Health Care Expenditures of Living With a Disability: Total Expenditures, Out-of-Pocket Expenses, and Burden, 1996 to 2004

Article Outline

• Abstract
• Methods
  • Data Source
  • Participants
    • Disability definitions
  • Dependent Variables
    • Annual total health care expenditure
    • Out-of-pocket spending on health care
    • Out-of-pocket spending burden
  • Statistical Methods
• Results
  • Alternative Specifications
    • Second definition of disability
    • Two-part model
• Discussion
  • Study Limitations
• Conclusions
• References
• Copyright

Abstract 

Mitra S, Findley PA, Sambamoorthi U. Health care expenditures of living with a disability: total expenditures, out-of-pocket expenses, and burden, 1996 to 2004.

Objective

To estimate the health care expenditures associated with a disability and their recent trends.

Design

Retrospective analysis of survey data.

Setting

Not applicable.

Participants

Data from multiple years (1996–2004) of the Medical Expenditure Panel Survey (MEPS) for a nationally representative sample of civilian, noninstitutionalized U.S. population.

Interventions

Not applicable.

Main Outcome Measures

Health care expenditures consisted of total health care expenditures, total out-of-pocket (OOP) spending, and burden (the ratio of OOP to family income). All the analyses accounted for the complex survey design of the MEPS.

Results

Between 1996 and 2004, 6% to 9% of persons in the working-age group (21–61y) were identified as having a disability. Persons with disabilities consistently had higher total health expenditures, OOP spending, and burden compared with their counterparts without disabilities. In 2004, the average total expenditures were estimated at $10,508 for persons with disabilities and at $2256 for those without disabilities. In a multiple regression framework, persons with disabilities were consistently found to have higher expenditures, OOP spending, and burden between 1996 and 2004. Although expenditures, OOP spending, and burden increased over time, after controlling for demographic, socioeconomic, and health status, these 3 health care costs were not found to change disproportionately for persons with disability.

Conclusions

During the 1996 to 2004 period, persons with disabilities were consistently found to have significantly higher health expenditures, OOP spending, and burden compared with their counterparts without disabilities, which may adversely affect their health and standard of living.

Key Words: Health expenditures, Rehabilitation

List of Abbreviations: ADA, American with Disabilities Act, GLM, generalized linear model, ICF, International Classification of Functioning, Disability and Health, MEPS, Medical Expenditure Panel Survey, OLS, ordinary least squares, OOP, out-of-pocket, SSDI, Social Security Disability Insurance, SSI, Supplemental Security Income

IT IS ESTIMATED THAT, in the United States, between 40 and 50 million people have a disability.¹ For working-age persons, recent estimates range between 20 and 23 million.² During the 1990s, the disability prevalence rate among the working-age population increased.³ At the same time, among all adults younger than 65 years, health care costs have been rising, and an increasing proportion of family budgets have been for OOP spending,⁴ particularly for people with chronic health conditions (ie, medical conditions that require ongoing treatment).⁵ For those younger than 65 years of age with chronic conditions, the highest OOP expense has been reported to be physician office visits, and if they are uninsured, they are 5 times less likely than those with insurance to see a medical provider in a given year.⁶ There have been studies of trends in expenditures for specific chronic conditions.⁷ However, such studies do not cover the expenditures associated with disability because disability is not a medical attribute of the person. Indeed, disability has increasingly been considered as taking place at the interface of the person and the physical and social environment. Having a disability does not require having a medical chronic condition, and in reverse, a medical chronic condition does not necessarily lead to a disability. Among persons with disabilities, several studies have demonstrated that they face a higher cost of living in general⁸ and higher health care expenditures in particular, compared with persons without disabilities.⁹ Activity limitations alone have also been noted to increase the cost of care.¹⁰ The financial burden of OOP spending as a proportion of total family income was 2.5 times greater for a family with a person with disability compared with families without a person with a disability in 1987,¹¹ even after controlling for availability of health care insurance.

Understanding how the increasing trend in health care expenditures, OOP spending, and burden for the general population is affecting persons with disabilities is essential for several reasons. Poor families with a member with a disability have been shown to resort to health care rationing.¹¹ Recent trends in health expenditures for working-age persons with disabilities may also have implications for future health care spending as this population becomes older and eligible for Medicare.¹² A major policy concern has been that the rising cost of health care contributed to the decline in the employment of persons with disabilities during the 1990s, but evidence of an impact on health insurance is unclear.¹³ Finally, several policy initiatives during the 1990s and early 2000s were expected to affect persons with disabilities' access to health care services and insurance (eg, Medicaid Buy-In programs for persons with disabilities).

Past research has shown that persons with disabilities overall have greater health care expenditures than their counterparts without disabilities, but estimates of the magnitude of extra health care expenditures are outdated. For example, using 1997 MEPS data, Yelin et al⁹ found that a person with a disability faces an increment in medical care expenditures of $2953 beyond what would be expected of similar persons without disabilities. Persons with disabilities often, but not always, need substantial health care services and support.¹⁴ For example, a person who loses a limb may experience a high demand for health care in the short-term, but not after the condition has stabilized.¹³ It is also possible that persons with disabilities underutilize health care services because of their limited access to health care providers and facilities,¹⁵, ¹⁶ which may result in lower health care expenditures.

However, for total health expenditures, we expect to find an increase for persons with disabilities that is disproportionate compared with that for persons without disabilities because of several key developments in federal policies—the 1990 ADA and the 1999 Olmstead decision in particular. One may expect greater access to health care services for persons with disabilities made possible by the ADA. Title II of the ADA protects qualified persons with disabilities from discrimination on the basis of disability in accessing services, programs, or activities provided by state and local governments. Title III prohibits discrimination against people with disabilities by places of public accommodation, and private health care providers are considered places of public accommodation. However, the ADA did not specifically prohibit private insurance companies from discriminating on the basis of actuarial risk. The Health Insurance Portability and Accountability Act of 1996 did attempt to extend coverage to persons with preexisting conditions, but this protection, together with that of the Consolidated Omnibus Budget Reconciliation Act of 1986, is hardly sufficient.¹⁷, ¹⁸ In addition, the Supreme Court's Olmstead decision of 1999 may have also put some upward pressure on health expenditures for persons with disabilities. This decision challenged states to provide community placements and develop more accessible services for persons with disabilities, and continued a move toward deinstitutionalization started in the early 1980s. Section 2176 of the Omnibus Budget Reconciliation Act of 1981 (Public Law 97-35) added section 1915(c) to the Social Security Act for the Medicaid program to create Home and Community-Based Services Waivers. States were allowed to waive or set aside some of the Medicaid provisions to allow long-term care services to be delivered in community settings outside of institutional settings for those with disabilities. Being in the community may generate expenditures to the person that were previously covered through an institution.

While total health expenditures are expected to have increased disproportionately more for persons with disabilities, OOP spending and burden, on the contrary, may have increased disproportionately less for persons with disabilities compared with their counterparts without disabilities. Indeed, several policy initiatives developed at the federal and state level in the 1990s and 2000s may have placed a downward pressure on OOP spending and burden among persons with disabilities. There have been several initiatives to make public health insurance benefits more widely available among persons with disabilities, and in particular to break the link between public health insurance benefits (Medicaid and Medicare) and income support benefits (ie, SSI and SSDI). Most notably, section 4733 of the Balanced Budget Act of 1997 and provisions of the Ticket to Work and Work Incentives Improvement Act of 1999 (Ticket to Work Act thereafter) encouraged, but did not require, states to implement Medicaid Buy-In programs, under which workers with disabilities who would qualify for SSI or SSDI but earn more than the allowable limits for regular Medicaid can purchase Medicaid benefits at a heavily subsidized premium. Thirty-two states have implemented the Medicaid Buy-in Program. Finally, OOP spending and burden may have been influenced by general developments related to disability benefits for persons with disabilities. Disability benefit programs such as SSDI and SSI have grown very rapidly through the 1990s and early 2000s,¹⁹ which may have had an effect on the personal income of persons with disabilities and reduced OOP spending through access to Medicare and Medicaid, respectively. Overall, the expected increase in public health insurance coverage leads us to expect that OOP spending and burden for persons with disabilities would have increased disproportionately less than for their counterparts without disabilities.

Despite increased disability prevalence among working-age people, very little is known about health care expenditures, OOP spending, and burden among persons with disabilities and their recent trends. To the best of our knowledge, no study has so far assessed, based on recent data, the disparity and trends in health care expenditures, OOP spending, and burden across disability status. The objective of this study is to compare health care expenditures, OOP spending, and burden for people with and without disabilities aged between 21 and 61 years and their trends between 1996 and 2004 using nationally representative household data from the MEPS. This study takes the perspective of the person living with the disability rather than the perspective of insurers or society at large coping with the costs, as so frequently reported.²⁰ Our hypotheses are that, for persons with disabilities compared with persons without disabilities, (1) expenditures, OOP spending, and burden are consistently higher, (2) total health expenditures have disproportionately increased, while (3) OOP spending and burden have increased relatively less during the 1996 to 2004 period.

Back to Article Outline

Methods 

Data Source 

We used data from the household component of the MEPS. The MEPS, which began in 1996, is a nationally representative household data set on health care use, expenditures, sources of payment, and insurance coverage for the noninstitutionalized civilian population. The MEPS also collects detailed information on activity and functional limitations, which are used in our article to define disability. The MEPS has an overlapping panel design: for each panel, 6 rounds of interviews are conducted over a period of 2.5 years. We used data from selected years only (1996, 1998, 2000, 2002, 2004) to avoid repeated observations.

Participants 

We restricted our study sample to people in the working-age group (ie, 21–61y). We used 61 years as the cutoff point instead of 64 years to avoid including persons who have transitioned to early retirement under the Social Security Administration Old Age program. We also excluded part-year observations (eg, persons who could not be located and persons who died during the calendar year). By removing persons who die during the year, we avoid the challenge of comparing part-year expenditures with full-year expenditures, and we reduce the skew in the data. However, health expenditures in the last year of life are known to be high, so removing persons who died during the year is likely to lead to an underestimate of actual expenditures and might bias our estimates if mortality occurs differentially across disability status. Thus, our final study sample consisted of 10,987 persons in 1996, 11,407 in 1998, 12,082 in 2000, 18,854 in 2002, and 16,459 in 2004.

Disability definitions 

Defining and measuring disability is challenging. Disability has been defined through different conceptual models that lead to different measures. (A detailed coverage of these models is available in Altman²¹ and Mitra.²²) For instance, in the medical model, disability is caused by a disease, an injury, or other health conditions and requires medical care in the form of treatment and rehabilitation. Under this model, persons with any impairment are considered disabled, regardless of whether they experience limitations in their life activities caused by the impairment. An impairment is an anatomic or physiologic loss (deaf, blind) caused by a pathologic condition. On the other hand, in the social model, disability is understood as a social construct; “disability is not the attribute of a person, instead it is created by the social environment and requires social change.”^22(p237) A third model of interest is the ICF developed by the World Health Organization. ICF is conceptually an integration of the medical and the social models.²³ In the ICF, disability is an umbrella term that covers impairments, functional limitations (eg, limitation in walking), and participation restrictions (eg, restrictions in employment). In addition to the conceptual challenge of defining disability, several researchers have empirically demonstrated the importance of using multiple disability measures.²⁴ Therefore, we use 2 measures of disability that can be understood as part of the ICF model. Our first measure of disability is based on major activity limitations, which is typically used in disability research for the working-age population. (We could not use definitions of disability based on limitations in activities of daily living and in self-care tasks [instrumental activities of daily living] because of small sample sizes. Such definitions are usually used for the elderly.) For each panel, in rounds 3 and 5, people were queried as to whether they had any limitation in work, housework, or school. Persons who answered positively in either round were recorded as disabled. It is important to note that this activity limitation measure may lead to an overestimate of the burden because persons with work limitations are likely to have lower earnings compared to persons without work limitations. This might affect our estimates of burdens for given years, but is not expected to affect the trend analysis. This limitation makes it important to use other measures of disability. Our second measure of disability was derived using functional limitations. In 2 rounds of the MEPS, respondents were queried if they had a walking limitation or a cognitive limitation (people were asked if they experienced confusion or memory loss, had problems making decisions, or required supervision for their own safety). In addition, in one round of the survey, functional limitation questions related to seeing and hearing were asked. For each of the functional limitation questions described above, the person can only answer yes or no. In our second measure of disability, a person who reported at least one functional limitation in one round was considered as having a disability.

Dependent Variables 

Annual total health care expenditure 

Within the MEPS, health care services that were paid for by third-party payers and/or persons themselves are defined as health expenditures and reported for each year. We used persons' annual total expenditures that were summed across inpatient, emergency department, outpatient (eg, clinic and office-based visits), pharmacy, and other (eg, home health services, vision care services, dental care, ambulance services, and medical equipment). All expenditures were adjusted for inflation and converted to constant 2004 dollars with the use of the consumer price index for medical care services.

Out-of-pocket spending on health care 

In the MEPS, total expenditures were also categorized by 11 major sources of payment. (These sources were (1) OOP expenditure by user or family; (2) Medicare; (3) Medicaid; (4) private insurance; (5) Veterans Administration, excluding Civilian Health and Medical Program of the Department of Veterans Affairs; (6) Tricare; (7) other federal sources—includes Indian Health Service, Military Treatment Facilities, and other care provided by the federal government; (8) other state and local sources—includes community and neighborhood clinics, state and local health departments, and state programs other than Medicaid; (9) workers' compensation; (10) other unclassified sources—includes sources such as automobile, homeowner's; and (11) other private sources—any type of private insurance payments reported for persons.) Total expenditures include OOP expenditures paid by the person or the family. Our measure of OOP expenditures does not include expenditures for health care insurance premiums because data on premiums paid are not available in the public use files of the MEPS.

Out-of-pocket spending burden 

We measured the burden of OOP spending as the percent of family income spent OOP, because the ability to pay for health care costs not covered by third-party payers is an important determinant of the use of health care services.

Statistical Methods 

Differences in levels and trends in total health expenditures, OOP spending, and OOP burden across disability status were examined using linear regression models. Because of the highly skewed nature of the data, we transformed total health expenditures, OOP expenditures, and burden into logarithmic terms using ln(outcome+1). Many studies use GLMs with log link to examine the relationship between expenditures and other variables.²⁵ We compared the deviance and kurtosis to assess the fitness of the GLM log-normal model as an alternative to the OLS log-normal model. We found in all years, the total health care expenditures were right-skewed and had a kurtosis value greater than 3, which suggested that the OLS log-normal model is preferable to the GLM model.²⁶ All analyses controlled for the complex sampling design. (The MEPS has a complex sample design with stratification, clustering, multiple stages of selection, and disproportionate sampling.²⁷, ²⁸) This complex survey design requires special adjustments with regard to variance estimation and analysis for results to be nationally representative. (Appropriate commands in the SAS^a version 9 software were used to have such an adjustment done in the analysis.)

The linear model described above does not distinguish between group composition changes and differences in regression coefficients across disability status. To understand the factors that contribute to the difference in health expenditures between persons with and without disabilities, we follow the Oaxaca decomposition method (eg, Oaxaca and Ransom²⁹). Oaxaca decomposes the gap in the mean of outcome between 2 groups; it allows for the possibility that the gap in outcome (log health expenditures, log of OOP expenditures, or burden) is caused in part by differences in the effects of determinants (regression coefficients). For example, the health expenditures of persons with disabilities may be less responsive to changes in insurance coverage. The Oaxaca decomposition starts with a multiple regression analysis of the outcome variable as follows:

where ln is the log of health expenditures of individual i, D and ND denote persons with and without disability respectively, is the intercept, is a set of characteristics, which include demographic, socioeconomic, health variables and a time variable, is a set of coefficients on those characteristic variables, and is the error term for person i. The difference in mean health expenditures between persons with and without disabilities can be decomposed as: On the right-hand side of equation 2, the first term is the difference in health expenditures that is attributable to differences in characteristics, while the second term represents the part of the expenditure difference that is unexplained. The second term may result from differences in unobservables and in returns to characteristics between the 2 groups.

Back to Article Outline

Results 

Disability prevalence as measured by the percentage of working-age persons with a major activity limitation has increased from 6.7% in 1996 to 9.4% in 2004 (.01≤P<.05). Persons with disabilities are significantly more likely to be older, poor, less educated, and to have public insurance and less likely to be employed than the nondisabled (table 1).

Table 1. Description of Sample by Disability (Activity Limitation) Status, Medical Expenditure Panel Survey—1996, 1998, 2000, 2002, 2004

NOTE. Based on persons aged between 21 and 61. Asterisks represent significant differences in sample composition by disability status based on chi-square statistic. Perceived health and mental health questions are inquired in each round of MEPS. Persons were categorized as having fair/poor health if they reported being in fair/poor health in any of the 3 rounds.

Abbreviations: NS, not significant; Sig, significance; USC, usual source of care; wt %, weighted percentage.

⁎.01≤P<.05;

†P<.001;

‡.001≤P<.01.

Table 2 displays the levels and changes over time in all 3 measures of health care expenditures between 1996 and 2004 by disability status. In 2004, median total health expenditures were estimated at $649 for those without disabilities, and $4449 for persons with disabilities, and median OOP spending was $280 and $703 for persons without and with disabilities, respectively. The median burden of OOP on family income was 1% for persons without disabilities but 4.4% for persons with disabilities. The descriptive evidence in table 2 thus provides support to our first hypothesis that persons with disabilities have substantially higher total health expenditures, OOP spending, and burden compared with persons without disabilities.

Table 2. Trends in Total Health Care Expenditures, OOP Spending (2004 $), and Burden by Disability Status, Medical Expenditure Panel Survey (1996–2004)

NOTE. Based on persons who were alive as of the end of calendar year and aged between 21 and 61. Total expenditures and OOP spending are expressed in 2004 dollars based on the consumer price index for medical services. The ratio is, for each indicator of interest (eg, mean total expenditures), that of persons with disabilities divided by that of persons without disabilities.

During the 1996 to 2004 period, the gaps in mean and median total health expenditures across disability status have increased. As shown in the ratio column, the median health expenditures in 1996 for persons with disabilities were 6 times higher than those for persons without disabilities, but 7 times higher in 2004. This result supports our second hypothesis that health expenditures have disproportionately increased for persons with disabilities. OOP spending has increased for both groups of interest, but increased more for persons with disabilities. Median OOP expenditures grew by 124% for persons with disabilities and by 62% for persons without disabilities. The median burden has increased by 25% from 0.8 to 1 for those without disabilities, and increased by 22% from 3.6 to 4.4 for those with disabilities. Our third hypothesis that OOP spending and burden have increased less for persons with disabilities is only partly supported by the data in table 2. For persons with disabilities, the burden has increased less than for persons without disabilities, but OOP spending has increased more.

The first column of table 3 shows unadjusted trends in total health care expenditures, OOP spending, and burden. In this model, only time (0–8y), disability status, and a time-disability interaction term were included as independent variables. The estimated parameters of the time and disability variables are positive in all 3 models, which indicate that health expenditures, OOP spending, and burden have increased during the period of interest and that there are significant differences in health expenditures, OOP spending, and burden over time and across disability status. The coefficient of the time-disability interaction term is positive and significantly different from zero in the regression of health expenditures and burden, but imprecisely estimated for OOP expenditures. This suggests that health expenditures and burden may have disproportionately increased for persons with disabilities compared with those without disabilities.

Table 3. Unadjusted and Adjusted Trends in Total and OOP Expenditures and Burden by Disability Status, Medical Expenditure Panel Survey—1996, 1998, 2000, 2002, 2004

NOTE. Based on persons who were aged between 21 and 61. Adjusted trend is based on ordinary least squares regression on logged expenditures for total and OOP spending. Model 1 adjusted for sex, race/ethnicity, age, marital status, region of residence, education, employment, and poverty status. Model 2 additionally included controls for physical illness, mental illness, and perceived physical and mental health status.

Abbreviations: HH, household; Sig, significance.

⁎.01≤P<.05;

†P<.001;

‡.001≤P<.01;

§.05≤P<.10.

In the second and third columns of table 3, we control for demographic and socioeconomic characteristics. Given that perceived health status and mental health status are highly correlated with disability status (as shown in table 1), we first leave out health characteristics in model 1 and then include them in model 2. After the health covariates are introduced, the regression coefficient of the disability binary variable is reduced but remains positive and significantly different from zero for health expenditures (1.895–0.988), OOP spending (0.988–0.498) and burden (0.828–0.432). In model 2, having a disability is associated with an increase in total health expenditures by 168%, in OOP spending by 65%, and in burden by 54%. The models with and without the health covariates suggest that disability status is an independent and significant predictor of expenditures, OOP spending, and the burden. These results provide support to our first hypothesis that persons with disabilities have higher total health expenditures, OOP spending, and burden.

Finally, in models 1 and 2, we find that, after characteristics of persons are controlled for, disability-time interaction terms become close to zero and insignificant for the 3 outcomes, which suggests that there has not been any disability-specific trend in outcomes during the 1996 to 2004 period. This result stands against our second and third hypotheses that there have been disproportionate trends in total health expenditures, OOP spending, and burden.

We next turn to the decomposition of the gap in health care expenditures. During the 1996 to 2004 period, we observe a gap in the mean log of total health care expenditures of −2.31 between persons with and without disabilities (7.69 vs 5.38). After adjusting for characteristics, the estimated gap is −1.88 (7.28 vs 5.40), or 35 percentage points. This negative difference reflects the fact that persons with disabilities have greater expenditures than those without disabilities. Of this gap, we find that −0.913 (or 48.4% of the gap) is explained by differences in characteristics between the 2 groups. The negative sign of the explained portion suggests that if people with disabilities had the characteristics of people without disabilities (eg, in terms of age, health status), the total expenditures of people with disability would be lower. When we examine in table 4 the total explained difference by domains such as demographics, socioeconomic status, access to care, and health status, we find that 40.3% of the gap is due to differences in health. Time contributed only 0.2% of the gap, so there are no significant trends associated with having a disability. This result is consistent with the result reached earlier in the regression analysis (see table 3) where the interaction term between time and disability status is not statistically different from zero. The same result was reached with the decompositions in the gaps in OOP spending and burden across disability status and is available from the authors.

Table 4. Decomposition of Total Health Care Expenditures by Disability Status, Medical Expenditure Panel Survey—1996, 1998, 2000, 2002, 2004

NOTE. Based on persons aged between 21 and 61. Decomposition is based on parameter estimates derived from 3 regressions (nondisabled, disabled, pooled regression without controlling for disability status). The model is adjusted for sex, race/ethnicity, age, marital status, region of residence, education, employment, poverty status, physical illness, mental illness, and perceived physical and mental health status.

Alternative Specifications 

We specified several alternative models to ensure the robustness of our findings with regard to disability and health care costs. These are summarized below.

Second definition of disability 

We repeated the analysis above using the second disability definition based on functional limitations. Disability prevalence based on functional limitations is higher compared with that based on activity limitation, but shows a similar increasing trend during the 1996 to 2004 period, from 17.2% to 19.4% (P<.01). Findings from the analysis using the second definition of disability did not substantially differ from those with our primary definition. (Results are available from the authors.)

Two-part model 

We found that nearly 20% of adults in the age group 21 to 61 years had zero total health care expenditures. Therefore, we also estimated a 2-part model for total expenditures. In the first part, the probability of nonzero health expenditures is estimated, while in the second part, the amount of health care expenditures is estimated for all persons with positive expenditures. This approach has been used in many studies (eg, Steinmann et al³⁰) and has been found to be superior to those that simultaneously estimate use and expenditures through Heckman-type selection models.³¹

In the first part, findings from the logistic regression to predict the likelihood among working-age adults to have positive total health care expenditures revealed that those persons with disabilities were nearly 3 times more likely than those without disabilities to incur health care expenditures during the study period. The adjusted odds ratio was 2.75 (95% confidence interval, 2.00–3.77). In the second part of the model, among working-age adults with positive health care expenditures, persons with disabilities had greater expenditures. The OLS parameter estimate for disability status was .73 (P=.000). Again, after controlling for demographic and socioeconomic characteristics, the interaction term between disability and time did not reveal any significant difference in the trend of total expenditures across disability status. (Results of the 2-part model are available from the authors.)

Back to Article Outline

Discussion 

Across all years, persons with disabilities had greater total health expenditures, OOP spending, and burden compared with adults without disabilities, suggesting there are additional health care costs associated with disability. Higher OOP spending associated with disability has implications for the measurement and estimation of the prevalence of poverty among persons with disabilities. As shown by Fujiura et al,³² the percentage of households living below the poverty line is significantly higher when the household has a person with a disability (28%) compared with a household without a person with a disability (8.3%). These estimates of poverty prevalence among households with and without a person with a disability were estimated by applying the standard poverty threshold, assuming that the minimum standard of resources encapsulated in the poverty threshold is sufficient to meet the needs of persons with disabilities. The estimate of poverty among households with members with a disability would be even higher if the poverty line was adjusted to reflect the extra health costs of living with a disability. In fact, She and Livermore³³ found that the conventional income-based measures of poverty do not adequately measure poverty among those with disabilities who require additional resources just to meet basic material needs, particularly medical needs and food security.

Although further research is needed to estimate the extent to which OOP spending leads to poverty among persons with disabilities and to assess the adequacy of disability benefits such as SSDI and SSI, and of mainstream benefits such as Temporary Assistance for Needy Families, one could speculate that the benefit amounts should be higher for persons with disabilities to reach a standard of living similar to that of persons without disabilities. In addition, researchers and policymakers need to assess whether the Health Savings Accounts created in 2003 to enable people to pay for current health expenses and save for future expenses on a tax-free basis have helped persons with disabilities cover their extra health care costs. Another policy that might alleviate the burden of extra costs of living with disability is to provide allowances to persons with disabilities as part of standalone programs that compensate for disability-related expenditures. Such programs are available in other developed countries. For example, in Sweden, special allowances are provided for a wide range of disability-related costs, including durable medical equipment and attendant care.³⁴ In Great Britain, the so-called disability living allowance compensates for the extra costs incurred due to the effects of a disability.³⁵ The allowance has 2 components, a care component and a mobility component, and is provided on a temporary or a permanent basis, irrespective of the work status of the person.

Our hypotheses regarding disability-specific trends in expenditures, OOP spending, and burden were not confirmed by the analysis. Between 1996 and 2004, we found significant increases in total health care expenditures, OOP spending, and burden for all working-age people. However, there was not any evidence of a disproportionate increase in total expenditures among persons with disabilities during the study period. This finding suggests that there might not have been an improvement in access to health care services, as expected after the ADA of 1990 and the Olmstead decision of 1999, which would have resulted in higher total health expenditures. Of course, this article does not provide any direct evidence on the effects of the ADA and the Olmstead decision, but certainly points toward the need for more research in this area.

In addition, we could not find evidence of a smaller increase in OOP spending and burden for persons with disabilities, which may point toward the inability of recent policy attempts to facilitate access to health insurance for persons with disabilities. Programs such as the Medicaid Buy-In programs do not appear to have curbed the increase in OOP spending for persons with disabilities relative to persons without disabilities. Again, more research is needed to assess the specific impact of such policy initiatives on OOP spending and generally on access to health care among persons with disabilities.

Study Limitations 

This study has several limitations. This study does not cover the institutionalized population with disabilities and is therefore not representative of the entire working-age population with disabilities. This study is focused on health-related expenditures, and it therefore does not capture other potential additional expenditures associated with disability such as transportation, which has been shown to be significant based on data for other countries.³⁶ The MEPS data set did not capture the complete expenditure for personal attendant care, care that has both a formal and an informal component, and can represent a significant portion of community-based expenditures for persons with disabilities.³⁷, ³⁸ The primary level of analysis for this article is the person. Further research on expenditures across disability status is required at the household level and at the insurer level, whether public or private. Furthermore, our study did not include insurance premiums, which changed during the study period in Medicare and in private health insurance plans, and might have affected persons with and without disabilities differentially.

Back to Article Outline

Conclusions 

Despite these limitations, this study fills an important knowledge gap. It uses a nationally representative data set of the noninstitutionalized population to analyze levels and recent trends in health care expenditures, OOP spending, and OOP burden across disability status. The study finds that substantial direct health care expenditures, OOP spending, and burden are associated with disability. Although expenditures, OOP spending, and burden increased over time, after controlling for demographic, socioeconomic, and health status, these 3 health care costs were not found to change disproportionately for people with disabilities. Our findings suggest that insurance coverage expansions as they have been put in place during the last decade or so may alone not be enough to reduce OOP spending and burden among persons with disabilities. Further research is needed on expenditures, OOP spending and burden associated with a disability at the household level and on the effectiveness of specific policy initiatives in reducing OOP spending and burden on persons with disabilities.

Supplier

Back to Article Outline

References 

1. In:  Field M,  Jette AM editor. Committee on Disability in America (The future of disability in America). Washington, DC: Institute of Medicine of the National Academies; 2007;
   • View In Article
2. Rehabilitation Research and Training Center on Disability Demographics and Statistics (2006 disability status report). Ithaca: Cornell Univ; 2007;
   • View In Article
3. Bhattacharya J, Choudhry K, Lakdawalla D. Chronic disease and severe disability among working-age populations. Med Care. 2008;46:92–100
   • View In Article
   • CrossRef
4. Banthin JS, Bernard DM. Changes in financial burdens for health care: national estimates for the population younger than 65 years, 1996 to 2003. JAMA. 2006;296:2712–2719
   • View In Article
   • CrossRef
5. Waters HR, Anderson GF, Mays J. Measuring financial protection in health in the United States. Health Policy. 2004;69:339–349
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (100 KB)
   • CrossRef
6. Hwang W, Weller W, Ireys H, Anderson G. Out-of-pocket medical spending for care of chronic conditions. Health Aff. 2001;20:267–278
   • View In Article
7. Yelin E, Murphy L, Cisternas MG, Foreman AJ, Pasta DJ, Helmick CG. Medical care expenditures and earnings losses among persons with arthritis and other rheumatic conditions in 2003, and comparisons with 1997. Arthritis Rheum. 2007;56:1397–1407
   • View In Article
   • MEDLINE
   • CrossRef
8. Zaidi A, Burchardt T. Comparing incomes when needs differ: equivalization for the extra costs of disability in the U.K.. Rev Income Wealth. 2005;51:89–114
   • View In Article
9. Yelin E, Cisternas M, Trupin L. The economic impact of disability in the United States. 1997; total and incremental estimates. J Disabil Policy Stud. 2006;17:137–147
   • View In Article
10. Chan L, Beaver S, Maclehose RF, Jha A, Maciejewski M, Doctor JN. Disability and health care costs in the Medicare population. Arch Phys Med Rehabil. 2002;83:1196–1201
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (110 KB)
    • CrossRef
11. Altman BM, Cooper PF, Cunningham PJ. The case of disability in the family: impact on health care utilization and expenditures for nondisabled members. Milbank Q. 1999;77:39–75
    • View In Article
    • MEDLINE
12. Bhattacharya J, Cutler DM, Goldman DP, et al. Disability forecasts and future Medicare costs. Front Health Policy Res. 2004;7:1052
    • View In Article
13. Hill SC, Livermore GA, Houtenville AJ. Rising health care expenditures and the employment of people with high-cost chronic conditions. In:  Stapleton DC,  Burkhauser RV editor. The decline in employment of people with disabilities: a policy puzzle. Kalamazoo: UpJohn Institute for Employment Research; 2003;
    • View In Article
14. Goodman NJ, Stapleton DC, Livermore GA, O'Day B. The health care financing maze for working-age people with disabilities. Ithaca: Cornell University Institute for Policy Research–Rehabilitation Research and Training Center for Economic Research on Employment Policy for Person with Disabilities; 2007;
    • View In Article
15. Iezzoni LI, Killeen MB, O'Day BL. Rural residents with disabilities confront substantial barriers to obtaining primary care. Health Serv Res. 2006;41:1258–1275
    • View In Article
    • MEDLINE
16. Iezzoni LI, O'Day BL. More than ramps: a guide to improving health care quality and access for people with disabilities. New York: Oxford Univ Pr; 2006;
    • View In Article
17. Atchinson BK, Fox DM. The politics of the Health Insurance Portability and Accountability Act. Health Aff. 1997;16:146–150
    • View In Article
18. White PH. Access to health care: health insurance considerations for young adults with special health care needs/disabilities. Pediatrics. 2002;110:1328–1335
    • View In Article
19. Social Security Administration. Trends in the Social Security and Supplemental Security Income Disability programs. Baltimore: SSA; 2006;
    • View In Article
20. Minden K, Niewerth M, Listing J, Biedermann T, Schontube M, Zink A. Burden and cost of illness in patients with juvenile idiopathic arthritis. Ann Rheum Dis. 2004;63:836–842
    • View In Article
    • MEDLINE
    • CrossRef
21. Altman BM. The handbook of disability. Thousand Oaks: Sage; 2001;
    • View In Article
22. Mitra S. The capability approach and disability. J Disabil Policy Stud. 2006;16:236–247
    • View In Article
23. World Health Organization. International classification of functioning, disability and health. Geneva: World Health Organization; 2001;
    • View In Article
24. Kruse D, Schur L. Does the definition affect the outcome?. In:  Stapleton DC,  Burkhauser RV editor. The decline in employment of people with disabilities. Kalamazoo: UpJohn Institute for Employment Research; 2003;p. 279–300
    • View In Article
25. Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY. Methods for analyzing health care utilization and costs. Annu Rev Public Health. 1999;20:125–144
    • View In Article
    • MEDLINE
    • CrossRef
26. Manning WG. The logged dependent variable, heteroscedasticity, and the retransformation problem. J Health Econ. 1998;17:283–295
    • View In Article
    • MEDLINE
    • CrossRef
27. Cohen J. Methodology report #1: design and methods of the Medical Expenditure Panel Survey household component. www.meps.ahrq.gov/data_files/publications/mr1/mr1.shtmlAccessed January 16, 2009
    • View In Article
28. Cohen SB, DiGaetano R, Goksel H. Methodology report #5: estimation procedures in the 1996 Medical Expenditure Panel Survey household component. www.meps.ahrq.gov/data_files/publications/mr5/mr5.shtmlAccessed January 16, 2009
    • View In Article
29. Oaxaca R, Ransom M. On discrimination and the decomposition of wage differentials. J Econ. 1994;61:5–21
    • View In Article
30. Steinmann L, Telser H, Zweifel PS. Aging and future healthcare expenditure: a consistent approach. Front Health Policy Res. 2007;10:1041
    • View In Article
31. Dow WH, Norton EC. Choosing between and interpreting the Heckit and two-part models for corner solutions. Health Serv Outcomes Res Methodol. 2003;4:5–18
    • View In Article
32. Fujiura GT, Yamaki K, Czechowicz S. Disability among ethnic and racial minorities in the United States. J Disabil Policy Stud. 1998;9:111–130
    • View In Article
33. She P, Livermore GA. Material hardship, poverty, and disability among working-age adults. Soc Sci Q. 2007;88:970–989
    • View In Article
34. Zeitzer I. The challenges of disability pension policy: three western European case studies of the battle against the numbers. In:  Fultz E,  Ruck M editor. Reforming worker protections: disability pensions in transformation. Budapest: International Labour Organization; 2002;p. 227–286
    • View In Article
35. Mitra S. Temporary and partial disability programs in nine countries: what can the United States learn from other countries?. J Disabil Policy Stud. 2009;20:14–27
    • View In Article
36. Jones A, O'Donnell O. Equivalence scales and the costs of disability. J Public Econ. 1995;56:273–289
    • View In Article
37. LaPlante MP, Kaye HS, Harrington C. Estimating the expense of a mandatory home- and community-based personal assistance services benefit under Medicaid. J Aging Soc Policy. 2007;19:47–64
    • View In Article
    • CrossRef
38. Grossman BR, Kitchener M, Mullan JT, Harrington C. Paid personal assistance services: an exploratory study of working-age consumers' perspectives. J Aging Soc Policy. 2007;19:27–45
    • View In Article
    • CrossRef

• a SAS software; SAS Institute, 100 SAS Campus Dr, Cary, NC 27513-2414.