The Impact of Disability Trends on Medicare Spending


U.S. Department of Health and Human Services

The Impact of Disability Trends on Medicare Spending

Executive Summary

Brenda C. Spillman

The Urban Institute

September 2, 2005

This report was prepared under contract #HHS-100-97-0010 between the U.S. Department of Health and Human Services (HHS), Office of Disability, Aging and Long-Term Care Policy (DALTCP) and the Urban Institute. For additional information about this subject, you can visit the DALTCP home page at or contact the ASPE Project Officers, William Marton and Hakan Aykan, at HHS/ASPE/DALTCP, Room 424E, H.H. Humphrey Building, 200 Independence Avenue, S.W., Washington, D.C. 20201. Their e-mail addresses are: and

The opinions and views expressed in this report are those of the authors. They do not necessarily reflect the views of the Department of Health and Human Services, the contractor or any other funding organization.

It is now generally agreed that there has been a significant decline in the elderly disability rate in the United States since the early 1980s, but much less is known about implications of the decline for Medicare spending. This is in part because uncertainty persists about the extent to which the aggregate disability decline reflects improvements in health. If declines are not due--or not entirely due--to improvements in health, this suggests uncertainty about the implications of disability declines for future Medicare costs.

Underlying the downward trend in aggregate disability, are different trends for different types of disability:

  • Most of the decline occurred for persons reporting chronic help with only instrumental activities of daily living (IADLs)--activities such as housekeeping and meal preparation associated with the ability to live independently.

  • The prevalence of chronic disability managed solely with the use of assistive devices increased significantly.

  • There was, at most a modest reduction in assistance with activities of daily living (ADLs), which are associated with poorer health and more serious disability.

  • The trend in the prevalence of institutional residence was flat.

The purpose of this study is to better understand the relationship between disability declines and Medicare costs, and what Medicare spending patterns may suggest about the relationship between disability and health. To that end, the study examines actual patterns of Medicare spending and utilization that occurred over the period of declining disability between 1984 and 1999, and how they differ from what might have been expected had disability not changed. Spending is examined in the aggregate, by service, and by type of disability, and implications for the relationship between disability, Medicare spending, and health are discussed. Based on the results, Medicare spending projections are developed under various assumptions about how disability and spending are likely to change over the over the next several years.


The basic analytic strategy in this study is to compare actual Medicare spending per capita--and per capita spending by persons with chronic disability relative to persons without chronic disability--with counterfactual estimates assuming that disability rate declines had not occurred. The counterfactual takes into account not only the aggregate trend in disability but differences in trends for the four different types of disability. Differences between the counterfactual and the actual values reflect the impact of declining disability on spending.

Data are from the 1984, 1989, 1994, and 1999 waves of the National Long-Term Care Survey (NLTCS), merged with Medicare claims data. The NLTCS is a nationally representative survey of Medicare enrollees aged 65 or older designed to identify those who are chronically disabled in one or more ADLs or IADLs and to collect detailed data on their disability, service use, family support, and health and demographic characteristics. The samples are drawn from Medicare enrollment files and represent both community and institutional residents.

Disability is defined as receiving help (including supervision) or using equipment to perform ADL activities, being unable to perform IADLs without help because of health or disability, or living in an institution, which is defined as a group setting with medical supervision. The ADLs included in the present analysis are eating, getting in and out of bed (transferring), getting around inside, dressing, bathing, and toileting. The included IADLs are light housework, laundry, shopping, meal preparation, getting around outside, managing money, taking medications, and telephoning. Chronic disability is defined as receiving help (including standby help, or supervision) or using equipment for any of these 14 activities for at least 90 days, or being in an institution.

Chronically disabled individuals are assigned to one of the four mutually exclusive disability groups identified above, defined by residence, type of disability, and receipt of help. Community residents are classified as receiving ADL assistance if they reported chronic help with any ADL, as receiving only IADL help if they report chronic help with IADL activities but no chronic ADL help, and as independent with equipment if they reported no chronic help with any activity but used assistive devices to perform at least one activity. The final group comprises all persons residing in institutional settings.

Medicare fee-for-service claims data for the years 1982-2000 have been linked to all persons surveyed in any wave of the NLTCS. Because Medicare data are not available for managed care enrollees they are excluded from the analysis. In order to link Medicare utilization and spending with a baseline disability measure, an analysis period was constructed for each survey year that begins with the earliest survey interview date and ends one year later. The analysis focuses on total spending and spending for four types of service that are most analytically important for trends: inpatient hospital spending; combined Part B physician supplier, durable medical equipment, and outpatient spending; skilled nursing facility (SNF) spending; and home health spending.

Spending and utilization estimates are standardized to the 1999 age and gender distribution. To examine the counterfactual of spending if the disability rates had not changed, comparison estimates were produced after further standardizing the sample to 1984 disability rates by age and gender.


  • As a group, the nondisabled, who made up an increasing proportion of the older population, became more expensive on a per capita basis.

  • Spending for persons with chronic disability fell progressively, so that the gap in spending between the disabled and nondisabled progressively narrowed.

  • These spending patterns were observed for all of the four major service areas examined.

Within the disabled population:

  • The two least expensive groups--persons managing their disability with only equipment, a group increasing in prevalence, and persons with help only with IADLs, a group with falling prevalence--became more expensive.

  • The two most costly groups--persons who received help with ADLs or were institutionalized, whose age-adjusted prevalence has fallen--became less expensive.

  • Increases in hospital, SNF, and home health use for the two less disabled groups, relative to the counterfactual, may indicate poorer or less stable health owing to changes in the composition of these groups.

Projections of future spending were made using two assumptions about how disability will change over the 20 years between 1999 and 2019 and three spending projection methods. The disability assumptions were:

  • The age and gender standardized aggregate disability rate will decline at the historical rate of 1.4 percent per year observed between 1984 and 1999.

  • Historical trends in the four underlying types of disability will continue, generating a lower, 0.5 percent per year decline in standardized aggregate disability.

Spending projections, at historical rates, were made using per capita total spending for the nondisabled and the disabled, per capita spending by service for the nondisabled and the disabled, and projected per capita total spending by type of disability. Projection results revealed that all spending projection methods that take disability rates into account generated higher future spending than naïve projections ignoring disability. The highest projected spending was for spending by service.

  • Savings under the assumption of a 1.4 percent per year age and gender standardized disability decline were modest, ranging from 5 percent to 7 percent depending on the spending method by 2019.

  • Savings under the assumption of a 0.5 percent per year decline, assuming that underlying standardized trends in type of disability continue, ranged from 2 percent to 2.6 percent.

  • When projected changes in the age and gender distribution were taken into account, the influx of younger persons after 2009 when the Baby Boom begins to enter Medicare moderated spending, savings for the larger disability decline fell, and savings for the smaller decline based on underlying trends by disability type rose.

  • Savings continued to be larger for the larger disability decline assumption, except for spending projected by service, underscoring the important of changes in service mix for future Medicare spending.


The Medicare spending patterns observed in the NLTCS appear to indicate subtle changes in both service use and composition of the nondisabled and disabled population, as well as for subgroups of the disabled population defined by type of activity and use of assistance. Specifically, as seen, when the population is standardized to the 1999 age and gender distribution, both persons without disability and those able to manage disabilities solely with equipment or requiring help only with IADLs appear to have become steadily more expensive for Medicare on a per capita basis than if disability not declined. The reverse was true for persons with more severe disability, defined here as requiring ADL help or being institutionalized, who appear to have become less expensive. As a result of these patterns, the gap in spending between the nondisabled, the disabled as a whole, and for persons with more serious disability progressively narrowed, while the spending gap for persons with less serious disability widened slightly, relative to the counterfactual if disability rates had not changed.

The results provide evidence that the link between disability declines, better health, and lower Medicare spending is complex. If persons who do not report chronic disability and persons who manage all disabilities with equipment continue to become relatively more expensive for Medicare and continue to increase in prevalence, Medicare savings from declining spending and prevalence of more serious disability requiring long-term care would be offset to some extent. Among the nondisabled, persons with a prior history of disability were found to be nearly as expensive for Medicare as the currently disabled and far more expensive than persons reporting no current or prior disability. Although this group is quite small and showed no consistent growth between 1984 and 1999, this suggests that “recovery” from disability and recovery of health are not synonymous, at least as measured by Medicare spending.

Finally, controlling for age and gender, both hospital and post-acute spending were higher for persons with less serious disability than they would have been for persons with these disability profiles in the past. It is possible that this indicates environmental improvements that allow more independent function--or even no perception of disability--among persons with medical conditions associated with higher costs. It is also possible that such improvements have reduced the ability to measure disability consistently over time, so that some of the decline observed reflects measurement error rather than improvements in physical functioning, further confounding the ability to rely on reductions in disability rates as a proxy for improved population health (Cornman, Freedman, and Agree 2005).

In the current policy and demographic climate, with the retirement of the Baby Boom looming, there is considerable pressure to assess implications of disability declines for public program costs and particularly for Medicare. In fact, the period covered by the projections in this study stops just short of the aging of the early Baby Boom cohorts into ages where disability rates are high. However, the findings in this study and others highlight the importance of more critical thinking about where research should focus in order to understand causes of disability rate declines and the importance of observed declines in current disability for future health costs.

The Full Report is also available from the DALTCP website ( or directly at