|
|
| |||||||||||||||||||||||||||||||
Background and Methods Persons of low socioeconomic status are known to have reduced life expectancy. In a study of the relation of socioeconomic status to disability-free or active life expectancy among older persons, we analyzed prospectively gathered data on 2219 blacks and 1838 whites who were 65 years of age or older in the Piedmont region of North Carolina. We defined disability as the inability to perform independently one or more basic functional activities such as walking, bathing, dressing, eating, and using the toilet. For subgroups defined by sex, race, and education, statistical models were used to estimate, for persons at each year of age, the probability of transition from not being disabled or being disabled at base line to not being disabled, being disabled, or having died one year later. These transition probabilities were then entered into increment-decrement life tables to generate estimates of total, active, and disabled life expectancy (with total life expectancy equal to active life expectancy plus disabled life expectancy).
Results Sixty-five-year-old black men had a lower total life expectancy (11.4 years) and active life expectancy (10 years) than white men (total life expectancy, 12.6 years; active life expectancy, 11.2 years), although the differences were reduced after we controlled for education. The estimates for 65-year-old black women (total life expectancy, 18.7 years; active life expectancy, 15.9 years) were similar to those for white women. Black men and women 75 years old and older had higher values for total life expectancy and active life expectancy than whites, and the differences were larger after stratification for education. Education had a substantially stronger relation to total life expectancy and active life expectancy than did race. At the age of 65, those with 12 or more years of education had an active life expectancy that was 2.4 to 3.9 years longer than the values for those with less education in all the four subgroups defined by sex and race. Overall, the subgroups with longer total life expectancy and active life expectancy also lived more years with a disability.
Conclusions Among older blacks and whites, the level of education, a measure of socioeconomic status, has a greater effect than race on total life expectancy and active life expectancy.
There has been limited research on the effect of socioeconomic status on disability-free or active life expectancy12. Active life expectancy is computed with life-table methods to estimate the number of remaining years of life expected to be free from disability at specific ages. Active life expectancy is useful as a measure of a population's health status, because the traditional data on specific diseases and mortality do not fully reflect functional status or the quality of life, factors that are important in an older population13. Furthermore, an understanding of the number of years a population group is expected to live in the nondisabled and disabled states permits valuable estimates to be made of the burden this group will place on the health care system, particularly for their longterm care needs. The impact of functional decline in the population is highlighted in the goals of the Public Health Service for the year 2000; an increase in the span of healthy life for U.S. residents is one of its three main goals for the future14.
An important issue of public concern is the lower life expectancy at birth for blacks as compared with whites in the United States15. At older ages, this difference in life expectancy shrinks, and there is a crossover in life expectancy at around the age of 8015 that may be related to the selective survival of healthier blacks16. National and local surveys reveal a slightly higher prevalence of disability among older blacks than among older whites, but these studies do not include estimates of active life expectancy according to race17,18.
We used data from a community-based population 65 years of age and older in north central North Carolina to examine life expectancy and active life expectancy among blacks and whites of both sexes with lower and higher levels of education.
Methods
Study Population
For these analyses, we used the base-line and first annual follow-up interviews of the Piedmont Health Survey of the Elderly (PHSE). This survey is one part of the Established Populations for Epidemiologic Studies of the Elderly, longitudinal studies of persons 65 years old and older funded by the National Institute on Aging. The base-line assessment took place from January 1986 through June 1987. The details of the study design and sampling strategy have been reported elsewhere18,19. Briefly, a four-stage sampling design was used to construct a probability sample of persons 65 years old and older in Durham, Franklin, Granville, Vance, and Warren counties, North Carolina. The sample was designed to contain at least 50 percent black persons. Of the 5226 eligible subjects, 4163 (80 percent) completed the base-line interview. Of these, no data were available on disability at base line or at follow-up for 63 (1.5 percent), 13 (0.3 percent) were lost to follow-up, and 30 (0.7 percent) were neither black nor white and were therefore excluded from these analyses. The final size of the sample was thus 4057.
Collection of Data
Base-line interviews were conducted in participants' homes, and annual follow-up interviews were conducted by telephone. If a telephone interview was not possible, an in-person interview was conducted. If a subject was unable to participate in an interview because of cognitive or physical impairment, selected information, including all data necessary for this analysis, was collected from a proxy respondent. Proxies were used for 3.9 percent of the respondents at base line and for 8.3 percent in the telephone follow-up. The subjects were classified at base line and at follow-up as having a disability in the activities of daily living if they needed help from another person in performing or were unable to perform one or more of the following activities: walking across a small room, bathing, dressing, eating, transferring from bed to chair, or using the toilet20,21.
Participants were asked to specify the highest grade or year of regular school they had completed. Educational status was classified as lower (less than 12 years) or higher (12 years or more). Death was ascertained at the time of the annual follow-up interview and confirmed by death certificate.
Statistical Analysis
The data were used to compute estimates of total life expectancy, active life expectancy, and disabled life expectancy, with use of increment-decrement or multistate life tables. Total life expectancy is the average number of years of remaining life expected for a person of a given age in a specific population group; active life expectancy is the average number of years of life free from disability in the activities of daily living at specified ages; and disabled life expectancy is defined residually as the total life expectancy minus the active life expectancy22.
Increment-decrement life tables were originally developed by demographers to model such population-mobility phenomena as migration among regions and movement among marital-status categories (single, married, divorced, and so on) during a lifetime23,24,25. For these phenomena, complete census data, vital-statistics data, or both usually provide stable age-specific estimates of movement rates or the probability of transition from one category to another. When only sample data are available, as in the estimation of working-status life tables, the observed age-specific estimates of rates or transition probabilities are less stable, so that considerable graduation (or smoothing) by means of moving averages or related techniques is required for the application of population-based life-table procedures26,27,28. Because of the limited number of participants in certain subgroups defined by age, sex, race, and education and the associated limitations on the number of person-years of exposure, the number of transitions in disability status, and the number of deaths, it was necessary in the present analysis to model the age-specific probability of transition from one state to another28.
We first classified all participants as nondisabled or disabled in activities of daily living at base line and as nondisabled, disabled, or dead one year after base line. The Rate computer program was then applied to the cross-classified individual-level transition data to estimate instantaneous transition rates for all possible transitions (to and from the disability states and from each to death) and to obtain iterative maximum-likelihood estimates of regression coefficients for the effects on the transition rates of age (for each year from 65 through 94 and for 95 or older) and the covariates sex, race, and education29. This regression method is based on the assumption of an isomorphism between the structure of the Markov process underlying a conventional specification of the increment-decrement life-table model and that of the conventional Markov panel-regression model. The maximum-likelihood estimates of the regression parameters were then used to compute smoothed transition probabilities for each combination of a single year of age and the covariates sex, race, and education.
The resulting graduated transition probabilities vary more smoothly across ages than the sample estimates and were used as input data for a piecewise constant-transition-rates increment-decrement life-table program (written specifically for this project, following conventional methods)25. The program proceeds by single years from age 65 to age 94 and is terminated by an open-ended interval for 95 and older. The sizes of the active and disabled life-table populations at the initial age, 65 years, were set proportional to logistically modeled values of the sizes that prevailed in the sample at base line. (Details of this synthesis of statistical estimates of transition-rate functions and increment-decrement life-table techniques are available from the National Auxiliary Publications Service.*)
In comparing total life expectancy and active life expectancy among subgroups defined by sex, race, and education, it would be helpful to know the precision of the estimates. Unfortunately, estimating the standard errors of life expectancies in increment-decrement life tables is a difficult problem for which no general solutions are available. Nonetheless, approximate estimates at 65, 75, and 85 years of age were calculated by first regressing the logarithm of the active life expectancy for age-sex-race-education subgroups on the covariates sex, race, and education. With the resulting estimated regression-coefficient vectors used in first-order (linear) approximations of a Taylor series expansion of the dependence of active life expectancy on the covariates, the conventional delta method30 was then applied to produce approximate standard errors of 1.0 years or less for all three ages. Standard errors of similar magnitude were obtained for the estimates of total life expectancy. Accordingly, differences among the subgroups of more than about two years in active life expectancy and total life expectancy may be regarded as statistically significant.
Results
The sample for these analyses consisted of 1427 men (35.2 percent) and 2630 women (64.8 percent); there were 2219 blacks (54.7 percent) and 1838 whites (45.3 percent); 3087 (76.1 percent) had less than 12 years of education, 905 (22.3 percent) had 12 or more years of education, and for 65 (1.6 percent) the number of years of education was unknown. Among those with known educational status, 12 or more years of education were reported by 112 black men (14.6 percent of black men), 189 black women (13.4 percent), 203 white men (31.9 percent), and 401 white women (34.0 percent).
Figure 1 shows unmodeled transitions for men and women in three age groups. For those with no disability in activities of daily living at base line, the large majority remained nondisabled, with the proportion decreasing with increasing age. The rate of transition to disability increased with age and was similar for men and women. The majority of those who were disabled at base line remained disabled. However, a substantial proportion of those who were disabled at base line reported no disability in activities of daily living one year later; this proportion generally declined with increasing age. Mortality was generally higher among men than among women, and those who were disabled had a mortality rate that was two to three times that of members of the same sex and age group who had no disability at base line. During one year of follow-up, 55 black men (7.0 percent) died, as did 40 white men (6.2 percent), 53 black women (3.7 percent), and 43 white women (3.6 percent).
|
|
In general, there were no large differences between blacks and whites in total life expectancy and active life expectancy when education was taken into account (Table 2). After stratification according to education, the differences between black and white men were one year or less at all ages. The differences according to race were somewhat greater among women after stratification according to education, with the greatest difference among women in the higher-education group who were 85 years of age or older; in this group blacks had a 3.6-year advantage in total life expectancy and a 3.2-year advantage in active life expectancy.
|
It is of particular interest that for each sex-race-education subgroup, the absolute number of years of remaining life during which the subject was expected to be disabled (disabled life expectancy) differed little among 65-year-olds, 75-year-olds, and 85-year-olds (Table 1). Because at older ages the total life expectancy was shorter, however, the percentage of remaining life during which the subject was expected to be disabled (calculated by dividing disabled life expectancy by total life expectancy) went up with age (Table 3). In Table 1, when higher-education subgroups were compared with lower-education subgroups, those who lived longer and had a longer active life expectancy also had a slightly longer disabled life expectancy. When women and men in education-race subgroups were compared, women had a disabled life expectancy about twice that of men in all cases. However, both the education-related and the sex-related differences in disabled life expectancy were absent when disabled life expectancy was expressed as a proportion of the remaining years of life (Table 3), because the subgroups with a longer disabled life expectancy also had a longer total life expectancy.
|
The increased risk of death associated with lower socioeconomic status is well documented,1 and recent research has also demonstrated an increased risk of disability among older persons with lower education and income levels31,32,33,34,35,36. It is not surprising, therefore, that the active life expectancy, which reflects the influence of both disability and mortality, should be strongly affected by educational status. Our findings, however, bring the magnitude of the risk associated with lower socioeconomic status into focus in terms of actual years of life. In this cohort, 65-year-old men and women with 12 or more years of education could expect 2.4 to 3.9 more years of life without disability than subjects of the same age and sex with less education.
The cohort studied was selected to be representative of the Piedmont region of North Carolina. Although not necessarily representative of the U.S. population, this cohort is drawn from both urban and rural areas and has a wide distribution of income and education levels among both blacks and whites,18 thus permitting the examination of differences in rates of disability and death among socioeconomic groups. Although our actual estimates of total life expectancy and active life expectancy cannot be extrapolated directly to the U.S. population, our central finding -- which it would be useful to evaluate in a national sample -- is the magnitude of the differences in active life expectancy associated with differences in educational status.
Our estimates of total life expectancy and active life expectancy are reasonably comparable to other available data. As compared with national estimates,15 total life expectancy at the age of 65 in our study was about two years shorter for black and white men, was nearly two years greater for black women, and was essentially the same for white women. Education-related differences in life expectancy have been demonstrated in large samples of the U.S. population in 19602 and 1979 through 198537. The 1979-1985 study showed clear gradations in life expectancy according to education among both men and women. At 65 years of age, there was a 3.3-year difference among white men and a 2.4-year difference among white women between those with 4 or fewer years of education and those with 17 or more years37. Differences between the results of these studies and ours may reflect real differences between the older population of the Piedmont region of North Carolina and that of the United States, but they may also be related to the relatively small sample size for each of the subgroups in our study.
Because previous estimates of active life expectancy have used different definitions of disability and different methods,12,22,38,39 it is difficult to make direct comparisons. Active life expectancy was calculated with increment-decrement life-table techniques for whites in the original three communities in the Established Populations for Epidemiologic Studies of the Elderly (East Boston, Mass.; Iowa and Washington counties, Iowa; and New Haven, Conn.), although modeling was not carried out to smooth the transition probabilities22. Estimates of active life expectancy in our study fall within the range of values for active life expectancy at the other three sites for both men and women at 65 and 75 years of age and are close to the estimates for men and women at 85 years. In a study of a representative sample of older persons in Massachusetts, the 65-year-old men and women who were poor were found to have a 2.4-year deficit in active life expectancy as compared with nonpoor subjects at age 6512.
In our study the differences in total life expectancy and active life expectancy according to race were not large. As compared with whites, only 65-year-old black men had shorter total life expectancy and active life expectancy, and the differences were small after stratification according to education. Nevertheless, the lower education level among black men (only 14.6 percent had 12 or more years of education, as compared with 31.9 percent of white men) means that, overall, black men are at a substantial disadvantage at age 65 as compared with white men. These findings are consistent with those of studies in populations with wider age ranges, which found that, after adjustment for socioeconomic status, the disadvantage for blacks in terms of mortality was not significant40,41.
Once black men and women in this cohort reached the age of 75, they had an advantage over whites in both total life expectancy and active life expectancy. Almost all the advantage in total life expectancy came from added years of active life expectancy, with only slightly more disabled life expectancy than is estimated for whites. These "crossovers," which favor blacks after 75 years of age, have been estimated from prospective cohort data, so they are not easily dismissed as attributable to inaccuracies in the data. Crossovers have been estimated not only for the previously studied total life expectancy, but also for active life expectancy.
Compatible with these findings are data on differences in health and functional status among older blacks and whites that have been extracted from a variety of local and national studies by Gibson42. In several studies of blacks of different ages, the "young old" (65 to 74 years old) were actually more likely to be in poor health than older persons. This pattern may be due to selective mortality among sicker persons, who never reach the oldest ages, and may also explain the crossover in active life expectancy seen in this study. In studies comparing the prevalence of disease and disability between blacks and whites, there was a slight disadvantage for blacks in the "young old" age group, a lesser disadvantage for the group 85 years of age and older, and little or no difference between the races among those 75 to 84 years old42. The somewhat higher prevalence of disability among older blacks may actually result from longer survival with disability and is not incompatible with a longer active life expectancy, which is also a function of survival time. In fact, a similar pattern emerges when women are compared with men and blacks with whites: higher prevalence of disability, longer disabled life expectancy, yet also longer total life expectancy and active life expectancy.
This study may offer some insight into the ongoing debate about the future prospects for a compression of morbidity,43,44,45,46 which may be defined as a reduction in disabled life expectancy and an increase in active life expectancy. Comparing groups with large differences in total life expectancy (men vs. women, those with lower education levels vs. those with higher education levels), we found that longer life entails more active years but, in absolute terms, more years of life with disability as well. In subgroups defined by education and race, women consistently had twice as many years of life with disability as men at all ages. For example, at age 65, black men with a lower education level had a disabled life expectancy of 1.4 years, as compared with 2.7 years for black women with a lower education level. In sex-race subgroups, those with a higher education level consistently had more disabled years than persons with a lower education level. For example, at age 85, black men with a lower education level had a disabled life expectancy of 1.6 years, as compared with 2.0 years for black men with a higher education level., and although the proportion of remaining life expected to be lived in the disabled state did not differ substantially between longer- and shorter-lived subgroups (Table 3), the impact on the public health must be considered in terms of the absolute number of years of disability. In the future, if those who currently have shorter total life expectancy (men and persons with less education) become more like today's groups with longer total life expectancy, then they can be expected to have a longer active life., and although about the same proportion of life will be lived with disability, there will be more years during which they are disabled, with the attendant burdens on the medical care and long-term care systems.
Perhaps the most noteworthy subgroup in this study is black women with higher education levels. This group had the highest total life expectancy and active life expectancy at each age and had a widening advantage over similarly educated white women with increasing age. Unfortunately, with long total life expectancy and active life expectancy also comes the longest disabled life expectancy of any subgroup. If these results can be confirmed in other studies, research focusing on these women would be valuable in elucidating the factors that promote their long active life expectancy. Understanding diseases and other factors that have the most important roles in adding years of disabled life expectancy in this long-lived group of women could be useful in devising preventive strategies to reduce disabled life expectancy as total life expectancy continues to increase in the future.
Our finding that educational attainment has a strong influence on total life expectancy and active life expectancy among both blacks and whites is of great importance, because education level, and socioeconomic status in general, are alterable risk factors. At least part of the disadvantage associated with low socioeconomic status relates to poorer lifelong health practices in this group,9 and efforts must continue to improve these practices. At the population level, however, raising the general level of socioeconomic status may have even more profound effects on health-related behavior and health outcomes. Increasing the likelihood that a person will attain a high level of education not only may be advantageous for that person's young and middle years as a wage earner but also may be a valuable investment in increasing his or her years of active, nondisabled life after retirement.
Supported under a contract (N01-AG-1-2102) with the National Institute on Aging in support of the Established Populations for Epidemiologic Studies of the Elderly (Duke University).
Source Information
From the Epidemiology, Demography, and Biometry Program, National Institute on Aging, Bethesda, Md. (J.M.G.); the Department of Sociology, Duke University (K.C.L.), and the Department of Psychiatry (D.B.) and the Center for the Study of Aging and Human Development (G.G.F.), Duke University Medical Center, Raleigh, N.C.; Boston University School of Medicine, Boston (L.G.B.); and Abt Associates, Cambridge, Mass. (L.G.B.).
Address reprint requests to Dr. Guralnik at the National Institute on Aging, 7201 Wisconsin Ave., Rm. 3C-309, Bethesda, MD 20892.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
