RACIAL/ETHNIC MORTALITY DISPARITIES AMONG THE ELDERLY

Overall Mortality Disparities Using Vital Statistics and Census Data

We begin by examining racial/ethnic disparities in older adult mortality. The National Center for Health Statistics (NCHS) constructs official mortality rates based on U.S. Vital Statistics (numerator) and Census (denominator) data. The advantages of these data sources are that they are large and cover the entire population, including individuals in nursing homes, long-term care institutions, and prisons. Although important and informative, there are some well-known limitations with the quality and reliability of the official death rates by race/ethnicity, especially among the elderly (Coale and Kisker, 1986; Elo and Preston, 1997; Kestenbaum, 1992; Lauderdale and Kestenbaum, 2002; Preston, Elo, Rosenwaike, and Hill, 1996; Rosenberg et al., 1999; Rosenwaike and Hill, 1996). One problem is reporting disparities between the two data sources. Disparities may occur because racial/ethnic identification on the Census is completed most often by a household member, while identification at the time of death is assigned most often by a funeral director (Rosenberg et al., 1999). Another problem is that a number of recent studies have shown significant levels of age misreporting among the elderly, which can seriously bias old-age mortality estimates (e.g., Preston et al., 1996). Third, Census undercount, particularly of racial and ethnic minority populations, can artificially bias mortality estimates for these groups upward, although adjustments can be made for the estimated undercount (Rosenberg et al., 1999). Despite these limitations, these official data remain a key source for describing racial/ethnic mortality disparities by age, sex, and geographic area.

Panel A of Table 3-1 presents official death rates per 100,000 by race/ ethnicity and sex for 5-year age groups among the U.S. elderly population in 1999 (Hoyert, Arias, Smith, Murphy, and Kochanek, 2001); Panel B presents rate ratios for the specific racial/ethnic, age, and sex groups vis-à-vis non-Hispanic white elders. As the ratios in Panel B demonstrate, the reported mortality rates of most of the racial/ethnic minority groups (e.g., persons of Hispanic, API, and Native American origin) are lower than or roughly equal to those of non-Hispanic whites at ages 65 to 69 and tend to become comparatively more advantaged at the advanced ages. Among non-Hispanic blacks, the mortality rates are 30 to 50 percent higher than non-Hispanic whites at ages 65 to 79, converge quite rapidly at ages 80 to 84, and eventually cross over among persons ages 85+. Although levels of mortality are higher among men than women for each racial/ethnic and age group, the relative disparities by race/ethnicity vary little by sex. Thus, these official rates depict non-Hispanic blacks to have the highest mortality among most of the elderly age groups, while rates for APIs, Hispanics, and Native Americans are generally lower than non-Hispanic whites.

TABLE 3-1

Death Rates per 100,000 by Race/Ethnicity and Death Rate Ratios Compared with Non-Hispanic Whites for the Elderly Population of the United States, Official U.S. Mortality Data, 1999.

Recent demographic work has been undertaken to evaluate and reestimate black, white, and Asian-American mortality estimates among the elderly for the various sources of bias mentioned (e.g., Elo, 2001; Hill, Preston, and Rosenwaike, 2000; Lauderdale and Kestenbaum, 2002; Preston et al., 1996). The reestimates suggest that the general mortality patterns for these three population groups described remain consistent; that is, black mortality remains significantly higher than that of whites for most elderly age groups, with the greatest disparities occurring among the young-old (ages 65-74), and then convergence and crossover at the oldest ages (Hill et al., 2000). Likewise, new estimates of Asian-American older adult mortality were shown to be lower than whites, although the advantage may not be as great as demonstrated in the official data (Lauderdale and Kestenbaum, 2002).

However, a debate continues about whether a real black-white mortality crossover occurs among the oldest-old (Nam, 1995). Although researchers for many years have documented such a mortality crossover using a number of different data sets and have concluded that it appears to be real (Johnson, 2000; Kestenbaum, 1992, 1997; Manton, Poss, and Wing, 1979; Manton and Stallard, 1997; Nam, Weatherby, and Ockay, 1978; Parnell and Owens, 1999), others have been more skeptical because of the data quality concerns (Coale and Kisker, 1986; Preston et al., 1996). The most recent, carefully produced evidence by a research team from the latter group continues to find a racial mortality crossover occurring at ages 90 to 94 for females and 95+ for males (Hill et al., 2000). Although the crossover is identified at an older age than a number of other researchers have found, the weight of the evidence, using a number of nationally based U.S. data sources, is strong that a black-white crossover exists. Probably more important, the evidence from Hill et al. (2000) shows that mortality rates among aged 80+ U.S. whites (and, given their similarity to white rates, black rates as well) are lower than reports in other low-mortality countries with good data, as originally documented for the U.S. white population by Manton and Vaupel (1995).

Notably, and in contrast to the in-depth work devoted to investigating data quality and refining old-age black and white mortality estimates, relatively few researchers have examined data quality or corrected for age misreports among other racial/ethnic groups (see Lauderdale and Kestenbaum, 2002, for an excellent exception that examines Asian-American mortality). Among the studies that have examined death rates across the life course, Rosenberg et al. (1999, p. 9) find that for population groups other than non-Hispanic whites and blacks, “levels of mortality are seriously biased from mis-reporting in the numerator and under-coverage in the denominator of the death rates.” Their findings suggest that officially reported death rates for Native Americans may be more than 20 percent too low, while those reported for APIs and Hispanics may be about 11 percent and 2 percent too low, respectively. On the other hand, Rosenberg et al. (1999) found that officially reported rates for non-Hispanic whites and non-Hispanic blacks were most likely 1 percent and 5 percent too high, respectively. Their refined estimates of age-adjusted death rates across the life course suggest that API and Hispanic death rates still remain the lowest (in that order), while Native American adjusted rates are higher than those of non-Hispanic whites but lower than those of non-Hispanic blacks. How these adjustments for known sources of error specifically influence elderly adult death rates is unknown, although one recent report found that Native American adult decedents were most likely to be misclassified at older adult ages (Stehr-Green, Bettles, and Robertson, 2002). Thus, the officially reported mortality disparities shown in Table 3-1 should be interpreted with great caution, with the low mortality levels for Native Americans, especially at the oldest ages, particularly suspect. We will present alternative estimates of racial/ethnic mortality disparities among the elderly, based on survey data linked to mortality follow-up information below.

Cause-Specific Mortality Disparities Using Vital Statistics and Census Data

Following up on the documentation of overall mortality by race/ ethnicity, Table 3-2 (also see Figure 3-1) presents cause-specific mortality rates (per 100,000 population) by race/ethnicity for the leading causes of death among the elderly population by gender. These rates are standardized to the gender-specific age distributions of the non-Hispanic white population. Similar to the overall mortality rates, substantial caution is warranted again, particularly for Hispanics, APIs, and Native Americans, because of problems of racial/ethnic misclassification of decedents, age misreporting, and Census undercount (Rosenberg et al., 1999).

TABLE 3-2

Racial/Ethnic Disparities in Cause-Specific Mortality Rates (per 100,000 population) for the Top 10 Causes of Death Among the U.S. Elderly Population, 1999.

FIGURE 3-1

Cause-specific age-standardized mortality rates by race/ethnicity for the five leading causes of death among the elderly, U.S. population, 1999.

For women (Panel A, Table 3-2), diseases of the heart, malignant neoplasms, and cerebrovascular diseases are the three leading causes of death for most racial/ethnic groups and account for 61 to 66 percent of deaths among all groups except Native Americans. For the two leading causes, non-Hispanic black rates are roughly twice as high as those reported for Hispanics, APIs, and Native Americans. For Native Americans, diabetes is the third leading cause of death and the three aforementioned causes account for just 53 percent of all deaths. High cause-specific rates of respiratory disease and Alzheimer's disease stand out among non-Hispanic whites, while non-Hispanic blacks have the highest rates for diseases of the heart, malignant neoplasms, cerebrovascular diseases, influenza/pneumonia, diabetes mellitus, nephritis, septicemia, and the residual category. API women stand out for having the lowest rate for several of the causes, exhibiting especially low rates of respiratory diseases, influenza/pneumonia, Alzheimer's disease, diabetes mellitus, and septicemia. For example, the age-standardized rate of respiratory disease mortality for API women (75.5) is roughly 69 percent lower than among non-Hispanic white women (294.9), and the rate of septicemia mortality for API women (31.9) is about 85 percent lower than exhibited by non-Hispanic black women (140.0).

For elderly men (Panel B, Table 3-2), non-Hispanic blacks and non-Hispanic whites, in that order, also exhibit by far the highest rates of heart disease and malignant neoplasm mortality. For all groups except non-Hispanic whites, cerebrovascular disease is the third leading cause, while for non-Hispanic whites, respiratory disease is the third leading cause. Indeed, similar to the rates for women, non-Hispanic white elderly men exhibit, by far, the highest reported mortality rate for chronic lower respiratory diseases. Non-Hispanic white men (like their counterpart women) are also characterized by the highest reported rates of Alzheimer's disease mortality. For most causes, however, rates are highest among non-Hispanic black male elders. For example, death rates for nephritis and related causes, septicemia, and diabetes mellitus are highest, by far, among non-Hispanic black men in comparison to all other groups. Relatively high rates of diabetes mellitus also stand out for Native American and Hispanic males, while accident mortality is the highest among Native American male elders. As with women, the API male population is characterized by the lowest overall rates of mortality and lowest rates for several specific causes of death

Overall Mortality Disparities Using a Survey-Based Data Set

Large, population-based survey data sets, with links to follow-up mortality information, provide another important source of information regarding mortality disparities among the elderly. Using survey-based data sets linked to follow-up death records (i.e., the National Death Index) to analyze adult mortality patterns in the United States offers some important advantages and disadvantages. Key advantages include the fact that the racial/ethnic identifying information is provided by the individual or a coresident of the individual, while, in contrast, official U.S. mortality data are based on reports of race/ethnicity provided by the funeral director that may or may not match the individual's own racial/ethnic identity reported in the Census (Rosenberg et al., 1999). Second, and perhaps most important, survey-based data sets provide an array of covariates measured at the time of the survey from which the racial/ethnic patterns of mortality can be understood more thoroughly (e.g., Rogers, Hummer, and Nam, 2000), an enormous advantage over the more limited vital statistics-based data.

On the other hand, the survey-based data sets do not cover the complete population: They are samples of the U.S. population and most often exclude the noninstitutionalized population (i.e., persons in nursing homes and prisons) by design. The matched data sets are also believed to miss between 2 and 5 percent of decedents during the mortality follow-up period (e.g., NCHS, 2000). This may particularly influence the findings for racial/ ethnic groups that have high percentages of immigrants. Because the identification of deaths is heavily influenced by matching Social Security numbers from the death file to the original survey report, the quality of matches has been shown to be lesser among heavily immigrant populations (Hummer, Rogers, Amir, Forbes, and Frisbie, 2000; Liao et al., 1998). Third, again for some racial/ethnic groups composed of a large percentage of immigrants (e.g., Mexican Americans), return migration to the country of origin, after their original inclusion in the survey, may also bias survey-based follow-up estimates of mortality downward, although one recent study suggests that return migration effects cannot account for the relatively low adult mortality rates that have been demonstrated for the U.S. Hispanic population (Abraido-Lanza, Dohrenwend, Ng-Mak, and Turner, 1999). However, this hypothesis has never been tested directly using mortality records from out-migration countries such as Mexico. Finally, sample sizes among the oldest-old population in most nationally based survey data sets tend to be quite small, particularly for mortality follow-up purposes, thus providing unstable estimates at the oldest ages and making detailed cause-specific and sex-specific analyses for relatively small racial/ethnic populations unstable.

The National Health Interview Survey—Multiple Cause of Death (NHIS-MCD) linked data set (NCHS, 2000) is perhaps the finest of this kind for the study of mortality patterns within the U.S. population. The National Health Interview Survey (NHIS) conducted by the National Center for Health Statistics is an annual health interview of a nationally representative sample of individuals. It is the primary source of information on the health of individuals in the United States. The annual survey includes information from approximately 100,000 people (encompassing nearly 40,000 households annually) regarding central items such as age, sex, race/ ethnicity, nativity, income, education, and self-reported health and activity limitation status. Moreover, its link to the National Death Index provides a unique opportunity to examine mortality patterns among racial/ethnic groups with a large prospective data set.

Table 3-3 and Figure 3-2 show estimated racial/ethnic disparities in older adult (65+) mortality for men and women combined at several specific ages. Non-Hispanic whites are specified as the reference group, as is the case in most studies of U.S. mortality. The disparities are estimated using results from a proportional hazards model of mortality risk applied to the NHIS-MCD linked data set (NCHS, 2000). The equations specified mortality risk as a function of race/ethnicity, sex, age, and race/ethnicity by age interaction terms (which account for the possible widening or narrowing of mortality disparities vis-à-vis non-Hispanic whites with increasing age). The race/ ethnicity by age interaction effects proved to be statistically significant for non-Hispanic blacks, APIs, and Native Americans and, thus, the racial/ethnic disparities in comparison to non-Hispanic whites for these groups are shown to vary by age. Data in Table 3-3 include 82,868 individuals aged 65 and older at the time of the baseline interviews, which were conducted in six different NHIS survey years, 1989 through 1994. Mortality follow-up was assessed through the end of 1997, which resulted in 20,145 deaths.

TABLE 3-3

Predicted Racial/Ethnic Mortality Disparities by Age for the Elderly Population of the United States, NHIS-MCD, 1989-1997.

FIGURE 3-2

Predicted racial/ethnic mortality disparities by age, United States, 1989-1997.

These survey-based, follow-up mortality results show non-Hispanic black mortality to be 40 percent higher than non-Hispanic white mortality at age 65, with convergence and eventual crossover at the oldest ages. This disparity is modestly smaller than what was shown in Table 3-1, where for adults aged 65 to 69, non-Hispanic black females and males were each shown to exhibit 47 percent higher mortality than their non-Hispanic white counterparts. The results in Table 3-3 also show 19 percent higher Native American mortality compared to non-Hispanic whites at age 65 (although the small number of Native American deaths yielded a statistically nonsignificant overall difference in comparison to non-Hispanic whites). Perhaps more important, the survey-based results in Table 3-3 for Native Americans do not show substantially lower mortality among Native Americans at the oldest ages, as exhibited in the official mortality data in Table 3-1. The survey-based data in Table3-3 continue to demonstrate lower Hispanic than non-Hispanic white mortality among the elderly. However, the Hispanic mortality advantage using these data is not nearly as wide as evidenced in the officially reported data seen in Table 3-1. Finally, API elderly mortality is significantly lower than non-Hispanic whites throughout the age range, with evidence of convergence to non-Hispanic white levels among the oldest-old. These disparities are quite similar to those reported by Elo and Preston (1997), who used survey-based (Current Population Surveys) follow-up mortality data from the 1979-1985 National Longitudinal Mortality Study data set. Perhaps the only difference of note is that the Hispanic groups presented in Table 3-3 exhibit only moderately (e.g., 13 to 16 percent) lower mortality than non-Hispanic whites, as opposed to the larger advantages (e.g., 21 to 37 percent) reported by Elo and Preston from the earlier time period.

In sum, the mortality results show that racial/ethnic disparities remain relatively unchanged in comparison to the results of Elo and Preston (1997), who used data from approximately a decade earlier. Non-Hispanic black mortality is roughly 40 percent higher at age 65 than non-Hispanic whites, with this disparity decreasing and crossing over at the very old ages. The mortality advantage for Hispanic elders compared to non-Hispanic whites is probably closer to 15 percent, as seen in the survey-based data set findings, rather than the officially reported 25 to 50 percent, although selective out-migration and poorer quality death matches may still be important, and unaccounted for, biasing factors, particularly for Mexican Americans. This Hispanic advantage also has been reported by a number of others at the national level (Hummer et al., 2000; Liao et al., 1998; Sorlie, Backlund, Johnson, and Rogot, 1993). API elderly mortality appears to be the most favorable of all, as consistently found now across several data sources (also see Elo and Preston, 1997; Lauderdale and Kestenbaum, 2002; Rogers, Hummer, Nam, and Peters, 1996), although some subpopulations (e.g., Native Hawaiians, Samoans) may experience higher mortality (Hoyert and Kung, 1997). A major question revolves around Native American elder mortality compared to non-Hispanic whites. Although officially reported data show large advantages for Native American elders, data quality issues are believed to be of greatest concern for this racial/ethnic group (Rosenberg et al., 1999; Stehr-Green et al., 2002). The survey-based results suggest that, at the very least, Native American older adult mortality is not lower than non-Hispanic whites. Nonetheless, data quality issues for Native Americans, as well as for Hispanics and APIs, should be kept in mind. Furthermore, note that these racial/ethnic groups are both internally very heterogeneous (e.g., by level of socioeconomic status, geographic dispersion, history of discrimination, cultural background) and, for Hispanics and APIs in particular, composed of a large percentage of immigrants. This heterogeneity is discussed in several chapters throughout this volume.

RACIAL/ETHNIC DISPARITIES IN HEALTH AND ACTIVITY LIMITATIONS AMONG THE ELDERLY

Disability and Active Life Expectancy

While mortality studies provide one set of evidence regarding racial/ ethnic patterns of health, a number of other outcome variables help to round out the general picture of health disparities among the U.S. elderly population. As with a great deal of research on mortality disparities, much work in the health and activity limitations area has compared the black and white populations, with some attention in recent work given to Hispanics (for excellent recent examples using nationally representative data, see Hayward, Crimmins, Miles, and Yang, 2000; Manton and Gu, 2001; Manton and Stallard, 1997; Smith and Kington, 1997a, 1997b). An important exception is the recent work of Hayward and Heron (1999), who utilized life table models with Public Use Microdata Sample data from the 1990 U.S. Census (to obtain disability estimates) and U.S. Vital Statistics data (to obtain mortality estimates) to compare patterns of disability and active life expectancy across five racial/ethnic populations. Measures of active life expectancy gauge “the number of years individuals can expect to live without a limitation of activity resulting from chronic disease or impairment” (Hayward and Heron, 1999, p. 77). Thus, although this particular measure is bounded by life expectancy for all racial/ethnic groups, it can fluctuate quite extensively based on the relative health and disability levels of a population group throughout the life course.

A subset of the Hayward and Heron (1999) findings is summarized in Table 3-4 (also see Figure 3-3). These results show substantial differences in disability prevalence and active life for U.S. elders by race/ethnicity. Panel A of Table 3-4, which shows the findings for men, demonstrates that Native Americans and blacks have the highest disability prevalence at each age group. Furthermore, these two groups have the fewest years of active life remaining (with the exception of Native Americans at age 80) and the lowest percentage of remaining active life at each specific age. On the other hand, API elderly men have the lowest disability prevalence at each age group and the highest level and percentage of active life remaining at each specific age. Finally, Hispanic males exhibit slightly lower levels of disability and higher percentages of remaining active life than non-Hispanic whites.

TABLE 3-4

Disability Prevalence Rates and Estimates of Active Life Expectancy by Race/Ethnicity, Age, and Sex, United States, 1990.

FIGURE 3-3

Racial/ethnic disparities in disability prevalence by age group, U.S. elderly population, 1990.

While disability prevalence for women (Panel B, Table 3-4; also see Figure 3-3) tends to be somewhat lower than men at age group 60 to 64 and higher than men at age group 80 to 84, racial/ethnic patterns are similar to men. That is, Native Americans and blacks exhibit the highest levels of disability at each age group among the elderly, and API rates are the lowest. Again, Hispanic and white patterns are quite similar. In all, however, the findings emphasize the relative good health of the API elderly population and the high levels of disability and inactive elder life among the black and Native American populations. Notably, the relatively low mortality rates documented for the Native American population, as reported in official mortality data, are contrasted with their very high levels of disability and inactive life among the elderly (Hayward and Heron, 1999, p. 88). However, the survey-based mortality results from Native Americans (from Table 3-3) correspond much more closely to these disability results. The high levels of disability among the black population corresponds with their high levels of mortality, although it is important to note that one study recently documented a sharper decline in black disability compared to nonblacks between 1994 and 1999 (Manton and Gu, 2001).

Self-Reported Health and Activity Limitations

Table 3-5 presents data on self-reported health status and activity limitations by race/ethnicity, drawn from 1989 to 1994 pooled National Health Interview Surveys, allowing for the specification of patterns across relatively small racial/ethnic populations. Self-reported health status is a five-point (excellent, very good, good, fair, poor), easily collected, and frequently utilized measure of general health status. Self-reported health has been demonstrated to have a powerful influence on subsequent mortality risk for the U.S. elderly population (Idler and Benyamini, 1997; Rogers et al., 2000) and for a number of racial/ethnic groups (McGee, Liao, Cao, and Cooper, 1999), although the predictive effect for subsequent mortality risk among Hispanic immigrants is much weaker (Finch, Hummer, Reindl, and Vega, 2002). Ratings of “poor” and “fair” are particularly risky categories when gauged with subsequent probabilities of survival (Rogers et al., 2000).

TABLE 3-5

Self-Reported Health and Activity Limitations by Race/ Ethnicity, Sex, and Age Group, United States, 1989-1994.

At the descriptive level, Panel A of Table 3-5 shows wide variation in male self-reported health status by race/ethnicity. While just 18.2 percent of API adults aged 65 to 74 report fair or poor health, 35.4 percent of Native Americans and 41.3 percent of non-Hispanic blacks do so. Mexican American elders aged 75 to 84 and 85+ also exhibit quite high levels of poor and fair self-reports, although this may be due, at least in part, to the generally lower levels of health reported among the Hispanic immigrant population (Angel and Guarnaccia, 1989; Cho, Frisbie, Hummer, and Rogers, in press; Shetterly, Baxter, Mason, and Hamman, 1996). It is clear that API and non-Hispanic white individuals exhibit the most favorable levels of self-reported health, while non-Hispanic blacks, Native Americans, and Mexican Americans display the least favorable results. Female patterns of self-reported health status by race/ethnicity (Panel B, Table 3-5) do not differ appreciably from the male patterns.

For activity limitations, the measure utilized in Table 3-5 groups together individuals who reported one of the following: (1) they were unable to perform their major activity, (2) they were limited in their major activity, or (3) they were limited in other activities. Each of these categories has been linked to higher subsequent mortality risk in a recent follow-up study, with the “limited in other activities” category displaying the most modest mortality effects (Rogers et al., 2000, pp. 190-191). Nevertheless, we group them together here in order to provide estimates of racial/ethnic disparities in activity limitation for the five population subgroups under consideration, by specific age group. Recall, too, that these data are nationally representative of the noninstitutionalized U.S. elderly population.

Similar to the Hayward and Heron (1999) disability findings abstracted in Table 3-4, the NHIS findings from Table 3-5 show that, for both men and women, non-Hispanic blacks and Native Americans are most apt to report limitations at ages 65 to 74, while API individuals are least likely to report limitations. Both male and female non-Hispanic blacks continue to display relatively unfavorable patterns among the oldest age groups as well, with the percentage of activity-limited persons being the highest or next-to-highest when compared to the other racial/ethnic groups. Aside from the API population, non-Hispanic whites generally exhibit the most favorable patterns across each of the age groups considered. The activity limitation patterns of Mexican Americans and other Hispanics are more favorable than those of Native Americans and blacks at ages 65 to 74, but less favorable at ages 85+. However, the Mexican American and other Hispanic figures for the oldest-old, as well as those for Native Americans and the API population, must be considered with great caution because of the very small sample sizes of these groups in the NHIS, even when using 6 years of pooled data.

In sum, several sources of health and activity limitation data point to the relatively favorable profiles of API and non-Hispanic white elders, respectively, in comparison to the other racial/ethnic groups. For non-Hispanic black, Native American, and, to a lesser degree, Hispanic elders, the portrait is much less favorable. These summary health and activity limitation results by race/ethnicity are not entirely consistent with those shown for mortality earlier, where Hispanics exhibited favorable patterns compared to non-Hispanic whites, and Native Americans, depending on the data source, displayed either lower or modestly higher levels of mortality compared to elderly non-Hispanic whites. Such inconsistencies between the mortality and health/activity limitation findings could be due to a number of data quality and reliability issues, including the mortality data limitations already discussed. The possible selective nature of the migration process, both in terms of immigration and emigration, may also have strong impacts on the patterns of health and mortality for elderly Hispanics and APIs. Furthermore, cultural and language differences in the interpretation and responses to the health questions in survey data may influence those findings as well. Despite these ambiguities, the weight of the evidence strongly suggests that black and Native American elders are the least well off in terms of health and mortality, while API elders tend to be the most advantaged across most of the outcomes examined. Hispanics, while displaying modestly lower elder mortality than non-Hispanic whites, exhibited higher levels of poor and fair health, slightly lower levels of active life expectancy, and a higher level of activity limitations than non-Hispanic whites.

COMPARING RACIAL/ETHNIC DISPARITIES AMONG THE ELDERLY WITH THOSE EXHIBITED BY YOUNGER AGE GROUPS

Mortality

The mortality and health disparities described previously focus on the elderly (aged 65+, or in some cases, aged 60+) population, based on findings from several nationally representative data sets. One important question that emerges is how these relative disparities compare with those exhibited during earlier parts of the life course. We first turn to officially reported mortality rates for different age groups, shown in Table 3-6. Mortality rates for 1999 are shown for U.S. infants, and for younger adults aged 20 to 24, 40 to 44, and 60 to 64, with the older adult rates for those aged 70 to 74 and 80 to 84 displayed for context. Males and females are combined in this table for the sake of parsimony.

TABLE 3-6

A Comparison of Racial/Ethnic Disparities in Mortality for Different U.S. Age Groups, 1999.

Rates for blacks are more than twice as high as whites during infancy and young adulthood, narrow but are still nearly twice as high during middle-aged adulthood, and then converge in older adulthood to near equity among persons aged 80 and over. These results are consistent with numerous NCHS publications over the years, and with other data sources in a number of major mortality studies over the last 30 years (Kitagawa and Hauser, 1973; Rogers et al., 2000; Sorlie, Backlund, and Keller, 1995). Major cause of death contributors for the especially high levels of black mortality during younger adulthood include elevated levels of homicide and infectious disease mortality relative to whites (Anderson, 2001; Rogers, 1992). Earlier-age major onsets of diabetes-, heart disease-, and cancer-related mortality among middle-aged blacks relative to whites are even more important in understanding this pattern. For example, non-Hispanic black adults aged 55 to 64 suffered from 3.11 times higher mortality due to diabetes, 1.96 times higher heart disease mortality, and 1.44 times higher cancer mortality compared to non-Hispanic whites aged 55 to 64 in 1999; these relative cause of death differences were 2.16, 1.23, and 1.22, respectively, at ages 75 to 84 (Anderson, 2001).

Officially reported Hispanic mortality is equal to or lower than that of non-Hispanic whites across the life course, with the exception of young adulthood (Table 3-6). As with non-Hispanic blacks, elevated risks of homicide and infectious disease mortality help to account for this excess young adult Hispanic mortality (Anderson, 2001; Hummer et al., 2000; Rosenwaike, 1991). Native American infant and younger adult mortality is also significantly higher than non-Hispanic whites, a pattern that is divergent from that shown for older adults vis-à-vis non-Hispanic whites. The exceptionally favorable pattern for the Native American elderly population, again, must be viewed as highly suspect; clearly, our survey-based results for older adult mortality do not show this same pattern. API mortality rates are the lowest of these racial/ethnic groups at each age group. Although data quality issues are at least partially accountable for the reported low mortality rates for the API population (Rosenberg et al., 1999), low API infant mortality provides additional evidence of this group's overall healthy profile. Indeed, racial/ethnic patterns of infant mortality are less subject to misreports of race/ethnicity in comparison to patterns of adult mortality because infant death statistics are based on maternal race/ethnicity, which is reported on the infant's birth certificate in the mother's presence at the time of birth (Hummer et al., 1999a). Infant mortality rate equity between Hispanics and non-Hispanic whites, likewise, is strong evidence for the relatively favorable overall mortality experience of Hispanics at the national level.

Health and Activity Limitations

Table 3-7 turns to health and activity limitation disparities at different ages, using pooled data from the 1989-1994 National Health Interview Surveys. Panel A, which focuses on self-reported health, demonstrates that non-Hispanic whites report the most favorable health for most age groups examined. For example, for adults aged 60 to 64, non-Hispanics blacks are 1.9 times as likely to report fair or poor health as non-Hispanic whites, Mexican Americans are 1.7 times as likely, other Hispanics are 1.5 times as likely, and Native Americans are 1.8 times as likely. Differences between the API and non-Hispanic white populations are modest at most ages, although the API population tends to report slightly worse health when compared to non-Hispanic whites. As is the case with mortality by age, the racial/ethnic disparities in self-reported health are largest during young adulthood and are narrower among the elderly, although unlike the mortality rates shown earlier, non-Hispanic whites demonstrate the most favorable self-reported health, even in old age.

TABLE 3-7

A Comparison of Racial/Ethnic Disparities in Self-Reported Health and Activity Limitations for Different Age Groups, United States, 1989-1994.

For most age groups, non-Hispanic blacks and Native Americans tend to report the poorest health. Mexican Americans and other Hispanics are not very different from one another or from Native Americans on this measure, reporting relatively poor health in comparison to non-Hispanic whites at every age (Markides, Rudkin, Angel, and Espino, 1997). The subjectivity involved in measures of self-reported health, particularly across racial/ethnic and highly concentrated immigrant populations (Angel and Guarnaccia, 1989; Finch et al., 2002), should temper any strong conclusions being made from such direct comparisons. Nevertheless, self-reported health has been demonstrated to be a strong predictor of subsequent mortality risk for each of these racial/ethnic groups at the national level, even net of socioeconomic and other health controls (McGee et al., 1999). Therefore, this measure must be considered yet another important indicator of the overall health of these populations.

Panel B of Table 3-7 turns to age-related racial/ethnic disparities in activity limitations. As with ratings of self-reported health, these assessments tend to be much less favorable when considering the older age groups, as might be expected. However, racial/ethnic disparities for this measure are much narrower than those exhibited for self-reported health and, like the mortality disparities shown in Table 3-6, favor APIs at each age group (see Hayward and Heron, 1999, for a similar comparison across younger and older age groups that also shows favorable rates of activity limitation among Asian Americans). Interestingly, Mexican Americans are characterized by their most favorable level of activity limitations compared to non-Hispanic whites at ages 20 to 24, when their mortality rates are highest compared to non-Hispanic whites. At ages 60 to 64 and 70 to 74, reports of activity limitations are modestly higher among Mexican Americans compared to non-Hispanic whites (Markides et al., 1997). Finally, for most age groups, non-Hispanic blacks and Native Americans again exhibit the least favorable results, with differentials reduced somewhat among the elderly groups, but remaining moderate in size.

HOW DO HEALTH AND MORTALITY DISPARITIES AMONG THE ELDERLY CORRESPOND WITH DIFFERENTIALS IN SOCIODEMOGRAPHIC FACTORS?

Another aim of this chapter is to compare the extent to which the observed racial/ethnic health and mortality disparities among the elderly correspond with differentials in sociodemographic factors across these population groups. Table 3-8 and Figures 3-4 through 3-6 summarize this endeavor by using pooled NHIS data from 1989 to 1994 to examine four key factors that have been linked to the health and mortality patterns of U.S. adults: gender, nativity/duration of residence in the United States, educational level, and family income. NHIS data are used to correspond with the health and mortality models that are estimated in the next section of this chapter. Thus, it should be kept in mind that these data are representative of the noninstitutionalized older adult population and do not reflect the total population of U.S. elders. The sociodemographic factors are examined for the elderly population as a whole (ages 65+, labeled “overall” in the table), as well as for the 65 to 74 and 75+ subgroups.

TABLE 3-8

Percentage Distributions for Selected Demographic and Socioeconomic Variables by Race/Ethnicity for Individuals 65 Years and Above, 1989-1994.

FIGURE 3-4

Nativity and duration by race/ethnicity for individuals 65 and older, U.S. noninstitutionalized population, 1989-1994.

FIGURE 3-6

Income level by race/ethnicity for individuals 65 and older, U.S. noninstitutionalized population, 1989-1994.

As might be expected, each racial/ethnic population includes more females than males, with this gender disparity larger among the oldest old. The descriptive tabulations demonstrate much greater heterogeneity in nativity/duration patterns by race/ethnicity (Figure 3-4). Consistent with recent immigration patterns, API (65.4 percent), other Hispanic (67.8 percent), and Mexican American (39.3 percent) elders are far more likely to be foreign born in comparison to the other groups. Interestingly, though, there is wide variation when comparing Mexican-Americans, other Hispanics, and APIs. Indeed, other Hispanic and API elders are the least likely to be U.S. born, and API individuals are most likely to be of short duration in the United States (e.g., 23.3 percent have resided in the United States less than 10 years). Thus, the health and socioeconomic characteristics of recent elderly migrants from Asia have vast demographic potential to make a major impact on this group's health and mortality patterns. On the other hand, a majority of Mexican American elders are U.S. born and less than 3 percent of this elderly group have resided in the United States for less than 10 years. Indeed, an overwhelming majority of Mexican American and other Hispanic foreign-born elderly residents report residing in the United States for 10 or more years. Among the predominantly U.S.-born racial/ethnic groups, non-Hispanic white (6.1 percent) elders are more likely to be foreign born than non-Hispanic blacks (3.2 percent) or Native Americans (4.2 percent), with most non-Hispanic white foreign-born elders residing in the United States for 10 or more years. Note that for Native Americans, foreign-born individuals are presumably those who have migrated from Canada or Latin America.

Nativity/duration has been shown to be associated with adult health (Cho et al., in press; Frisbie, Cho, and Hummer, 2001) and mortality risks (Elo and Preston, 1997; Hummer et al., 1999b; Kestenbaum, 1986; Rogers et al., 2000; Singh and Siahpush, 2001) within a variety of racial/ethnic populations at the national level. For the most part, immigrants to the United States have been shown to exhibit favorable levels of adult health and mortality in comparison to the native-born population (with cause-specific mortality exceptions; see, e.g., Singh and Siahpush, 2001; Toussaint and Hummer, 1999), with foreign-native born disparities usually reported to be wider for health outcomes than for mortality risk. Explanations for favorable immigrant health and mortality include the selectivity of healthy individuals migrating to the United States, the selectivity of unhealthy individuals emigrating from the United States, the protective effects of the culture of origin from various sending countries, and the negative influences of U.S. culture, particularly for health behaviors such as cigarette smoking and dietary intake. The favorable levels of health and mortality for immigrants may also wear away with increasing time spent in the United States (Cho et al., in press; Frisbie et al., 2001), although few studies have had the necessary longitudinal data at the national level to directly test this proposition. Chapter 7 in this volume deals with the issue of immigrant health in much greater detail.

Differences in education (Figure 3-5) and family income (Figure 3-6) across these racial/ethnic groups are also substantial and, most notably in terms of education, do not always correspond with the health and mortality disparities outlined previously. Mexican American elders have the lowest educational levels of any of these groups: 71.6 percent attended 8 or fewer years of school, with the percentage even higher among Mexican American individuals aged 75 and older. More than half of Native American, non-Hispanic black, and other Hispanic elders also have fewer than 9 years of schooling, although again, age group differences within the elderly are substantial given generally increasing levels of schooling throughout the 20th century for all racial/ethnic groups. Corresponding more closely with their relative health and mortality profiles, non-Hispanic white elders (52 percent), followed by APIs (46.7 percent), are most likely to have completed at least 12 years of schooling, with this pattern consistent both for the 65 to 74 and 75+ groups. API elders (12.7 percent) exhibit the highest percentage of 16+ years of education.

FIGURE 3-5

Years of education by race/ethnicity for individuals 65 and older, U.S. noninstitutionalized population, 1989-1994.

Levels of family income by race/ethnicity correspond quite well with the health and mortality patterns of the elderly. Non-Hispanic black and Native American elders are characterized by very high percentages of individuals who live in the lowest family income category ($0-15,999), both overall and for the 65 to 74 and 75+ subgroups. Correspondingly, individuals in these two groups are also least likely to live in families in the highest income category. Mexican American elders also exhibit a highly unfavorable income distribution compared to non-Hispanic whites and APIs. Indeed, API individuals report living in the most favorable income situation of any racial/ethnic group, with just 23.7 percent reporting living in the low category and 31.2 percent reporting living in the highest ($45,000+) category. For all racial/ethnic groups, income levels are more favorable among individuals ages 65 to 74 compared to those 75+.

In sum, the descriptive data show especially wide nativity/duration and socioeconomic differences across racial/ethnic groups. The educational and income variables clearly show the socioeconomically advantaged position of the non-Hispanic white and API populations, with the income advantage for non-Hispanic whites and APIs particularly striking in comparison to non-Hispanic blacks, Hispanics, and Native Americans. On the other hand, non-Hispanic black, Native American, and Mexican American elders exhibit much worse off socioeconomic profiles that, at least for blacks and Native Americans, correspond with their health outcomes. Widely differing nativity/duration profiles, even among the heavily immigrant populations, also point to the substantial complexity of understanding health and mortality patterns across these groups.

BASIC RACIAL/ETHNIC MODELS OF OLDER ADULT HEALTH, ACTIVITY LIMITATIONS, AND MORTALITY

The last major section of this chapter presents basic models of racial/ ethnic disparities in older adult health, activity limitations, and mortality, using data taken from pooled samples of the National Health Interview Survey from 1989 to 1994. Because of its size and the ability to pool multiple years together, this is one of the only nationally representative data sets that allow for modeling of health outcomes for the elderly population across the racial/ethnic groups under consideration here. Indeed, Preston and Taubman (1994, p. 291) described the NHIS as the “most authoritative source of national data on socioeconomic differences in health status.” We note that the same is true for examining racial/ethnic health disparities at the national level. Here, we focus on the influences of basic demographic and socioeconomic factors on the racial/ethnic disparities, and particularly how controlling for such factors in a progressive manner (e.g., Mirowsky, 1999) influences the baseline differences.

Table 3-9 focuses on racial/ethnic disparities in self-reported health status among the elderly, using logistic regression models that contrast those individuals with poor/fair self-reported health versus those with good/ very good/excellent health. Model 1 shows the baseline racial/ethnic disparities in comparison to non-Hispanic whites, controlling only for age and sex. Non-Hispanic blacks, Native Americans, Mexican Americans, and other Hispanics all exhibit higher odds of poor/fair self-reported health compared to non-Hispanic whites, while the API population demonstrates lower odds. Interestingly, these disparities are largely unaffected with the addition of a race/ethnicity by age interaction term (Model 2), although it is the case that the non-Hispanic black-white disparity is largest at age 65 and decreases with age. The inclusion of nativity/duration (Model 3) also has a limited impact on the racial/ethnic health differences. In this case, the foreign born of less than 10 years in the United States are moderately (43 percent) more likely to report poor/fair health compared to the U.S.-born elderly, which may be due to the interpretation of, and response categories for, the self-reported health question (Angel and Guarnaccia, 1989; Finch et al., 2002).

TABLE 3-9

Odds Ratios of Racial/Ethnic Differences in Elderly Self-Reported Health, United States, 1989-1994.

Model 4 adds educational level. Education is an especially important indicator of socioeconomic status for the elderly because it is usually determined early in life, can be assessed quite easily for all individuals, influences health behavior and the use of health services, and is very important in influencing occupational position and the generation of wealth throughout the life course (Preston and Taubman, 1994). Thus, a great deal of research has assessed educational differences in the health and mortality of U.S. adults and has continued to find sizable, graded differences (Christenson and Johnson, 1995; Elo and Preston, 1996; Feldman, Makuc, Kleinman, and Cornoni-Huntley, 1989; Freedman and Martin, 1999; Rogers et al., 2000; Ross and Mirowsky, 1999; Smith and Kington, 1997a, 1997b), with some studies even pointing to growing inequalities since 1960 (Pappas, Queen, Hadden, and Fisher, 1993; Preston and Elo, 1995). For the elderly population considered in Table 3-9, the educational differences in self-reported health are graded and wide; for example, individuals with 0-8 years of education are 2.6 times as likely to indicate poor/fair self-reported health compared to individuals with 16+ years of education. Furthermore, the addition of education results in a measurable reduction in the racial/ethnic disparities in self-reported health, particularly for non-Hispanic blacks, Mexican Americans, and Native Americans compared to non-Hispanic whites. However, even with the addition of this powerful socioeconomic variable, sizable racial/ethnic disparities remain in self-reported health, particularly between non-Hispanic blacks and whites.

Model 5 additionally considers family income. While recent studies on the health of the elderly have justifiably criticized reliance on household income as the proxy for the availability of household economic resources (Smith and Kington, 1997a, 1997b), the yearly NHIS contains no consistent measure of household assets, liabilities, or overall net worth. Thus, there remain important, unmeasured components of socioeconomic inequality in this set of models (also see Kaufman, Cooper, and McGee, 1997, for a critique of socioeconomic measurement in studies of racial/ethnic health disparities). Furthermore, caution is warranted in interpreting the effects of income as fully causal; indeed, substantial literature points to the effects of health status on income as well as income on health (e.g., Mulatu and Schooler, 2002; Smith and Kington, 1997b).

The results for income are consistent with those for education in that a strong, graded association is evident, with individuals in the lowest family income category more than twice as likely to be in poor/fair health compared to individuals in the highest income category. Racial/ethnic disparities in self-reported health are further reduced with the addition of income, although significant differences remain between non-Hispanic blacks and non-Hispanic whites and between Mexican Americans and non-Hispanic whites. Nevertheless, it is clear that socioeconomic factors are instrumental in helping to account for the relatively poor self-reported health for non-Hispanic black, Mexican American, other Hispanic, and Native American elders in comparison to non-Hispanic whites. In contrast, API elders display marginally favorable self-reported health compared to non-Hispanic white elders, with education and family income differences between the API and non-Hispanic white groups having little impact.

Table 3-10 turns to racial/ethnic disparities in activity limitations for the elderly. Model 1 shows that API adults exhibit the most favorable patterns, with more than 40 percent lower odds of limitation compared with non-Hispanic whites. This baseline advantage is relatively unaffected by controls for nativity/duration and socioeconomic factors. On the other hand, Native Americans, non-Hispanic blacks, and Mexican Americans exhibit significantly higher levels of activity limitations than non-Hispanic whites that are, again, largely unaffected by controlling for nativity/duration, but substantially reduced with controls for education and income. In fact, the Mexican American difference with non-Hispanic whites is reduced to nonsignificance after controlling for education and income differences across groups, while the higher odds for non-Hispanic blacks and Native Americans are substantially reduced. Again, the important role of socioeconomic variables for poorer health outcomes among elderly Native Americans, non-Hispanic blacks, and Mexican Americans is demonstrated.

TABLE 3-10

Odds Ratios for Racial/Ethnic Differences in Elderly Activity Limitations, United States, 1989-1994.

Finally, Table 3-11 displays results from proportional hazard models of racial/ethnic disparities in elder mortality risk. As exhibited in the descriptive portion of this chapter earlier, mortality disparities display somewhat different patterns than health disparities among the elderly. Model 1, for example, shows that non-Hispanic blacks exhibit just a 13 percent higher overall risk of mortality than non-Hispanic whites, controlling for age and sex, while Mexican Americans, other Hispanics, and, especially, APIs exhibit lower elderly mortality than non-Hispanic whites. The age by racial/ ethnicity interactions shown in Model 2 reconfirm the converging black-white and API-white mortality disparities with age and demonstrate the substantial mortality disadvantage for blacks and mortality advantage for APIs, relative to non-Hispanic whites, at age 65. For example, at age 65, Model 2 demonstrates that non-Hispanic blacks exhibit a 41 percent higher risk of mortality compared to non-Hispanic whites. The nativity/duration association with mortality risk, exhibited in Model 3, is moderately strong and maintains that strength through the complete set of models. The foreign born of less than 10 years of duration in the United States exhibit a 26 percent lower risk of mortality compared to the native born, while the foreign born of 10 or more years of duration display an 18 percent lower risk compared to the native born. Again, the interpretation of these differentials is difficult because of immigrant and emigrant migration selectivity as well as the generally higher quality of death matches among the native-born population in comparison to the foreign born. Interestingly, though, the racial/ethnic disparities are only modestly influenced with the control for nativity/duration, demonstrating that the observed, rather favorable, elder mortality patterns for Mexican Americans, other Hispanics, and APIs are not solely due to the substantial foreign-born composition of these groups.

TABLE 3-11

Hazard Ratios for Racial/Ethnic Differences in Elderly Mortality Risk, United States, 1989-1997.

Models 4 and 5 further include education and family income, respectively. Although each of these factors exhibits a graded association with mortality risk in the expected direction, the socioeconomic differences are moderate in size and of weaker magnitude than for the health outcomes examined earlier. The somewhat weaker association of socioeconomic factors with older adult mortality has been noted in a number of national studies (e.g., Kitagawa and Hauser, 1973; Rogers et al., 2000), although the explanation for this phenomenon remains elusive. Moreover, although controlling for education and income in Models 4 and 5 works to help reduce the black-white mortality gap, differences between these two groups remain. Moreover, the relative mortality disparities between the other racial/ethnic populations and whites change only slightly with the introduction of the socioeconomic factors.

CONCLUSION

Despite the national-level priority on understanding and eliminating health disparities, we know far less about racial/ethnic differences in older adult health, activity limitations, and mortality than is the case among infants, children, and younger adults. It is imperative that the research community push for a greater understanding of these disparities, particularly given the continuing documentation of disparities across groups, the changing racial/ethnic composition of the nation, and an aging population. This chapter set out to document racial/ethnic disparities in older adult health and mortality using large, recent, nationally based data sets; to compare the disparities to earlier portions of the life course; to document how the health and mortality disparities correspond with racial/ethnic differences in sociodemographic factors; and to demonstrate how health and mortality disparities are influenced by controlling for those basic demographic and social factors.

To briefly summarize the findings, we have documented continuing racial/ethnic disparities in health, activity limitations, and active life among U.S. elders. Non-Hispanic black, Native American, and, to a lesser degree, Mexican American and other Hispanic elders were shown to have overall worse health across a number of indicators compared to non-Hispanic whites, while API elders displayed more favorable patterns across a number of the indicators. Mortality disparities were found to be less pronounced than health disparities among the elderly, although non-Hispanic blacks were clearly found to have the highest mortality risks using several different data sources. Excess black mortality, in comparison to whites, is concentrated among the younger elderly population, with negligible differences beyond age 80. Nevertheless, a comparison of the present mortality differentials with those of Elo and Preston (1997) from roughly a decade previous show no evidence of closure of the black-white mortality gap among the younger elderly population, even in the recent context of declining mortality for all groups. Racial/ethnic disparities in elderly health and mortality were also found to be generally of smaller magnitude than the disparities shown for younger U.S. adults, although levels of poor health, activity limitations, and mortality risks increase for all racial/ethnic groups with age.

This chapter also documented wide socioeconomic differences between racial/ethnic groups in old age. Our models also showed that education and income differences across groups continue to play an important part in the overall worse health of non-Hispanic blacks, Native Americans, and to a lesser degree, Hispanics in comparison to non-Hispanic whites and the API population. Other demographic and social factors were shown to differ quite markedly across groups. The widely varying immigration experiences across racial/ethnic groups will continue to differentiate Hispanic and API elders from blacks, whites, and Native Americans. Further nativity/duration differences were noted even within the highly concentrated immigrant populations; for example, API foreign-born elders are much more likely to have recently migrated to the United States in comparison to Hispanic foreign-born elders.

In our discussion and analyses, we have focused on the associations between sociodemographic factors, including age, sex, nativity, education, and income, and racial/ethnic disparities in health and mortality. But we must not overlook other important factors that may be associated with these disparities, such as health behaviors—tobacco use, alcohol consumption, diet, and exercise; health conditions—such as obesity and diabetes; mental and addictive disorders—such as drug abuse; and environmental and neighborhood factors—including crime and safety (Rogers, Hummer, and Krueger, 2004). Thus, although we have identified some of the major social and demographic factors that influence racial/ethnic health and mortality patterns, there are surely other areas of research that can further clarify the mechanisms that contribute to these patterns. Importantly, untangling the causal associations between socioeconomic factors and health outcomes, and understanding these associations within each racial/ethnic subgroup, should be a major research priority in the coming years.

Important data availability and reliability issues continue to hinder the more complete understanding of racial/ethnic health and mortality disparities among the elderly at the national level. Indeed, much of what is known about mortality among the elderly is based on vital statistics data, which have well-known limitations (e.g., very limited range of covariates, some mismatches of racial/ethnic reports between numerator and denominator data, and undercounts of minority groups in the denominator data), particularly for understanding minority group mortality levels and patterns. Matches between large survey-based data sets and mortality follow-up records (e.g., the NHIS-MCD, the National Longitudinal Mortality Study) have vastly improved our understanding in this area, but even those data sources have limited sets of variables to allow for the full comprehension of mortality patterns across racial/ethnic groups, as well as coverage gaps and matching problems. Nevertheless, such large survey-based data sources currently present the greatest opportunities for most thoroughly documenting and modeling racial/ethnic patterns of mortality across a wide variety of racial/ethnic groups.

Longitudinal studies of health among the elderly have been limited, with a major recent improvement made by the Health and Retirement Study (HRS) data collection program. Still, even the HRS surveys offer very limited numbers of Hispanics, Asian Americans, and Native Americans, undermining their utility for understanding the full spectrum of racial/ ethnic disparities in older adult health. Understanding Hispanic and API older adult health and mortality will be a substantial future challenge, particularly with continued large-scale migration and the possibility of circular and/or return migration to Mexico, other countries of Latin America, and Asia.

The diverse ethnic, cultural, and geographic factors that characterize all of these broad racial/ethnic groups, but perhaps especially the Hispanic and API populations, will pose a real challenge to the understanding of elderly health and mortality patterns in the United States. Racial/ethnic groups are characterized by internal variation by language, religion, geographic dispersion, immigration patterns, socioeconomic status, health behaviors, and other factors. Indeed, Hispanics include Mexican Americans, Cubans, Puerto Ricans, Central and South Americans, and other persons of Spanish descent. APIs include Chinese, Japanese, Filipinos, Indians and Pakistanis, Vietnamese, Koreans, Hawaiians, and many other individuals from varying backgrounds. Currently, we know relatively little about the health and mortality patterns of most of the component ethnic populations of the broad racial/ethnic groups, and documenting and understanding such patterns will be an important research challenge in the coming decades.

Perhaps most important, racial/ethnic disparities in health and mortality among U.S. elders cannot be fully understood without placing such patterns in a life-course context. The research community will need much better data to understand how life-course factors (e.g., migration experiences, socioeconomic fluctuations, childhood and early adult health and illness, family backgrounds, discriminatory histories) impact older adult health and mortality across various racial/ethnic groups. Disparate patterns of older adult health and mortality among racial/ethnic groups come about because of the varying demographic, social, economic, behavioral, and health and health care experiences of these groups over many years; thus, our more complete understanding of old-age health consequences relies on better tapping this differential accumulation of experiences.

In conclusion, the United States experienced a remarkable mortality decline and great improvements in health across the 20th century. All racial/ethnic groups participated in and contributed to these substantial changes, although some important health and mortality disparities remain. With further improvements in standards of living, safety and security, nutrition, and medicine, we can anticipate further health gains and longer lives among all racial/ethnic groups in the coming decades as well, and we can hope that Healthy People 2010's goal of eliminating health disparities among racial/ethnic groups becomes a reality.

REFERENCES

1. Abraido-Lanza AF, Dohrenwend BP, Ng-Mak DS, Turner JB. The Latino mortality paradox: A test of the “salmon bias” and healthy migrant hypotheses. American Journal of Public Health. 1999;89:1543–1548. [PMC free article: PMC1508801] [PubMed: 10511837]
2. Anderson RN. Deaths: Leading causes for 1999. National Vital Statistics Reports. 2001;49(11) [PubMed: 11682979]
3. Angel R, Guarnaccia PJ. Mind, body, and culture: Somatization among Hispanics. Social Science and Medicine. 1989;23:1229–1238. [PubMed: 2660279]
4. Cho Y, Frisbie WP, Hummer RA, Rogers RG. Nativity, duration of residence, and the health of Hispanic adults in the United States. International Migration Review. in press.
5. Christenson BA, Johnson NE. Educational inequality in adult mortality: An assessment with death certificate data from Michigan. Demography. 1995;32:215–230. [PubMed: 7664961]
6. Coale AJ, Kisker EE. Mortality crossover: Reality or bad data? Population Studies. 1986;40:389–401.
7. Elo IT. New African American life tables from 1935-1940 to 1985-1990. Demography. 2001;38:97–114. [PubMed: 11227848]
8. Elo IT, Preston SH. Educational differentials in mortality: United States, 1979-1985. Social Science and Medicine. 1996;42:47–57. [PubMed: 8745107]
9. Elo IT, Preston SH. Racial and ethnic differences in mortality at older ages. In: Martin LG, Soldo BJ, editors. Racial and ethnic differences in the health of older Americans. Washington, DC: National Academy Press; 1997. pp. 10–42. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council.
10. Feldman J, Makuc D, Kleinman J, Cornoni-Huntley J. National trends in educational differences in mortality. American Journal of Epidemiology. 1989;129:919–933. [PubMed: 2705434]
11. Finch BK, Hummer RA, Reindl M, Vega WA. Validity of self-rated health among Latino(a)s. American Journal of Epidemiology. 2002;155(8):755–759. [PubMed: 11943694]
12. Freedman VA, Martin LG. The role of education in explaining and forecasting trends in functional limitations among older Americans. Demography. 1999;36:461–473. [PubMed: 10604075]
13. Frisbie WP, Cho Y, Hummer RA. Immigration and the health of Asian and Pacific Islander adults in the U.S. American Journal of Epidemiology. 2001;153:372–380. [PubMed: 11207155]
14. Hayward MD, Heron M. Racial inequality in active life among adult Americans. Demography. 1999;36:77–91. [PubMed: 10036594]
15. Hayward MD, Crimmins EM, Miles TP, Yang Y. The significance of socioeconomic status in explaining the racial gap in chronic health conditions. American Sociological Review. 2000;65:910–929.
16. Hill ME, Preston SH, Rosenwaike I. Age reporting among white Americans aged 85+: Results of a record linkage study. Demography. 2000;37:175–186. [PubMed: 10836175]
17. Hoyert DL, Kung H. Asian or Pacific Islander mortality, selected states, 1992. Monthly Vital Statistics Reports. 1997;46(1) [PubMed: 9404389]
18. Hoyert DL, Arias E, Smith BL, Murphy SL, Kochanek KD. Deaths: Final data for 1999. National Vital Statistics Reports. 2001;49(8) [PubMed: 11591077]
19. Hummer RA, Biegler M, DeTurk PB, Forbes D, Frisbie WP, Hong Y, Pullum SG. Race/ethnicity, nativity, and infant mortality in the United States. Social Forces. 1999a;77:1083–1118.
20. Hummer RA, Rogers RG, Nam CB, LeClere FB. Race/ethnicity, nativity, and U.S. adult mortality. Social Science Quarterly. 1999b;80:136–153.
21. Hummer RA, Rogers RG, Amir SH, Forbes D, Frisbie WP. Adult mortality differentials among Hispanic subgroups and non-Hispanic whites. Social Science Quarterly. 2000;81:459–476. [PubMed: 17879490]
22. Idler E, Benyamini Y. Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior. 1997;39:21–37. [PubMed: 9097506]
23. Johnson NE. The racial crossover in comorbidity, disability, and mortality. Demography. 2000;37:267–283. [PubMed: 10953803]
24. Kaufman J, Cooper R, McGee D. Socioeconomic status and health in blacks and whites: The problem of residual confounding and the resiliency of race. Epidemiology. 1997;8:621–628. [PubMed: 9345660]
25. Kestenbaum B. Mortality by nativity. Demography. 1986;23:87–90. [PubMed: 3956809]
26. Kestenbaum B. A description of the extreme aged population based on improved Medicare enrollment data. Demography. 1992;29:565–580. [PubMed: 1483542]
27. Kestenbaum B. Recent mortality of the oldest old, from Medicare data; Paper presented at the annual meeting of the Population Association of America; Washington, DC. 1997.
28. Kington RS, Nickens HW. Racial and ethnic differences in health: Recent trends, current patterns, 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. Commission on Behavioral and Social Sciences and Education, National Research Council.
29. Kitagawa EM, Hauser PM. Differential mortality in the United States: A study in socioeconomic epidemiology. Cambridge, MA: Harvard University Press; 1973.
30. Lauderdale DS, Kestenbaum B. Mortality rates of elderly Asian American populations. Demography. 2002;39(3):529–540. [PubMed: 12205756]
31. Liao Y, Cooper RS, Cao G, Durazo-Arvizu R, Kaufman JS, Luke A, McGee DL. Mortality patterns among adult Hispanics: Findings from the NHIS, 1986-1990. American Journal of Public Health. 1998;88:227–232. [PMC free article: PMC1508177] [PubMed: 9491012]
32. Manton KG, Gu S. Changes in the prevalence of chronic disability in the United States black and non-black population above age 65 from 1982 to 1999. Proceedings of the National Academy of Sciences. 2001;98(11):6354–6359. [PMC free article: PMC33472] [PubMed: 11344275]
33. 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. 43–104. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council. [PubMed: 23115810]
34. Manton KG, Vaupel JW. Survival after the age of 80 in the United States, Sweden, France, England, and Japan. New England Journal of Medicine. 1995;333:1232–1235. [PubMed: 7565998]
35. Manton KG, Poss SS, Wing S. The black/white mortality crossover: Investigation from the perspective of the components of aging. The Gerontologist. 1979;19:291–300. [PubMed: 551024]
36. Markides KS, Rudkin L, Angel RJ, Espino DV. Health status of Hispanic elderly. In: Martin LG, Soldo BJ, editors. Racial and ethnic differences in the health of older Americans. Washington, DC: National Academy Press; 1997. pp. 285–300. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council. [PubMed: 23115810]
37. Martin LG, Soldo BJ. Introduction. In: Martin LG, Soldo BJ, editors. Racial and ethnic differences in the health of older Americans. Washington, DC: National Academy Press; 1997. pp. 1–9. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council. [PubMed: 23115810]
38. Mathews TJ, MacDorman MF, Menacker F. Infant mortality statistics from the 1999 period linked birth/infant death data set. National Vital Statistics Reports. 2002;50(4) [PubMed: 11837053]
39. McGee DL, Liao Y, Cao G, Cooper R. Self-reported health status and mortality in a multiethnic U.S. cohort. American Journal of Epidemiology. 1999;149:41–46. [PubMed: 9883792]
40. Mirowsky J. Analyzing the associations between mental health and social circumstances. In: Aneshensel CS, Phelan JC, editors. Handbook of the sociology of mental health. New York: Kluwer Academic/Plenum; 1999. pp. 105–123.
41. Mulatu MS, Schooler C. Causal connections between SES and health: Reciprocal effects and mediating mechanisms. Journal of Health and Social Behavior. 2002;43(1):22–41. [PubMed: 11949195]
42. Nam CB. Another look at mortality crossovers. Social Biology. 1995;42:133–142. [PubMed: 7481916]
43. Nam CB, Weatherby NL, Ockay KA. Causes of death which contribute to the mortality crossover effect. Social Biology. 1978;25:306–314. [PubMed: 757307]
44. National Center for Health Statistics. National Health Interview Survey: Multiple cause of death public use data file, 1986-1994 survey years. Hyattsville, MD: Public Health Service; 2000. [Computer file and documentation.]
45. Pappas G, Queen S, Hadden W, Fisher G. The increasing disparity in mortality between socioeconomic groups in the United States, 1960 and 1986. New England Journal of Medicine. 1993;329:103–109. [PubMed: 8510686]
46. Parnell AM, Owens CR. Evaluation of U.S. mortality patterns at old ages using the Medicare enrollment database. 1999. [Accessed July 2, 2003]. Available: http://www​.demographic-research​.org/Volumes/Vol1/2. [PubMed: 12178150]
47. Preston SH, Elo IT. Are educational differentials in adult mortality increasing in the United States? Journal of Aging and Health. 1995;7:476–496. [PubMed: 10165966]
48. Preston SH, Taubman P. Socioeconomic differences in adult mortality and health status. In: Martin LG, Preston SH, editors. Demography of aging. Washington, DC: National Academy Press; 1994. pp. 279–318. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council.
49. Preston SH, Elo IT, Rosenwaike I, Hill ME. African-American mortality at older ages: Results of a matching study. Demography. 1996;33:193–209. [PubMed: 8827165]
50. Rogers RG. Living and dying in the USA: Sociodemographic determinants of death among blacks and whites. Demography. 1992;29:287–303. [PubMed: 1607053]
51. Rogers RG, Hummer RA, Nam CB, Peters K. Demographic, socioeconomic, and behavioral factors affecting ethnic mortality by cause. Social Forces. 1996;74:1419–1438.
52. Rogers RG, Hummer RA, Nam CB. Living and dying in the USA: Behavioral, health, and social differentials of adult mortality. San Diego, CA: Academic Press; 2000.
53. Rogers RG, Hummer RA, Krueger PM. Adult mortality. In: Poston DL Jr, Micklin M, editors. Handbook of population. New York: Kluwer Academic/Plenum; 2004.
54. 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. Vital and Health Statistics. 1999;2(128) [PubMed: 10611854]
55. Rosenwaike I. Mortality of Hispanic populations. New York: Greenwood Press; 1991.
56. Rosenwaike I, Hill ME. The accuracy of age reporting among elderly African Americans: Evidence of a birth registration effect. Research on Aging. 1996;18:310–324.
57. Ross CE, Mirowsky J. Refining the association between education and health: The effects of quantity, credential, and selectivity. Demography. 1999;36:445–460. [PubMed: 10604074]
58. Shetterly SM, Baxter J, Mason LD, Hamman RF. Self-rated health among Hispanic versus non-Hispanic white adults: The San Luis Valley Health and Aging Study. American Journal of Public Health. 1996;86:1798–1801. [PMC free article: PMC1380737] [PubMed: 9003141]
59. Singh GK, Siahpush M. All-cause and cause-specific mortality of immigrants and natives born in the United States. American Journal of Public Health. 2001;91:392–399. [PMC free article: PMC1446566] [PubMed: 11236403]
60. Smith JP, Kington RS. Demographic and economic correlates of health in old age. Demography. 1997a;34:159–170. [PubMed: 9074837]
61. Smith JP, Kington RS. Race, socioeconomic status, and health in late life. In: Martin LG, Soldo BJ, editors. Racial and ethnic differences in the health of older Americans. Washington, DC: National Academy Press; 1997b. pp. 105–162. Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council. [PubMed: 23115810]
62. Sorlie PD, Backlund E, Johnson NJ, Rogot E. Mortality by Hispanic status in the United States. Journal of the American Medical Association. 1993;270:2464–2468. [PubMed: 8031341]
63. Sorlie PD, Backlund E, Keller J. U.S. mortality by economic, demographic, and social characteristics: The National Longitudinal Mortality Study. American Journal of Public Health. 1995;85:949–956. [PMC free article: PMC1615544] [PubMed: 7604919]
64. Stehr-Green P, Bettles J, Robertson LD. Effect of racial/ethnic misclassification of American Indians and Alaskan Natives on Washington state death certificates, 1989-1997. American Journal of Public Health. 2002;92:443–444. [PMC free article: PMC1447096] [PubMed: 11867327]
65. Toussaint DW, Hummer RA. Differential mortality risks from violent causes for foreign- and native-born residents of the United States. Population Research and Policy Review. 1999;18:607–620.
66. U.S. Department of Health and Human Services. Healthy people 2000: National health promotion and disease prevention objectives. Washington, DC: Public Health Service, Author; 1991.
67. U.S. Department of Health and Human Services. Tracking healthy people 2010. Washington, DC: Author; 2000.
68. Williams DR. Racial variations in adult health status: Patterns, paradoxes, and prospects. In: Smelser NJ, Wilson WJ, Mitchell F, editors. America becoming: Racial trends and their consequences. II. Washington, DC: National Academy Press; 2001. pp. 371–410. Commission on Behavioral and Social Sciences and Education, National Research Council.
69. Williams DR, Collins C. U.S. socioeconomic and racial differences in health: Patterns and explanations. Annual Review of Sociology. 1995;21:349–386.