In 2006, DAHE accounted for $397.8 billion (26.7%) of all health-care expenditures for U.S. adults, which represents a substantial portion of U.S. health-care expenditures. In part, the magnitude of DAHE stems from the high disability prevalence in the adult population, with 18.2% of all adults reporting a limitation in any way in any activities because of physical, mental, or emotional problems. The NHIS-based definition of disability used in this study was broad and, therefore, included a large number of people, which contributed to the magnitude of DAHE. Even so, other definitions of disability, such as one used in a 2005 report14
by the U.S. Surgeon General, provide even higher estimates (22.0%) of the proportion of the population with a disability. If we had chosen a narrower disability definition, such as having only any deficits with ADLs or IADLs, the prevalence of people with disabilities would be lower, although we would likely have captured the highest-cost people.
The disabilities included in our study definition varied in their duration, severity, and cause. Some disabilities, such as an inability to walk because of a broken foot, may be temporary and have low treatment costs, but are experienced more frequently than a permanent disability. Conversely, a permanent disability may more likely be accompanied by chronic conditions, with significantly higher treatment costs than a short-term disability. We were not able to assess the effects on DAHE of short-term vs. permanent disabilities because of the lack of relevant data, nor were we able to distinguish the cause of the reported disability. Both issues could be explored in future work.
The proportion of DAHE of 26.7% was higher than the prevalence of disability of 18.2% because people with disabilities use disproportionately more services than people without disabilities.3
Even so, DAHE are relative to the health-care costs of all people with chronic conditions. Many more people have one or more chronic conditions or diseases than report disability,15
and DAHE are smaller than the total cost of chronic conditions and diseases in the U.S. adult population.
The DAHE national estimate masks substantial variation across states, and across payers within states. This variation was associated with variations in the two factors that we multiplied to estimate DAHE for each state and payer within each state: (1) the proportion of all adult DAHE and (2) total adult health-care expenditures. The proportion of a state's total DAHE was associated with the relative distribution of costs across payers within states, with each payer having different proportions of DAHE. Generally, higher disability prevalence among Medicare and Medicaid beneficiaries in a state was positively associated with the proportion of DAHE. (Detailed information on disability prevalence across payers is available online in the Technical Appendix.)
In some cases, states may have similar DAHE but very different proportions of total DAHE. For example, Colorado had a relatively low proportion (20.8%) of DAHE and total health-care expenditures of $21 billion in 2006, giving it DAHE of $4.4 billion. Alternatively, Mississippi had a much higher proportion (32.5%) of DAHE but only two-thirds that of Colorado's total health-care expenditures ($14 billion), giving it approximately the same DAHE ($4.5 billion).
Variations in DAHE in the community population across states are driven by demographic differences, primarily age. States with a larger proportion of older people (e.g., Pennsylvania and West Virginia) are more likely to have higher disability prevalence and, therefore, higher DAHE. Conversely, state variation in DAHE per person living in an institution or per community-dwelling recipient of long-term-care services is driven by the generosity of the states' Medicaid programs.
The Medicaid program is not only the largest payer of DAHE among all payers nationally, but it is also the largest payer of DAHE in two-thirds of the states (33 of the 50 states plus DC) for two reasons. In 2007, Medi-caid paid for the care of almost two-thirds (64.6%) of the 1.5 million Americans in nursing homes;16
payments for this care accounted for 30.6% of all Medicaid DAHE. Second, Medicaid programs in states in the Northeast (e.g., New York) have particularly generous home- and community-based long-term-care programs.17
In addition, while the difference in mean disability prevalence between the Medicaid and Medicare programs was less than 8 percentage points (46.3% vs. 39.0%), the proportion of DAHE was almost twice as high for Medicaid as for Medicare (68.7% vs. 38.1%) because of the high cost of institutional care paid by Medicaid. Therefore, the Medicaid program disproportionately bears the largest share of DAHE in the U.S.
Although DAHE per person with disability were similar for states in the West and Midwest regions, they varied more for states within the Northeast and Southeast regions, primarily because of greater state-to-state differences in disability prevalence and in the prevalence and cost of institutionalization. DAHE per capita were particularly high in the Northeast region (even though disability prevalence rates in Northeast states were generally below the national mean) and in the western half of the Southeast region. DC had the highest DAHE per capita and per person with -disability in part because of the high rates paid to nursing homes.18
New York, which had the second highest DAHE per capita and per person with disability, has by far the largest Medicaid personal care program in the U.S.17
This study's findings demonstrate the need for interventions to prevent or delay disability. Some disability risk factors (e.g., falling, inactivity, and depression) can often be addressed through individual-level interventions. In a review of the literature on individual-level interventions, Freedman et al. found that in the short term, multicomponent fall-prevention interventions would likely have a greater effect on reducing the risk for disability than exercise or depression-treatment interventions alone.19
These findings also provide a financial motivation to identify cost-effective strategies to effectively manage the treatment and costs of people with disabilities today. DAHE may be reduced by using preventive care services and health promotion interventions, and by improving access to medical care for people with disabilities to reduce the incidence of secondary conditions to disability through early diagnosis and intervention. For example, disease management programs may help maintain functional independence by preventing or delaying the onset of chronic conditions, resulting in decreased hospitalization and premature nursing home use. At a system level, programs to integrate Medicare and Medicaid payment through capitation for acute and long-term-care services (e.g., the Program of All-inclusive Care for the Elderly [PACE]) or pay-for-performance initiatives20
might motivate providers to deliver high-quality care cost-effectively across targeted settings.
This research had several limitations. First, because we lacked state-level data containing information on both disability prevalence and health-care expenditures, we had to use data from several sources to develop a synthetic estimate of state-level DAHE and, thus, were unable to calculate standard errors for our state-level estimates (see Technical Appendix).
The use of multiple data sources had several ramifications. To develop our DAHE estimate, we inferred that our NHIS/MEPS model estimates applied to the BRFSS population, even though the characteristics of people with disabilities differed between the two datasets. The BRFSS sample population with disabilities was considerably younger, much better educated, considerably less poor, and less likely to have Medicare and Medicaid than the NHIS/MEPS sample population with disabilities. In addition, we had to make predictions of BRFSS sample member insurance status using NHIS/MEPS estimates because the BRFSS sample lacked indicators of insurance status. Consequently, we adjusted our BRFSS DAHE estimate and payer distribution to the NHIS/MEPS DAHE estimate and payer distribution because the NHIS/MEPS population is more like the U.S. adult population.
Second, the DAHE estimate was large in part because we did not explicitly control for injuries and specific diseases and chronic conditions often associated with disability. Thus, some—but not all—of those costs were included in our estimate. Had we controlled for these additional health conditions, the DAHE estimate would have been smaller than reported. However, we used a standard econometric approach common in health services research to estimate the costs of single conditions or health problems.9–13
Finally, both BRFSS and NHIS/MEPS rely on self-reported data, and responses can vary by survey administration mode. Recent work by Walsh and Khatutsky21
has shown that in-person surveys underreport disability levels. In this study, while we used a common measure of disability from both surveys, the disability prevalence in the BRFSS (a telephone survey) was 18.2%, while the disability prevalence in the NHIS (an in-person survey) was 13.6%.