| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Rush Institute for Healthy Aging
2 Department of Internal Medicine
3 Department of Preventive Medicine
4 Rush Alzheimer's Disease Center
5 Department of Neurological Sciences
6 Department of Psychology, Rush University Medical Center, Chicago, Illinois.
Address correspondence to Dr. Carlos F. Mendes de Leon, Rush Institute for Healthy Aging, Rush University Medical Center, 1645 W. Jackson Blvd., Suite 675, Chicago, IL 60612. Email: cmendes{at}rush.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. Our data came from a longitudinal, population-based study of 6,158 Black and White adults aged 65 and older from the south side of Chicago. Data were collected during three face-to-face interviews between 1993 and 2002. We ascertained disability using three self-report measures and a performance-based measure of physical function. Using longitudinal data analysis, we examined overall racial differences in disability and the degree to which they varied by age and gender.
Results. After an average of 6 years of follow-up, Blacks reported significantly higher disability levels than Whites after adjustment for age and sex. There was a significant increase in disability during follow-up on all four measures. Racial differences in disability did not vary consistently by age at baseline or over time but were greater among women. Findings were largely similar for self-reported and performance-based measures of disability. Adjustment for socioeconomic status substantially reduced racial differences in disability, although some of the differences remained significant, especially among women.
Conclusions. Racial disparities in disability have not been eliminated, are greater among women, and have their origins earlier in adulthood.
ELIMINATION of racial disparities in health remains an important public health challenge in the United States and a top priority of the Healthy People 2010 program of the Department of Health and Human Services (U.S. Department of Health and Human Services, 2003
). Progress toward this goal requires continuous monitoring of health indices at the population level and comparing these indices across subpopulations defined on the basis of race or socioeconomic status (SES). The purpose of this research is to provide recent information on differences in health between older non-Hispanic Blacks and Whites, focusing on disability as a sensitive and commonly used indicator of the total burden of morbidity in the oldest segment of the population (Manton, Corder, & Stallard, 1997; Pope & Tarlov, 1991).
Previous studies have generally pointed toward a substantially higher prevalence of disability among older Blacks relative to non-Hispanic Whites (Clark, 1997; Clark & Maddox, 1992; Kelley-Moore & Ferraro, 2001; Manton & Gu, 2001; Mendes de Leon et al., 1997). These findings reflect disability differences in the 1980s and early 1990s but do not inform on current Black–White differences in disability. Recent trends in population health suggest that overall racial disparities may have narrowed during the 1990s. Using national statistics, Margellos, Silva, and Whitman (2004) found evidence for a decrease in disparities in 11 of the 14 health indicators they analyzed. Of most relevance to health in late life, there were important reductions in disparities in all-cause and stroke mortality, but a slight increase in the disparity in heart disease mortality.
There is little information on secular changes in disparities in age-related disability. Recent data suggest a slowly declining prevalence of age-specific disability levels (Freedman, Martin, & Schoeni, 2002). For example, Manton and Gu (2001) have estimated that chronic disability among persons aged 65 years and older declined by 0.26% per year between 1982 and 1989, by 0.38% per year between 1989 and 1994, and by 0.56% per year between 1994 and 1999. It remains unclear, however, whether the decline in overall prevalence of disability has led to a concomitant reduction in racial disparities in disability. Although some recent findings appear consistent with a decline in racial disparities in disability (Manton & Gu), others have failed to find support for such a trend (Schoeni, Freedman, & Wallace, 2001).
There are two additional unresolved issues that have led to an incomplete understanding of racial differences in disability. The first issue is the influence of the aging process itself. The prevailing assumption is that Black persons experience poorer health relative to Whites once they reach the later stages of life due to the cumulative effects of a lifelong disadvantage in social and economic conditions (Ferraro & Farmer, 1996). However, a recent report has argued persuasively that a lifecourse perspective based on cumulative disadvantage may be confounded by a tendency in the opposite direction, that of selective survival (Kelley-Moore & Ferraro, 2004). In essence, cumulative disadvantage may lead to greater premature mortality, resulting in survivors of a comparatively healthier group of older adults, offsetting the effects of disadvantage experienced earlier in life. Mortality data suggest a convergence or crossover in health between Blacks and Whites at later stages of life (Corti et al., 1999; Mendes de Leon et al., 1997; Ng-Mak, Dohrenwend, Abraido-Lanza, & Turner, 1999). There has also been some evidence for a decrease in racial differences in disability at older ages (Johnson, 2000; Mendes de Leon et al.), although others have failed to find such a pattern (Clark, 1997; Liao, McGee, Cao, & Cooper, 1999). In their own longitudinal analyses of the North Carolina Established Populations for the Epidemiologic Studies of the Elderly (EPESE) data, Kelley-Moore and Ferraro (2004) note that the Black disadvantage in disability increased over time, that is, with aging. Further analysis revealed that this aging effect was reduced to nonsignificance after adjustments were made for socioeconomic and health-related variables. Although this was one of the few reports that has specifically focused on the age dependency of racial differences in disability, the data for this study are now more than a decade old, and more recent data on this topic are lacking.
The second unresolved issue is whether racial disparities in disability disproportionately affect Black women. In general, women report a higher prevalence of disability than do men (Leveille, Penninx, Melzer, Izmirlian, & Guralnik, 2000; Oman, Reed, & Ferrara, 1999). Although there have been few systematic investigations of racial disparities by gender, there are reasons to believe that these disparities are greater among women. A number of studies have shown that the Black disadvantage in medical conditions and other risk factors for disability tends to be greater among women. For example, Black women have a higher prevalence of cardiovascular disease (Geronimus, Bound, Waidmann, Hillemeier, & Burns, 1996; Gillum, 1994; Gillum, 1996; Ng-Mak et al., 1999), diabetes (Robbins, Vaccarino, Zhang, & Kasl, 2000), and obesity (Flegal, Carroll, Ogden, & Johnson, 2002; Freedman, Khan, Serdula, Galuska, & Dietz, 2002) compared with White women, and these differences are larger than those observed among men. There is initial evidence that Black women also fare more poorly in disability-related outcomes. Using cross-sectional data from the third National Health and Nutrition Examination Survey (NHANES III), Ostchega, Harris, Hirsch, Parsons, and Kington (2000) found that Black women were disabled in a greater number of specific functional tasks (e.g., rising from a chair, walking a quarter mile) relative to White women than Black men were relative to White men. Like those of most other studies, these findings date from the late 1980s and early 1990s and do not reflect the influence of the recent decline in the overall prevalence of disability. In addition, prevalence data may give an incomplete picture of true differences in the rate of development of disability due to the influence of selective survival and healthy participation effects in cross-sectional studies.
In summary, the purpose of this descriptive analysis was to provide an update on racial differences in disability, using data from a longitudinal study of older adults. We specifically tested whether Blacks have higher levels of disability compared with Whites, whether racial differences in disability change over time in old age, and whether they were greater among women than men. Because racial differences in health develop in the context of substantial socioeconomic disadvantage for Blacks (Schoenbaum & Waidmann, 1997; Williams, 1999), we also evaluated the degree to which racial differences in disability are modified by SES.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Evans et al., 2003). The CHAP study was conducted in a geographically defined community area that comprised three adjacent neighborhoods on the south side of Chicago. Together, these neighborhoods contained an approximately equal number of Black and White residents, with a reasonably wide spectrum of SES among both Blacks and Whites in the study area, although, on average, SES levels were lower among Blacks. However, both Blacks and Whites in this population had somewhat higher SES levels compared with national averages. The study began with a complete census of the community area conducted between 1993 and 1996. All residents aged 65 years and older were identified and asked to participate. Of the 7,813 eligible residents, 6,158 (78.9%) enrolled and participated in an in-home interview at baseline. Of the 6,158 participants, 3,781 (61.4%) identified themselves as non-Hispanic Black or African American, and 2,321 (37.7%) as non-Hispanic White. The remaining 56 (0.9%) were excluded from the present analysis. The baseline interview began in 1993 and was completed in 1997. This phase was followed by two successive follow-up interviews at approximately 3-year intervals in the homes of the participants, conducted from 1998 to 2000, and 2000 to 2002, respectively. The interviews included structured questions about sociodemographic characteristics, health, and lifestyle, as well as performance-based tests of physical and cognitive function. The study was approved by the institutional review board of Rush University Medical Center, and all participants (or legal guardians) provided written consent.
Study Variables
Disability Outcomes
All three interviews included identical sets of questions on disability status as well as performance tests of physical function. Self-reported information on disability status relied on three widely used measures of basic daily functions. The first measure focuses on activities of daily living (ADLs) and assesses the ability to perform six basic self-care tasks without help (e.g., bathing, dressing, eating; Branch, Katz, Kniepmann, & Papsidero, 1984). The second measure focuses on the ability to perform tasks requiring mobility and strength and includes three items—walking up and down stairs, walking half a mile, and doing heavy work around the house (Rosow & Breslau, 1966). The third measure assesses the ability to perform four basic upper- and lower-extremity functions—pulling or pushing large objects, stooping, crouching, or kneeling, reaching or extending arms above shoulder level, and writing or handling small objects (Nagi, 1976). For ease of presentation, we have referred to the last 2 measures as Rosow-Breslau and Nagi disability, respectively. We formed a summary score for each measure by adding the number of tasks the subject reported being able to do without help, with higher scores indicating less disability. We applied reverse coding to graphically present disability changes over time as a process of decline and to facilitate comparisons with results of the performance-based measure of physical function, which is normally expressed in the same direction (see below).
We assessed physical function by three performance tests that focus on lower-extremity strength, balance, and gait—tandem stand, measured walk, and repeated chair stands. These are commonly used tests of physical function in elderly populations with reasonable reliability (Jette, Jette, Ng, Plotkin, & Bach, 1999; Tager, Swanson, & Satariano, 1998) and well-established predictive validity (Guralnik et al., 2000; Guralnik, Seeman, Tinetti, Nevitt, & Berkman, 1994). The tandem stand test measures the duration that a full tandem stand can be maintained (up to 10 seconds). The measured walk tests the time to complete an 8 ft. walk, and the chair stand measures the time to get up from a chair 5 times. In keeping with procedures established previously (Guralnik, Simonsick, et al., 1994; Guralnik, Seeman, et al.), we converted recorded times into quintiles of timed performance, with an additional category for those who were unable to complete the test. This resulted in scores ranging from 0 to 5, which were summed across the three tests for an overall summary measure of physical function (range = 0–15). Higher scores indicated a higher level of physical function.
Covariates
Other variables included in the analysis were age (in years), sex, race (ascertained using the 1990 U.S. Census Bureau question and classification), and SES. We constructed a composite measure of lifetime SES based on four components of SES that characterize different stages of the lifecourse, including childhood, adulthood, and older age. The first component was a previously developed measure of childhood SES based on father's education, mother's education (both in years of schooling completed), father's occupational prestige score (see below), and financial status during childhood (rated on a 5-point scale from very poor to very well off; Everson-Rose, Mendes de Leon, Bienias, Wilson, & Evans, 2003). The other three components were the participant's own level of education, occupational status, and current income. We measured education in years of schooling completed and ascertained occupational status at age 30 using questions from the U.S. Census, classified according to the 1990 U.S. Census indices of industries and occupations. Using the method described by Hauser and Warren (1996), we assigned occupational prestige scores based on occupational earnings and educational requirements data derived from the 1990 U.S. Census. We measured total household income using a color-coded card with 10 income categories, ranging from less than $5,000 per year to equal to or greater than $75,000 a year. We considered responses a rank-order variable with values ranging from 1 to 10. We obtained a measure of lifetime SES after z-scoring each of the four component measures and then averaging across the nonmissing values of each component. We set the lifetime SES measure as missing if two or more indicators had missing values. Intercorrelations between the four component indicators (childhood SES, education, occupation, income) ranged from 0.30 to 0.66, suggesting that these indicators were meaningfully related to each other but also that each contributed unique information to a person's lifetime SES.
Analysis
We computed baseline comparisons between Black and White respondents using t tests for continuous variables and chi-square tests for categorical variables. We used the generalized estimating equations (GEE) method to fit the longitudinal disability data as a function of follow-up time since baseline and to account for the correlated structure of disability data across repeated measurement (Zeger & Liang, 1986). Due to the nonnormal distribution of the self-reported disability measures, we considered each disability score as the proportion of tasks a person was able to do without help out of the total number of tasks. We analyzed the proportions using a logistic link function and binomial error structure and the performance-based outcome variable using an identity link and normal error structure.
The initial step of the analysis was to fit a base model for each outcome on the basis of time since baseline, age, sex, and age squared, to allow for a nonlinear relationship between age and disability, as well as the interactions between age and sex with time since baseline, to allow for different rates of change in disability over time as a function of baseline age and sex. We then entered a main effect for race to test for the presence of overall Black–White differences on each disability outcome. The term for race in these models represented differences in absolute disability levels averaged across the three interviews, or the cross-sectional association of race with disability. At the next step, we tested specifically whether racial differences in disability varied as a function of age and sex. To test the age interaction, we entered interaction terms between race and baseline age and time since baseline. The Race x Age term tested the degree to which racial differences in disability varied as a function of age at baseline (a cross-sectional effect). The Race x Time term tested the degree to which racial differences in disability changed over time, or as people aged (a longitudinal effect). We added a race by gender interaction term to test if racial differences in disability were greater among women than men. In the final model, we entered the lifetime SES variable to examine the extent to which SES modified the cross-sectional or longitudinal effects of race on disability. All longitudinal analyses were performed using the GENMOD Procedure of the SAS Version 8 (SAS Institute, 2000).
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
To illustrate the results of the second series of models, we computed predicted disability scores by race and sex and plotted these as a function of baseline age and follow-up time (see Figure 1). The main line in each plot shows the increase in disability (depicted as decline on each disability measure) with age (65 to 95) at baseline. The shorter lines that part from the main lines show the increase in disability during follow-up at ages 65, 75, and 85. The lines indicate that disability increases much more rapidly over time (i.e., with aging) than would be predicted from the cross-sectional associations between age and disability. A similar conclusion can also be derived from inspecting the difference in coefficients between the cross-sectional age and longitudinal time effects in Table 2. In each instance, the longitudinal coefficient was at least twice the magnitude of the cross-sectional coefficient, with the difference captured by the quadratic coefficient for the physical function summary measure. The underestimation of age-related declines in health is a common limitation of cross-sectional data, due to selective survival effects. The figure illustrates that the racial differences remain mostly constant with increasing age and during follow-up, but are greater among women than men.
|
|
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Overall, our findings provide evidence for the continued existence of important racial disparities in disability through at least the second half of the 1990s. This suggests that racial disparities in the total burden of morbidity in older age, as marked by disability levels, have not been eliminated. The results are consistent with other reports based on national data, which corroborate the persistence of racial disparities in disability (Manton & Gu, 2001; Schoeni et al., 2001). It should be noted that our findings do not directly inform on trends in the magnitude of racial disparities in disability since the late 1980s and early 1990s, the period during which the age-specific prevalence of disability has been decreasing (Freedman et al., 2002). Although there appears to be some evidence that the downward trend in overall disability levels has also led to a decrease in racial disparities in disability (Manton & Gu), the exact magnitude of this trend remains unclear (Schoeni et al.) and requires further investigation.
Our data pertain to a particular urban population, which may limit their generalizability to other parts of the country or to rural populations. The study population was reasonably characteristic of urban populations more generally. On average, Blacks had lower SES than Whites, but there was a fairly wide spectrum of SES in both groups. In addition, the ratios of differences in common indicators of SES such as education and income between Blacks and Whites were similar to those at a national level. For example, according to the 1990 U.S. Census, Blacks aged 65 years and older nationwide were 21% less likely than Whites to have a household income of $25,000 or more, compared to 22% less likely in the study community area. Blacks nationwide were also 25% less likely than Whites to have had more than high school education, compared to 30% less likely in the study area (U.S. Bureau of the Census, 2004). It appears, however, that progress in the reduction of racial disparities in health in the city of Chicago has lagged progress seen at the national level (Margellos et al., 2004). This may be due to a number of factors, for example the more extreme unequal distribution of socioeconomic resources between Blacks and Whites in the city of Chicago compared to national averages. The findings require replication from other parts of the country before firm conclusions can be reached regarding recent trends in racial disparities in disability.
Several aspects of our study increase confidence that the observed racial differences in disability reflect actual differences in the total burden of morbidity. Disability differences were evident across a range of functional tasks, including ADLs, mobility-related activities, and basic physical functions. Racial differences were also evident on a performance-based measure of physical function, which provides validation for the racial differences observed in self-reported disability measures. This is important given the limitations of self-report measures, which may be subject to various sources of bias. An important source of bias that has the potential of differentially affecting self-reports of disability by race is response shift. Response shift is the tendency to modify one's self-evaluation of the target health outcome due to a change in calibration of the standards of measurement or to a redefinition of the health construct itself (Daltroy, Larson, Eaton, Phillips, & Liang, 1999; Schwartz & Sprangers, 1999). Older Blacks, especially Black women, have a tendency to appraise their own health more negatively than Whites (Ferraro, 1993), which may induce response shift. It appears unlikely, however, that the observed differences in disability levels were due to a systematic overreporting of disability among Blacks, in view of the higher levels of impairment they exhibited on the performance-based measures of disability. To the best of our knowledge, there have been no previous longitudinal, population-based studies of racial differences in disability based on information from both self-report and performance-based measures of disability.
Our longitudinal data did not reveal clear evidence for an attenuation of racial differences in disability as the cohort was aging. Although such a pattern appeared evident for two of the self-report measures of disability, the performance-based measure suggested a trend in the opposite direction, that is, a widening of disability differences over time. The exact reasons for these discrepant results are unclear. It is possible that the performance-based measure of physical function is a more sensitive indicator of the total burden of morbidity, given that it does not rely on subjective appraisal of functional capacity. Hence, this measure may capture changes in underlying physical health with greater accuracy compared with self-report measures and provide a more valid test of group differences in changes in health over time. Disability by self-report, on the other hand, involves the subjective appraisal of one's functional capacity, which may reflect in part the continuous adjustments older adults make in their perceived levels of functional ability as their health declines. In other words, perceived declines in function may lag actual declines in physical health, which may have contributed to the apparent attenuation of racial differences in self-reported levels of disability over time. It should be noted, however, that the absolute magnitude of the time-dependent effects was small compared with absolute differences in self-reported or performance-based measures of disability, and these effects could therefore represent mostly random variations of relatively stable differences in disability over time. Thus, there was no clear evidence for a convergence or crossover of disability levels at more advanced ages, as has been observed for mortality (Corti et al., 1999; Mendes de Leon et al., 1997; Ng-Mak et al., 1999).
Our findings are consistent with predictions based on the lifecourse perspective of cumulative disadvantage, reflecting the counter-balancing effects of premature mortality and continuing health disadvantage among minorities (Kelley-Moore & Ferraro, 2004). They suggest that racial disparities in disability do not emerge or widen in older age, but are already present at age 65, even if average disability levels at that age tend to be low and racial differences small in magnitude. This may well reflect the fact that health risks and chronic conditions that give rise to the development of disability affect Blacks at an earlier stage in life than they do Whites. Improved identification and prevention of these health risks will be essential to reducing racial disparities in disability in older age.
Racial disparities in disability were significantly greater among women than men, and even at age 65, Black women already display considerable levels of disability. This finding is consistent with observations of the disproportionate burden of disabling health conditions among Black women throughout adulthood (Brownson et al., 2000; Flegal et al., 2002; Freedman, Khan, et al., 2002; Gillum, 1996; LeClere, Rogers, & Peters, 1998; Robbins et al., 2000). Less favorable social conditions due to poverty (U.S. Bureau of the Census, 1996) and greater family and caregiver burdens (McCann et al., 2000) may also contribute to the higher disability levels among Black women. Blacks also tend to have more restricted access to high-quality health care (Schneider, Zaslavsky, & Epstein, 2002; Smedley, Stith, & Nelson, 2003; White-Means, 2000), leading to inadequate treatment and prevention of health conditions that cause disability later in life. Lack of adequate care may adversely affect Black women in particular, given the higher prevalence of potentially disabling health risks and health conditions they experience throughout adulthood.
There is abundant evidence that the root causes of racial disparities in health need to be sought in the substantial disparities in socioeconomic resources that exist among racial groups (Williams, 1999). Our own findings indicate that disparities in disability at older ages largely follow the same pattern, although residual differences remained evident after adjustment for SES, especially for women and on the performance-based measure of physical function. The latter finding may reflect the aforementioned greater sensitivity of the performance-based measure to capture gradations in overall burden of morbidity. The reasons for the profound social disparities in health are complex and involve pathways related to differential adoption of health risk behaviors, greater exposure to environmental health risks, higher levels of chronic stress and unequal access to care (Anderson & Armstead, 1995; Gallo & Matthews, 1999; Lowry, Kann, Collins, & Kolbe, 1996; Macintyre, Maciver, & Sooman, 1993; Williams & Collins, 1995), all of which may lead to the development of chronic disease and disability. Furthermore, specific socioeconomic resources, such as education, may not produce the same social value or socioeconomic return across racial groups, for example, due to differences in the quality of education or labor market discrimination (Cancio, Evans, & Maume, 1996; O'Neill, 1990; Smith & Welch, 1989). Minorities, and Blacks in particular, may also be exposed to social conditions that are less explicitly tied to socioeconomic success yet confer health risks on their own, such as racial discrimination (Kessler, Mickelson, & Williams, 1999; Krieger, Rowley, Herman, Avery, & Phillips, 1994; Ren, Amick, & Williams, 1999), lack of collective efficacy (Sampson, Raudenbush, & Earls, 1997), or neighborhood stress (Ellen, Mijanovich, & Dillman, 2001; Elliott, 2000; Ewart & Suchday, 2002). The extent to which social conditions are at the origin of racial disparities in health raises the specter that individual-level approaches aimed at the treatment or prevention of health risks may ultimately be insufficient to achieve complete elimination of these disparities.
In the background of growing evidence for a declining prevalence in disability in the general population aged 65 years old and older, we conducted an analysis of racial differences in disability using recent data from a population-based study of older adults. The overall pattern of findings suggests three broad conclusions. First, older Blacks still experience substantial disparities in disability across a range of functional tasks and abilities, with greater disparities among women than men. Second, racial disparities in disability do not tend to develop or exacerbate in old age itself, but rather seem to be present at the onset of older age, suggesting that they have their origins in earlier adulthood. Third, factors related to socioeconomic disadvantage play an important role in shaping racial disparities in disability.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication November 11, 2004. Accepted for publication April 26, 2005.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. Liang, J. M. Bennett, B. A. Shaw, A. R. Quinones, W. Ye, X. Xu, and M. B. Ofstedal Gender Differences in Functional Status in Middle and Older Age: Are There Any Age Variations? J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2008; 63(5): S282 - S292. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Cornman and V. A. Freedman Racial and Ethnic Disparities in Mobility Device Use in Late Life J. Gerontol. B. Psychol. Sci. Soc. Sci., January 1, 2008; 63(1): S34 - S41. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. D. Dunlop, J. Song, L. M. Manheim, M. L. Daviglus, and R. W. Chang Racial/Ethnic Differences in the Development of Disability Among Older Adults Am J Public Health, December 1, 2007; 97(12): 2209 - 2215. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. F. Mendes de Leon Aging and the Elapse of Time: A Comment on the Analysis of Change J. Gerontol. B. Psychol. Sci. Soc. Sci., May 1, 2007; 62(3): S198 - S202. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
