and McKinlay, 1977). Even analysts admiring of the impact of medical science on health, for example, estimate that only about 5 years of the 30-year increase in life expectancy in the United States in the twentieth century has been due to preventive or therapeutic medical care (Bunker et al., 1994). The remainder is attributable primarily to increasing socioeconomic development and associated gains in nutrition, public health and sanitation, and living conditions. Second, improvements in access to medical care occasioned by the introduction of national health insurance or service plans have, quite unexpectedly, done little or nothing to reduce socioeconomic differences in health. The rediscovery of the importance of socioeconomic disparities in health as a major public health problem was probably stimulated most by the publication in England in 1980 of the Report of the Working Group on Inequalities in Health, better known as the Black Report after the chair of the working group, Sir Douglas Black, then chief scientist of the U.K. Department of Health and subsequently president of the Royal College of Physicians. The report showed that occupational class differences in health were greater than differences by gender, race, or regional background and, most distressingly, had actually increased between 1949–1953 and 1970–1972 over the first quarter-century of existence of the British National Health Service. Nor did things improve between the early 1970s and 1980s (Marmot et al., 1987). During the 1980s and early 1990s, the British experience was replicated in other developed countries including Canada, where the introduction of national health insurance in the early 1970s had little effect on socioeconomic differences in health (Wilkins et al., 1989). Finally, adjustments for gross access to and utilization of medical care have contributed little or nothing to explaining socioeconomic and racial/ethnic differences in health in our and other data.

However, we believe that the role of medical care in socioeconomic and racial/ethnic health differences deserves renewed examination and research. First, compared to whites, racial/ethnic minorities have lower levels of access to medical care in the United States (Blendon et al. 1989; Trevino et al. 1991). Second, higher incidence rates for racial/ethnic minorities do not fully account for the higher death rates (Schwartz et al. 1990). Later initial diagnosis of disease, comorbidity, delays in medical treatment, and disparities in the quality of care also play a role. There is growing evidence of large racial/ethnic differences in the quality of medical care. Many studies have found racial/ethnic differences in the receipt of therapeutic procedures for a broad range of conditions even after adjustment for insurance status and severity of disease (e.g., Wenneker and Epstein, 1989; Harris et al. 1997). These disparities exist even in contexts where differences in economic status and insurance coverage are minimized, for example, the Veterans Administration Health System (e.g., Whittle et al., 1993) and the Medicare program (e.g., McBean and Gornick, 1994). Recent studies document that these differences in medical treatment adversely affect the health of minority group members (Peterson et al., 1997; Hannan et al., 1999). Moreover, medical care appears to play a modest role in accounting for racial differences in mortality (Woolhandler et al., 1985; Schwartz et al., 1990), and other evidence suggests that medical care has a greater impact on the health status of vulnerable racial and low-SES groups than on their more advantaged counterparts (Williams, 1990). More generally, behind declining socioeconomic and racial/ethnic disparities in gross levels of access to and utilization of medical care may lie in persisting differences in access to more continuous care from a concerned and responsive provider, associated differences in access to and utilization of important standards of preventive care (e.g., blood pressure, prostate and colorectal screening, Pap smears, mammograms, and professional advice on health behaviors), and differences in the timeliness and appropriateness of access to state-of-the-art standards of therapeutic care. Thus, socioeconomic and racial/ethnic disparities in standards and appropriateness of medical care merit increased attention in research and policy.

Psychosocial and Environmental Risk Factors

As evidence grew of the more limited impact of medical care in explaining socioeconomic and racial/ethnic disparities in health, research increasingly established a growing and even predominant role of behavioral and psychosocial risk factors in the etiology and course of human health and disease. First, and still most compelling, was the evidence of the adverse effects of cigarette smoking on mortality and morbidity, especially adult lung cancer and heart disease (e.g., USPHS, 1964). This was followed by increasing evidence of the health risks of other behaviors, including immoderate levels of eating (and hence weight) and of alcohol consumption, lack of exercise, and dietary composition, e.g., fat and fiber (Lalonde, 1975; Berkman and Breslow, 1983, USDHHS, 1990). Evidence has also accumulated on the deleterious health effects of chronic and acute stress in work and life (Theorell, 1982), hostility and depression (Scheier and Bridges, 1995), lack of social relationships and supports (House et al., 1988), and lack of control, efficacy, or mastery (Rodin, 1986), with the impact of lack of social relationships, for example, on all-cause mortality being not incomparable to that of cigarette smoking (House et al., 1988). Notably, these psychosocial factors all tend to affect a broad range of health outcomes, rather than being focused on a single outcome.

Socioeconomic Status and Psychosocial and Environmental Risk Factors

One factor (e.g., smoking) or a small set of factors (e.g., health behaviors) is sometimes seen as crucial in explaining and alleviating problems of premature morbidity and mortality more generally (e.g., USDHHS, 1990; McGinnis and Forge, 1993) and socioeconomic and racial/ethnic disparities in health in particular (Mechanic, 1989; Satel, 1996). However, increasing evidence suggests that there are few or no analogues in our current problems of public health to the necessary and sufficient microbial causes of many infectious diseases, nor are there any comparable “magic bullets” for treating, preventing, or eradicating them. Rather, the current causes of morbidity and mortality, especially from major chronic diseases, are broadly multifactorial, with no one or few decisive, but the accumulation of many being as debilitating or deadly as a virulent infectious agent (Kunitz, 1987). For example, all major health behaviors combined (i.e., cigarette smoking, immoderate weight and drinking, and lack of physical activity) appear able to explain only 10–20%, at most, of socioeconomic differences in mortality (Lantz et al., 1998). Rather, what is most striking and important about socioeconomic (and, to a lesser degree, racial/ethnic) status is the degree to which it shapes exposure to, and perhaps also the impact of, a wide range of psychosocial and environmental risk factors for health. Our data (House et al., 1992, 1994) and those of others (e.g., Marmot et al., 1991; Lynch et al., 1996) show that lower-SES individuals have a higher prevalence of almost all major psychosocial risk factors for health. That is, they manifest higher levels of risky health behaviors such as smoking, lack of exercise, immoderate eating and drinking, and high-fat-low-fiber diets. They also experience more chronic and acute stress due to, for example, their more vulnerable economic status and the higher rates of ill health and death among friends and relatives. They generally report lower levels of social relationships and supports and of personal efficacy or control, along with higher levels of hostility and depression.

Figures 2A and 2B show, for national samples of the U.S. population, the distribution and range of these psychosocial risk factors by education and income. The prevalence of each psychosocial risk factor is always highest among those with the lowest level of education and income, with rate ratios (of the lowest- to the highest-SES groups) ranging from 1.1 to 3.8 and averaging 2.0 (see Lantz et al., 1998, for parallel data on health behaviors).

FIGURE 2A

Psychosocial risk factors by education for persons 45–64 in the United States. SOURCE: University of Michigan Americans' Changing Lives Survey (1986) and MMPI-II restandardization (1984–1986).

FIGURE 2B

Psychosocial risk factors by income for persons 45–64 in the United States. SOURCE: University of Michigan Americans' Changing Lives Survey (1986) and MMPI-II restandardization (1984–1986).

Persons of lower socioeconomic status in adulthood are also more likely to live and work in physical-chemical-biological environments that are hazardous to health. Further, they are more likely to have grown up in lower-socioeconomic environments, which may have residual adverse effects on health in adulthood that research is just beginning to examine (e.g., Power et al., 1990).

At this point we are only beginning to explore the causal pathways and complexities that link socioeconomic position to exposure to behavioral, psychosocial, and environmental risk factors and, in turn, link these risk factors to health outcomes. Our own analyses and those of others find that although any single behavioral, psychosocial, or environmental risk factor, or small set thereof, can account for only a small fraction of the association between SES and health, a set of 10–20 such risk factors can account for 50–100% of the association between a given SES indicator and health outcome. Marmot et al. (1991) and Lynch et al. (1996) have reported that adjusting for 10–20 such risk factors reduces the predictive association of SES with mortality by 50–100%. We get similar results for the cross-sectional relationship between education or income and functional status (House et al., 1994).

Potentials and Limits of Understanding and Intervening in Pathways

Better understanding of the pathways and mechanisms linking socioeconomic and racial/ethnic status to health is often and appropriately seen as crucial to reducing socioeconomic and racial/ethnic disparities in health. Once an intervening variable or mechanism is understood, efforts can be made to reduce this variable or mechanism or to weaken its impact on health, hence reducing the socioeconomic or racial/ethnic disparities in health. To the extent that we are able to decrease smoking and improve other health behaviors, lessen stress, enhance social relationships and supports, modify risky psychological dispositions or reduce exposure to environmental exposures, especially in disadvantaged socioeconomic or racial/ethnic groups, or to moderate their effects on health (e.g., develop a safe cigarette or nicotine delivery system, see Paper Contribution K), we should consequently reduce socioeconomic and racial/ethnic disparities in health. Such intervention strategies have had some beneficial effects in these regards.

However, we wish to call attention to the often insufficiently recognized limitations of such an approach and to highlight the potential, and indeed necessity, of another and complementary approach. The potential of the pathways strategy is limited in two crucial ways. First, modifying any psychosocial risk factor is a difficult process, and the impact of modifying one or a single risk factor is necessarily limited for the reasons discussed above. Second, where we have had substantial success in reducing a behavioral or psychosocial risk factor (e.g., smoking), this success is often greatest among the more advantaged due to the persistence or even accentuation (e.g., targeting of disadvantaged groups for cigarette advertising) of the causal forces giving rise to the risk factor in less advantaged groups. Thus, as overall levels of smoking have decreased in the United States, socioeconomic differences in smoking have increased (see Paper Contribution K). Finally, it is important to recognize that the mechanisms that currently link socioeconomic status to health are not the same ones that did so in the past or will in the future. Prior socioeconomic differences undoubtedly had more to do with differences in exposure to infectious agents and access to medical care than is currently the case. Indeed, many of the currently most important diseases and risk factors—such as coronary heart disease and its risk factors of smoking, lack of exercise, and high-fat diet—were at one time more characteristic of upper socioeconomic strata, but have become more incident and prevalent in lower socioeconomic strata as these diseases and risk factors have become more deleterious to health. A similar trend has characterized AIDS, which emerged first as a disease among higher socioeconomic strata but has rapidly become more prevalent in disadvantaged socioeconomic and racial/ethnic groups. Thus, whatever the major diseases (or risk factors for them) are in future years, they too are likely to be importantly determined by socioeconomic status (Link and Phelan, 1995).

In sum, as Link and Phelan (1995) have cogently argued, we need to think increasingly of disparities in socioeconomic position as the “fundamental cause” of socioeconomic disparities in health. Hence, ameliorating these socioeconomic disparities themselves may be the best strategy for reducing disparities in health. A parallel, but not identical, argument can be made regarding racial/ethnic disparities, to which we now turn, before returning to evidence that socioeconomic improvement and policy are a major form of health improvement and policy in our and other societies, arguably the most important and essential one for reducing socioeconomic disparities in health.

Race/Ethnicity and Health

National mortality data reveal that African Americans (or blacks) have an overall death rate that is more than 1.5 times higher than that of whites (NCHS, 1998). The magnitude of the racial difference in death rates varies by the specific cause of death, but a pattern of elevated death rates for blacks compared to whites exists for almost all the leading causes of death in the United States. In contrast, all other racial/ethnic groups have an overall death rate that is lower than that of whites. However, there is considerable variability for subgroups of these populations and for specific health conditions. All nonblack minorities have considerably lower rates than whites for the two leading causes of death (heart disease and cancer) but higher rates for some conditions. Hispanics have higher mortality rates than non-Hispanic whites for tuberculosis, septicemia, HIV/AIDS, chronic liver disease and cirrhosis, diabetes, and homicide (Sorlie et al., 1993; Vega and Amaro, 1994). Subgroups of the Asian and Pacific Islander population also have elevated mortality rates for some health conditions (Lin-Fu, 1993). For example, the Native Hawaiian population has the highest death rate due to heart disease of any racial group in the United States (Chen, 1993). Similarly, American Indians who receive care from the Indian Health Service (60% of that population) have age-adjusted mortality rates higher than the national average for tuberculosis, alcoholism, diabetes, accidents, homicides, suicides, and pneumonia and influenza (NCHS, 1993).

What Is Race?

Early studies of racial variations in health viewed race as primarily reflecting biological homogeneity and racial differences in health as largely genetically determined. This view predated modern scientific theories of genetics and carefully executed genetic studies. In contrast, scientific evidence suggests that our current racial categories are more alike than different in terms of biological characteristics and genetics (Lewontin, 1972; Gould, 1977; Latter, 1980). All human beings are identical for about 75% of known genetic factors, with about 95% of human genetic variation existing within racial groups (Lewontin, 1982). Thus, there is more genetic variation within races than between them, and racial categories do not capture biological distinctiveness. Race is thus more of a social than a biological category, and racial classification schemes have been influenced by larger social and political considerations (Cooper and David, 1986; Williams, in press).

Race and SES

Although not useful as biological markers, current racial/ethnic categories capture an important part of the inequality and injustice in American society (See and Wilson, 1988). There are important power and status differences between groups. For example, in 1995 the poverty rate for Asians was almost twice that of whites, while the rate for blacks and Hispanics was more than three times that of non-Hispanic whites (NCHS, 1998). Data on poverty tell only a part of the story of economic vulnerability. In addition to persons who actually fall below the government's poverty threshold, a large number of persons are only slightly above this level. Many of these persons are at a high risk of becoming poor. The combination of the poor and near-poor (annual income above the poverty threshold but less than twice the poverty level) categories reveals that one in every three persons in the United States falls into this economically vulnerable category—26% of whites, 33% of Asians, 54% of blacks, and 62% of Hispanics (NCHS, 1998). Although there is a strong relationship between race and SES, they are not equivalent. For example, the rate of poverty is three times higher for blacks than for whites, but two-thirds of blacks are not poor, and two-thirds of all poor Americans are white.

Race, SES, and Health

Research reveals that SES differences between races account for much of the racial differences in health. Adjusting racial (black-white) disparities in health for SES sometimes eliminates, but always substantially reduces, these differences (Krieger et al., 1993; Williams and Collins, 1995; Lillie-Blanton et al., 1996). However, race often has an effect on health independent of SES: within levels of SES, blacks still have worse health status than whites.

Table 1 illustrates these issues with life expectancy data. At age 45, white males have a life expectancy that is almost 5 years more than black males (NCHS, 1990). Similarly, the life expectancy at age 45 for white females is 3.7 years longer than that of similarly aged black women. However, there is considerable socioeconomic variation in life expectancy within both racial groups (NCHS, 1998). When we consider the distribution of life expectancy by race and income, two important trends emerge. First, for both racial groups, income is strongly linked to health status. Consistently, persons of lower levels of income report lower life expectancy than their more economically favored peers. Black men in the highest-income group live 7.4 years longer than those in the lowest-income group. The comparable numbers for whites was 6.6 years. Thus, the SES difference within each racial group is larger than the racial difference across groups. A similar pattern is evident for women, although the SES differences are smaller. At age 45, black women in the highest-income group have a life expectancy that is 3.8 years longer than those in the lowest-income group. Among whites, the SES difference is 2.7 years. Moreover, for men and women of both racial groups, increasing levels of income are associated with longer life expectancy. The power of SES in shaping racial differences in health is clearly evident by comparing the highest-SES blacks with the lowest-SES whites, especially among males. High-income black males have a life expectancy that is 5.3 years longer than low-income white males. Thus, the disproportionate concentration of African Americans at lower SES levels is a major factor behind the overall racial differences in health.

The second pattern that clearly emerges in these data is that race is more than socioeconomic status. Consistently, there is an independent effect of race even when SES is controlled. At every level of income, for both men and women, African Americans have lower levels of life expectancy than whites. In these data, the differences are greater at the two lower levels of income than at the two higher-economic- status categories. However, for some indicators of health status such as infant mortality, the racial gap becomes larger as SES increases (NCHS, 1998).

Role of Racism or Discrimination

The construct of racism can structure and inform our understanding of racial inequalities in health (Cooper et al., 1981; Krieger et al., 1993; Hummer, 1996; LaVeist, 1996; Williams, 1997; in press). The term racism refers to an ideology of inferiority that is used to justify the differential treatment of members of racial outgroups by both individuals and societal institutions, usually accompanied by negative attitudes and beliefs toward these groups. Racism has been a central organizing principle within American society and has played a key role in shaping major social institutions and policies (Omi and Winant, 1986; Quadagno, 1994). Historically, ideologies about racial groups were translated into policies and societal arrangements that have limited the opportunities and social mobility of stigmatized groups. The strong association between race/ethnicity and SES in the United States reflects the successful implementation of social policies that were designed to limit societal resources and rewards to socially marginalized groups.

There have been important positive changes in the racial attitudes of whites toward blacks in recent decades and broad current support for the principle of equality in most societal institutions (Schuman et al., 1997). At the same time, there is considerably less support for policies that would actually implement equal access to education, housing, jobs, and so forth (Schuman et al., 1997). Moreover, national data on stereotypes reveal that whites view blacks, Hispanics, and Asians more negatively than themselves, with blacks viewed more negatively than all other groups and Hispanics twice as negatively as Asians (Davis and Smith, 1990). Such a high level of acceptance of negative stereotypes of minority groups is an ominous harbinger of widespread societal discrimination. Psychological research indicates that the endorsement of negative racial stereotypes leads to discrimination against minority groups (Devine, 1995; Hilton and von Hippel, 1996). Moreover, well-learned stereotypes are resistant to disconfirmation (Stangnor and McMillan, 1992), and their activation is an automatic process, with individuals spontaneously becoming aware of relevant stereotypes after encountering someone to whom the stereotypes are applicable (Devine, 1989; Hilton and von Hippel, 1996).

Research reveals that considerable racial/ethnic discrimination persists in the United States in domains that affect socieconomic mobility, such as housing and employment (Kirschenman and Neckerman, 1991; Neckerman and Kirschenman, 1991; Fix and Struyk, 1993). Thus, the advent of civil rights legislation and changes in the racial attitudes of whites have not been sufficient to eradicate discrimination, and there has been remarkable stability over time on multiple dimensions of racial inequality (Economic Report of the President, 1998). For example, the median income of African Americans was 59 cents for every dollar earned by whites in 1996—identical to what it was in 1978.

Racism affects disparities in health in multiple ways. First, racism restricts and truncates socioeconomic attainment. The consequent racial differences in SES and poorer health reflect, in part, the impact of economic discrimination produced by large-scale societal structures. Residential segregation has been a primary mechanism by which racial inequality has been created and reinforced. Racial segregation has determined access to educational and employment opportunities that has importantly led to truncated socioeconomic mobility for blacks and American Indians (Jaynes and Williams, 1987; Massey and Denton, 1993). Residence in segregated neighborhoods can lead to exposure to environmental toxins, poor-quality housing, and other pathogenic living conditions, including inadequate access to a broad range of services provided by municipal authorities (Collins and Williams, 1999). These conditions importantly account for the large racial difference in homicide. The combination of concentrated poverty, male joblessness, and residential instability leads to high rates of single-parent households, and these factors together account for variation in the levels of violent crime (Sampson and Wilson, 1995). Importantly, the association between these factors and violent crime for whites was virtually identical in magnitude with the association for African Americans. Several studies have found a positive association between both adult and infant mortality and residence in segregated areas. One recent study has documented elevated mortality rates for both blacks and whites in cities high on two indices of segregation compared to cities with lower levels of segregation (Collins and Williams, 1999). This pattern suggests that beyond some threshold of segregation, the adverse conditions linked to highly segregated cities may negatively affect the health of all persons who reside there.

Moreover, because of racism, SES indicators are not commensurate across racial groups, which makes it difficult to truly adjust racial differences in health for SES (Kaufman et al., 1997). There are racial differences in the quality of education, income returns for a given level of education or occupational status, wealth or assets associated with a given level of income, the purchasing power of income, the stability of employment, and the health risks associated with occupational status (Williams and Collins, 1995; Kaufman et al., 1997).

Racial differences are especially marked for wealth. Eller (1994) shows that while white households have a median net worth of $44,408, the median net worth is $4,604 for black households and $5,345 for Hispanic ones. Moreover, racial/ethnic differences in wealth are evident at all levels of income and are greatest at the lowest income level. For persons in the lowest quintile of income in the United States, the net worth of whites is 10,000 times higher than that of blacks ($10,257 versus $1).

As noted earlier, systematic discrimination can also affect the quantity and quality of services received, including medical care. Recent research has focused on the potential health consequences of subjective experiences of discrimination. Racism in the larger society can also lead to systematic differences in exposure to personal experiences of discrimination. These experiences of discrimination may be an important part of subjectively experienced stress that can adversely affect health. A growing body of evidence indicates that self-reported measures of discrimination are adversely related to physical and mental health in a broad range of racial/ethnic minority populations (Amaro et al., 1987; Salgado de Snyder, 1987; Krieger, 1990; Dion et al., 1992; Jackson et al., 1996; Krieger and Sidney, 1996; Kessler, et al., 1999; Noh et al., 1999). Two recent studies suggest that exposure to discrimination plays a role in explaining observed racial differences in self-reported measures of health (Williams et al., 1997; Ren, et al., 1999).

A small body of research suggests that the prevalence of negative stereotypes and cultural images of stigmatized groups can adversely affect health status. First, the widespread societal stigma of inferiority can create specific anxieties, expectations, and reactions that can affect health indirectly by having an adverse impact on socioeconomic performance and mobility (Fischer et al., 1996; Steele, 1997). There may also be more direct health effects. Researchers have long identified that one response of minority populations would be to accept the dominant society's ideology of their inferiority as accurate. A few studies have operationalized the extent to which African Americans internalize or endorse these negative cultural images. These studies have found that internalized racism is positively related to psychological distress, depressive symptoms, substance use, and chronic physical health problems (Taylor and Jackson, 1990; Taylor et al., 1991; Williams and Chung, in press).

Medical Care

What we know and do not know about the role of medical care in producing and alleviating socioeconomic and racial/ethnic disparities in health is illustrative. As already noted, there can be no question that wider availability of effective therapeutic and preventive medical care has improved population health, though to a more limited degree than is often presumed. However, Preston and Haines (1991) argued that improvements in medical care may also have exacerbated socioeconomic and racial/ethnic disparities in health to the extent that differential access to care has become more consequential for health. As we have seen, the implementation of national health services or insurance has failed to reduce socioeconomic and racial/ethnic disparities in health to the degree we had hoped or expected, if at all. Growing evidence suggests, however, that gross equalization of access or utilization fails to equalize access to maximally appropriate and effective care.

Addressing three kinds of issues offers promise of reducing socioeconomic and racial/ethnic disparities in the quality and appropriateness of care and hence in health. The first is to recognize that differentials in the ways the medical care system deals with different groups can lead to problems in the delivery of care. If disadvantaged socioeconomic and racial/ethnic groups do not have access to the same type and quality of providers and the same kind of relationships and communication with them as more advantaged persons, the result is likely to be their receiving less regular, preventive, and appropriate care. Second, the focus of the medical care system on different types of care differentially affects the health of different groups. Disadvantaged groups may benefit more from improvements in basic primary and preventive care; advantaged groups, from secondary and tertiary care. Thus, relatively poor societies (e.g., China, Costa Rica, Sri Lanka, Kerala State of India) have achieved “good health at low cost” by focusing their limited resources on ensuring equal access to basic primary and preventive care (Halsted et al., 1985). Finally, as discussed above, even secondary and tertiary care appears to be distributed inequitably by racial/ethnic groups and probably also by socioeconomic status. This may explain why the United States generally lags behind other developed nations in population life expectancy, but surpasses all in life expectancy at age 80, where a high technology system supports a relatively elite set of survivors in the population.

Finally, we must also give more attention in medical care to identifying the ways in which the lives of individuals are constrained by broader social, economic, and political forces. Some evidence suggests that the effectiveness of behavioral interventions varies by the degree that they attend to social situations in which individuals are embedded. Syme's (1978) study of 244 hypertensive patients clearly illustrates how addressing underlying social and economic conditions appears to enhance the management of hypertension and improve the effectiveness of antihypertensive therapy. The patients in this study were matched on age, race, gender, and blood pressure history and randomly assigned to one of three groups. The first group received routine hypertensive care from a physician. In addition to routine hypertensive care, the second group also attended 12 weekly clinic meetings providing health education with regard to hypertension by a health educator and nurse practitioner. In addition to routine hypertensive care, the third group was visited by community health workers who had been recruited from the immediate community and provided with one month of training to address the diverse social and medical needs of persons with hypertension. These outreach lay workers provided information on hypertension but also discussed family difficulties, financial strain, and employment opportunities and, as appropriate, provided support, advice, referral, and direct assistance.

After seven months of follow-up, patients in the third group were more likely to have their blood pressure controlled than patients in the other two groups. In addition, those in the third group knew twice as much about blood pressure and were more compliant with taking their hypertensive medication than patients in the other two groups, and the good compliers in the third group were twice as successful at controlling their blood pressure as good compliers in the health education intervention group. Thus, even the effectiveness of the pharmacological treatment appeared to be enhanced in the group that also addressed the underlying stressful conditions of these hypertensive persons.

A study by Buescher and colleagues (1987) further illustrates how addressing underlying economic and social issues can improve the impact of medical care. This study compared the effectiveness of two approaches to delivering prenatal care in a population of predominantly black low-SES women in Guilford County, North Carolina. One group received prenatal care at the county health department. The other group received prenatal care from private practice physicians. Women who received care from the community-based physicians were twice as likely to have a low-birthweight baby, compared to those visiting the health department. The health department's prenatal care program attempted to comprehensively address the medical and social needs of the pregnant mothers. Prenatal care was provided by nurse practitioners, instead of physicians. Time was devoted during prenatal care visits to counseling the women about nutrition and other aspects of personal care. As appropriate, referrals were made to the Women, Infants and Children Program, which provides nutritional supplements to poor women. These referrals, as well as missed clinic appointments, were followed up aggressively. James (1993) argues that the positive cultural features of this program may have been very important. It appears that the county health department's program offered low-income women an extended network of social support, capable of meeting their needs in much the same way that older, more knowledgeable women have traditionally guided and supported young inexperienced mothers (James, 1993).

Adaptive Attributes of Disadvantaged Groups

One of the major paradoxes in U.S. population health is that African Americans and Latinos are not as disadvantaged on some aspects of health as their socioeconomic positions would lead us to expect. Thus, although African Americans tend to have higher levels of ill health than whites for most indicators of physical health and are also disadvantaged compared to whites on indicators of subjective well-being such as life satisfaction and happiness (Hughes and Thomas, 1998), they have comparable or better health status than whites for other indicators of mental health. Community-based studies using measures of psychological distress show an inconsistent pattern of black-white differences. Some studies show that blacks have higher rates of distress compared to whites, while other studies show higher rates of psychological distress for whites compared to blacks (Dohrenwend and Dohrenwend, 1969; Neighbors, 1984; Vega and Rumbaut, 1991; Williams and Harris-Reid, 1999). However, when rates of psychiatric illness are considered, African Americans have comparable or lower rates of mental illness than whites. In the Epidemiologic Catchment Area Study (ECA), the largest study of psychiatric disorders ever conducted in the United States, there were very few differences between blacks and whites in the rates of both current and lifetime psychiatric disorders. Anxiety disorders, especially phobias, stand out as one area in which blacks had considerably higher rates than Caucasians. In the National Comorbidity Study blacks do not have higher rates of disorder than whites for any of the major classes of disorders (Kessler et al., 1994). Instead, lower rates of disorders for blacks than whites are especially pronounced for the affective disorders (depression) and the substance abuse disorders (alcohol and drug abuse).

These findings emphasize the need for renewed attention to identify the cultural strengths and health-enhancing resources within the black community. Two social institutions—the family and the church—stand out as crucial for the black population. Strong family ties and an extended family system are important resources that may reduce some of the negative effects of stress on the health of black Americans. At the same time, a recognition of the strengths of black families should not be used to romanticize them as if they were a panacea for a broad range of adverse living conditions. While these networks of mutual aid and support do facilitate survival, they are also likely to provide both stress and support. Moreover, it is likely that cutbacks in government-provided social services in recent years have increased the burdens and demands on the support services provided by the black family. The black American church has been the most important social institution in the black community. These churches have historically been centers of spiritual, social, and political life. Black churches may promote mental health by providing a broad range of social and human services to the African-American community, serving as a conduit to the formal mental health system, providing a base for friendship networks, and facilitating collective catharsis and stress reduction through religious rituals and participation (Williams, 1998).

A second paradox is evident for Mexican Americans. In spite of high rates of poverty and comparatively low levels of access to medical care, Mexican Americans tend to have similar or better levels of health than the white population. Moreover, across a broad range of health status indicators foreign-born Hispanics have a better health profile than their counterparts born in the United States. This pattern may reflect the impact of migration. Rates of infant mortality, low birthweight, cancer, high blood pressure, adolescent pregnancy, and psychiatric disorders increase with length of stay in the United States for Hispanics (Vega and Amaro, 1994). It is likely that increasing length of stay and greater acculturation of the Hispanic and Asian population will lead to worsening health. Early studies of acculturation found that rates of heart disease among Japanese increased progressively as they moved from Japan to Hawaii to the U.S. mainland (Marmot and Syme, 1976). As groups migrate from one culture to another, immigrants often adopt the diet and behavior patterns of the new culture. Several behaviors that adversely affect health status appear to increase with acculturation. These include decreased fiber consumption, decreased breast feeding, increased use of cigarettes and alcohol—especially in young women, driving under the influence of alcohol, and use of illicit drugs (Vega and Amaro, 1994). However, the association between acculturation, length of stay in the United States, and the prevalence of disease may be complex. Migration studies of the Chinese and Japanese show that the rates of some cancers, such as prostate and colon, increase when these populations migrate to the United States, while the rates of other cancers, such as liver and cervical, decline (Jenkins and Kagawa-Singer, 1994). Research is needed to identify the extent to which there are specific aspects of culture that promote health and the strategies that may be utilized to facilitate their maintenance over time.

Macrosocial Change and Health

Not to be ignored is the evidence of history that improvements in the educational, occupational, and income levels of populations have produced massive improvements in population health that in turn improve the human capital necessary for further socioeconomic advancement (see Figure 4). This has included the reduction of health disparities by socioeconomic position, gender, and/or race/ethnicity. Not surprisingly, educational and economic development are major priorities of less developed societies, and also of the developed countries.

Social Welfare and Health

Similarly, evidence from the more developed countries strongly suggests that ensuring that the fruits of development are broadly distributed, especially to the broad lower range of the population, assists in promotion of health. The rise of Sweden and then Japan to the highest levels of population health in the world, and the clearly reduced socioeconomic disparities in the case of Sweden, must be significantly attributed to their emphasis on ensuring good conditions of life for all, albeit via different mixes of social policy and provision of private and public goods.

More Focused Social Policies

One would like to have clearer evidence, however, that specific policies that improve the social and economic status of disadvantaged groups also improve their health. Unfortunately, we have not always evaluated the effects of such policies, and rarely their health impacts, and this remains an agenda for future research. However, limited and developing evidence is at least consistent with our thesis.

At least one study from the evaluation of the negative income tax experiments of the early 1970s shows positive effects on maternal and child health. In a study of expanded income support, Kehrer and Wolin (1979) found that the birthweight of infants born to mothers in the experimental income group was higher than that of those born to mothers in the control group although neither group experienced any experimental manipulation of health services. Improved nutrition, probably a result of the income manipulation, appeared to have been the key intervening factor. Evaluation of the long-term effects of early childhood interventions, such as the Perry preschool study, also suggests long-term effects on socioeconomic factors (median income and rates of home ownership) as well as on other behavior and beliefs (e.g., staying in school, avoiding delinquency and crime, achieving better jobs) that should promote health.

Major Exogenous Income Policies

An ideal test of our thesis would be the examination of social policies that have markedly improved the economic status of all or some of the disadvantaged population. In the United States, two stand out. One is Social Security which has provided an income support program for the elder portion of the population that has, over time, lifted or kept most of them out of poverty. We (Arno and House, in progress) have undertaken to see if one can find evidence that Social Security has also improved the health of the elderly. We have sought to do so by examining the mortality experience of different adult age groups in the population over the course of the century. We hypothesized that the two largest exogenous income improvements from Social Security occurred first at its inception in the late 1930s and second when it was indexed to inflation in the late 1960s and early 1970s. Hence, we expected that one should see discontinuous improvements in the health of the elderly, but not in other age groups, after these two periods. Our initial results, shown in Figure 6, seem to us consistent with this. We recognize that there may be other compelling explanations of these two discontinuities, but we do not believe that others have or would have predicted both of them ex ante or that other explanations can explain both as well or as parsimoniously ex post. Further, these trends are consistent with the evidence that socioeconomic differences in health are markedly reduced among the elderly as compared to working-age adults (House et al., 1990, 1994). The other major income support program in our country has been the earned income tax credit, but we know of no research on its effects and would certainly see such research as a high priority for the future.

FIGURE 6

Total mortality, United States, 1900–1995.

Investment in Public Goods and Infrastructure

Limited data also indicate that efforts to improve the public goods and infrastructures of communities improve the health of their residents. Some research suggests that policy changes to improve neighborhoods can importantly enhance health. Dalgard and Tambs (1997) provide findings from a 10-year follow-up study of residents in five neighborhood types in Norway. This study found that residents in a poorly functioning neighborhood that had experienced dramatic change in its social environment over time reported improved mental health 10 years later. The improvements in the neighborhoods included a new public school, playground extension, establishment of a sports arena and park, organization of activities for adolescents by the sports association of the municipality, establishment of a shopping center with restaurants and a cinema, and a subway line extension into the neighborhood. This effect was not explained by selective migration. Similarly, an intervention in England for a poorly functioning neighborhood also had dramatic effects (Halpern, 1995). Over a 2-year period this intervention refurbished housing, with a special emphasis on making it safe and sheltered from strangers. Changes included improved traffic regulations, improved lighting and strengthening of windows, enclosure of gardens for apartments, closure of alleyways, and landscaping. In this project, residents were involved in the planning process. A 1-year follow-up study, conducted after the intervention had been in place, documented that the changes in the physical environment were associated with changes in the social environment and mental health as well. That is, the contact between neighbors had increased and neighbors reported more trust in each other. Levels of optimism and belief in the future had increased, and residents felt a stronger identification with their neighborhood. In addition, levels of anxiety and depression were significantly reduced among residents. This study reveals that improvement in the quality of life in a neighborhood can increase both the quality of social interaction or cohesion and health (see Paper Contribution I).

The Case of Racial/Ethnic Disparities

Racial/ethnic disparities in health should clearly be reduced by policies that reduce absolute and relative socioeconomic deprivation. However, they also need special approaches, obviously not aimed at changing race/ethnicity per se, but, rather, at changing the way race/ethnicity and associated racial/ethnic status are socially defined and constructed. Again, available evidence suggests there are reduced racial/ethnic disparities in health. Mullings (1989) has suggested that the civil rights movement, for example, had important positive effects on black health. By reducing occupational and educational segregation, it improved the SES of at least a segment of the black population and also influenced public policy to make health care accessible to larger numbers of people. Consistent with this hypothesis, one study found that between 1968 and 1978, blacks experienced a larger decline in mortality rates (on both a percentage and absolute basis) than whites (Cooper et al., 1981).