|
TRENDSAre The Young Becoming More Disabled?
This paper investigates trends in disability in the U.S. population, particularly among people under age fifty. Even as the elderly have become less disabled, reported disability has risen for younger Americans, especially those ages 30–49. We suggest some possible explanations for rising disability levels, such as obesity, technological advances in medicine, and changing disability insurance laws. Obesity and its attendant disorders seem particularly associated with these trends, although the data are not definitive. Whatever its sources, rising disability among the young could have adverse consequences for public programs such as disability insurance, Medicare, and Medicaid.
During the early 1970s a great deal of concern was raised about apparent increases in disability among the old.1 Several researchers argued that increases in longevity amounted to extensions of the time that the elderly spend in disability. Thus, they predicted, increases in longevity would inevitably be accompanied by growth in the incidence of disability.2 These fears subsided as the 1980s began. Indeed, a consensus is beginning to emerge that the health of the elderly has been improving along some important dimensions since the early 1980s and that these improvements might have accelerated during the 1990s.3 Given the enormous increases in diseases such as obesity and diabetes among the young, it seems important to ask whether disability trends have been different for younger age groups; indeed, the rate of obesity alone has more than doubled for the young over the past two decades.4 Other researchers have argued that as a result of changes to public disability insurance programs, the young face growing incentives to report disability, even if their actual level of disability has remained constant.5 The young seem to be facing greater disability risk from disease, and more powerful incentives to report disability. In this paper we use data from the National Health Interview Survey (NHIS) to present a detailed analysis of disability trends in the population ages 18–69. This analysis reveals substantial growth in reported rates of disability in population younger than age fifty but not among the elderly. These trends do not appear to be the result of changes in the composition of the population or of changes in specific survey procedures. In the interest of stimulating further research, we present several different explanations for rising disability levels: obesity, technological advances in medicine, and more generous disability insurance. We briefly outline the evidence that is consistent and inconsistent with each. None fits the data perfectly, but all contain important insights and the seeds for future analysis. Finally, we discuss the possible impact of disability among the young on public programs like Social Security Disability Insurance (SSDI), Medicare, and Medicaid.
We used NHIS data from 1984 to 2000, which contain individual-level information on the demographic and health status of a nationally representative sample of the U.S. civilian, noninstitutionalized population; these data have been collected every year since 1957.6 Although the survey was redesigned in 1982 and again in 1997, one can construct consistent estimates of disability from 1984 to 1996. (In 1982 and 1983 the NHIS did not ask the questions we use to measure disability.) Moreover, it is possible to construct a separate, internally consistent series from 1997 to 2000. Unfortunately, comparisons across the two series are not possible, because the survey instrument underwent considerable changes between 1996 and 1997. Nonetheless, from 1982 to 1996 and again from 1997 to 2000 there were no changes to the disability sections of the questionnaire. For all of the available NHIS years, we constructed estimates of disability that account for the complex survey design.7 From its inception, the NHIS has asked respondents whether their health limited their ability to work or perform housework. These questions are problematic for the measurement of disability because they are influenced by a person’s decision to stay in the labor force as well as by a person’s choice of occupations. For example, a given physical impairment could represent a work limitation for a bicycle messenger but not for an accountant. The NHIS began asking questions more suited to measuring disability in 1984. All respondents over age five reporting an activity limitation were asked if they needed help with personal care or other routine needs. These questions move toward a more common baseline of normal activity: Bicycle messengers might differ from accountants in the requirements of their jobs, but they probably differ by much less in their personal care requirements. Based on the answer to the personal care question, each respondent is placed in one of three ordered categories: (1) unable to attend to personal care needs; (2) limited in performing other routine tasks; and (3) not limited in attending to personal care or performing routine tasks. We say that a person has a "routine-needs" disability if he or she is in category 1 or 2 and that a person has a more severe "personal-care" disability if he or she is in category 1. We interpret category 2 to be less severe, and thus less inclusive, than category 1. Since both levels of disability seem to be caused by similar types of illnesses, the difference between them would appear to be one of severity and degree, rather than a fundamental qualitative difference in the kind of disability being measured. For example, by far the most common causes of both kinds of disabilities among people ages 18–39 are musculoskeletal (especially back) problems and mental illness.8 Disability data are also available in the 1997–2000 NHIS data sets, albeit with some modifications. Because of several subtle changes in the wording of the disability questions and in survey methodology, it is inappropriate to compare data from before and after the 1997 survey. However, we were still able to use the disability data from this period to group respondents into one of the three categories listed above. To smooth the disability trends within age categories, we pooled adjacent years of data. From the seventeen years of survey data, we constructed nine periods: 1984, 1985–86, 1987–88, 1989–90, 1991–92, 1993–94, 1995–96, 1997–98, and 1999–2000. We measured the raw change in disability over time, from 1984 to 1996 and from 1997 to 2000, as well as whether or not these raw changes differ statistically from zero. We used multivariate regression to assess whether or not the aggregate trends are driven by changes in the composition of the population. We adjusted for sex, race, Hispanic origin, employment status, and schooling.9 The adjusted numbers report the growth in disability that would have occurred absent changes in the composition of the population by race, sex, Hispanic origin, educational attainment, and employment status.
Aggregate trends. Exhibit 1  measures disability as the proportion of the population unable to attend to personal care needs or perform other routine tasks (either category 1 or category 2 above), and alternatively as the more restricted population unable to attend to personal care needs alone (category 1 only). It is worth reemphasizing that the numbers before and after 1997 are not directly comparable, although comparisons can be made within the two periods.
From 1984 to 1996 routine-needs disability expanded significantly for people ages 18–59 (Exhibit 1  ). There were significant expansions in the more severe personal care–limitation category for all respondents ages 18–39, although the effects were more modest for those ages 18–29. The sharpest increases in disability seem centered in the population ages 30–39. For populations ages 30–59 the absolute growth in both disability categories was roughly 65 per 10,000. The largest percentage increase occurred for those ages 30–39, whose rate of disability rose by more than 50 percent over this period. According to either definition of disability, the population ages 60–69 actually became less disabled. This is consistent with the findings from previous research. Moreover, it is quite likely that we are understating improvements in the health of the old, both because the NHIS excludes the institutionalized population and because the data set does not have very much statistical power at these older ages. Other researchers using the National Long-Term Care Survey (NLTCS) have found improvements in health even for those older than age eighty-five.10 The trends in personal-care and routine-needs limitations are similar, although growth in personal-care limitations seems to be confined to the youngest populations, ages 18–39, while routine-needs limitations grow even among the population ages 40–59. From 1997 to 2000 there were continuing, but modest and primarily insignificant, increases in disability for people under age fifty and insignificant declines for those over age fifty. It is likely that more years of data are required before we can draw firm conclusions about the period beginning in 1997, but a few interesting facts are worth noting. First, growth in the more severe personal-care-limitations measure seems to have occurred for those ages 40–49, who were the only ones under age fifty to escape such increases in the earlier period. Second, the biggest absolute declines in routine-needs disability occurred for those over age fifty. While these trends are not statistically significant, they are consistent with the differences between younger and older populations.
Accounting for compositional change.
Although the raw changes in disability are informative, we would like to know whether the changes in disability have been driven simply by changes in the composition of the population. Exhibit 2
From 1984 to 1996 composition-adjusted disability rates rose significantly for population groups ages 30–60 but did not change for those ages 60–69. There was significant growth over part of the period for those ages 18–29, but a retrenchment in 1996 eliminates significance. In percentage terms, the largest growth occurred for those ages 30–39, although those ages 40–49 experienced a more than 50 percent increase in disability. A comparison of Exhibits 1 and 2 reveals that adjusting for composition results in larger estimates of disability growth. At a minimum, therefore, compositional changes are not responsible for increases in disability; indeed, changes in the composition of the population dampened disability growth. This finding is consistent with our analysis of trends within smaller socio-demographic groups: Average disability rates seem to be rising for whites and nonwhites, for people inside and outside the labor force, and for all education groups. For example, among college-educated people ages 30–59, the age-adjusted rate of disability rose by 31 per 10,000, an increase that is statistically significant at the 95 percent level. Once again, the evidence for 1997–2000 is inconclusive. There are no statistically significant changes over this period, probably because the time frame is too short. If we take the point estimates at face value, however, it appears that growth in disability might be leveling off for the groups under age fifty, which are also the groups that experienced the largest increases from 1984 to 1996.
The traditional focus of disability research has been on the elderly, with good reason. Disability is much more prevalent among the elderly, and it has more direct impact on the demand for medical care in this age group. However, the emerging growth in disability among the young merits further attention by researchers and policymakers, because it has an important social and economic impact. It affects the demand for both medical care and publicly provided disability insurance; it also affects disability among tomorrow’s elderly. Developing and justifying a definitive explanation for the growth in reported disability from 1984 to 1996 is beyond the scope of this paper. However, here we suggest some possible candidates, along with the evidence that is consistent and inconsistent with each. There are at least two types of explanations, which are not mutually exclusive. The growth in disability could be the result of real deterioration in underlying health, or it could be the result of increased incentives to report disability.
Influence of obesity.
The recent rise in obesity levels represents one plausible reason why there might have been real deterioration in health. For each age group that we examined, the rate of obesity has increased by at least 700 per 10,0000 from 1984 to 1996, or by seven percentage points. Exhibits 3
Exhibit 3 illustrates that the obese tend to be more disabled than the nonobese, so rising obesity rates shift people from a less disabled to a more disabled state. The problem is even more serious than this factor alone would suggest, though, because rates of disability have also been growing more rapidly among the disabled than among the nondisabled. This has been especially acute among the population ages 50–59, in which disability has remained completely flat among the nonobese, even as it has risen by 151 per 10,000 among the obese.
To estimate obesity’s effect on disability, Exhibit 4
A more comprehensive view of the links between various diseases and disability appears in Exhibit 5
Technological advances in medicine. Paradoxically, another explanation for rising disability rates might be found in the advance of medical technology. Disability growth might actually reflect improvements in overall health, when viewed in the larger context of survival and mortality. Lifesaving treatments could forestall death for chronically frail people. This could lower the proportion of deceased people in a cohort but raise the proportion of living disabled people. Such growth in disability is actually consistent with overall improvements in health and welfare from advancing medical technology. This might be the most optimistic interpretation that emphasizes deterioration in underlying health. Its weakness lies in the difficulty of testing or disproving an explanation that emphasizes technological change: It is difficult to characterize the levels of disability and mortality that would have been obtained in the absence of medical breakthroughs. Role of disability insurance incentives. The above explanations seek to show why the health of the population might have deteriorated, but it is also possible that trends in disability reflect forces other than health, namely, changes in the economic incentives for disability insurance. As a result of the benefits formula employed, disability insurance has become relatively more attractive to less-skilled workers in recent years. Disability insurance benefits are inflated according to average wage growth, but as a result of expanding wage inequality, less-skilled workers are experiencing below-average wage growth. As a result, their potential disability benefits might be growing faster than their potential market wages. Several researchers have found that this trend could have encouraged less-skilled workers to file for disability insurance more often, even though they might not have become more disabled.14 Growth in per capita disability claims indeed occurred between 1984 and 1996, as did expansion in the incentives for disability insurance.15 It is possible that people who are no more disabled are nonetheless more likely to claim disability insurance and thus self-report disability in surveys. Just like the earlier explanations, however, this has weaknesses of its own. As mentioned earlier in our discussion of compositional change, rates of personal-care and routine-needs disability rose uniformly across all education groups and for people both outside and inside the labor force. In contrast, the disability insurance explanation would predict declining disability reports within the labor force, and it would predict that disability growth ought to be more pronounced among less educated groups. It is difficult to find these trends in our data.
Disability is not just a feature of old age. Economic development and technological change in health care have allowed people of all ages to live in frailty with greater ease than at any other time in history. From the mid-1980s to mid-1990s, reported rates of disability have grown among the young, even as the old have become relatively healthier. Moreover, this growth seems to have occurred within all demographic and economic groups, rather than within isolated segments of the population. Although the source of these trends is not yet clear, it is clear that rising rates of reported disability will have adverse consequences for public programs. Even if increased disability is the result of changing incentives for disability benefits and not the result of actual deterioration in health, the SSDI system is likely to feel the effects. David Autor and Mark Duggan report that the proportion of nonelderly adults receiving SSDI benefits rose from 3.1 percent to 5.3 percent between 1984 and 2000; in absolute terms, the number of nonelderly recipients doubled.16 These numbers are much larger in magnitude than the number of people with routine-needs or personal-care limitations, presumably because work-limiting disabilities are less severe than those that limit personal care. This suggests that the increase in routine-needs limitation could understate the impact on the SSDI program, which is even affected by less severe forms of disability. If—as we have argued—at least part of the increase in disability is attributable to deterioration in health among the young, the effects are likely to be felt in the U.S. health care system. Elsewhere, we have explored the future impact of disability among the young on the long-term care industry (and the Medicaid program that funds much of it) and on Medicare. Using forecasts based on data from the NHIS and the Medicare Current Beneficiary Survey (MCBS), we suggest that the recent growth in disability among the young could lead to a future nursing home population that is 10–25 percent larger than it might otherwise have been.17 Similar forecasting methods suggest that Medicare expenditures could be 10–15 percent higher than they would have been in the absence of this disability expansion.18 Given the inevitable increase in the level of resources devoted to medical care for an aging population, these percentage increases loom even larger for the future of the nation’s health care system.
Darius Lakdawalla is an economist at RAND in Santa Monica, California. Jay Bhattacharya is an assistant professor of medicine at the Center for Primary Care and Outcomes Research, Stanford University School of Medicine, in California. Dana Goldman is director and corporate chair, Health Economics, at RAND. The authors are grateful for the comments and suggestions of Michael Hurd, Geoffrey Joyce, and Constantijn Panis. This research was funded by the Centers for Medicare and Medicaid Services (Contract no. 500-95-0056) and the National Institute on Aging (Grant no. 5P01AGO8291). The statements and opinions contained in this paper are solely those of the authors, who assume full responsibility for their accuracy and completeness.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Study Methods
Study Results
