Health Care Expenditure Among People With Disabilities

Abstract

Workplace health-promotion programs have the potential to reduce health care expenditures, especially among people with disabilities. Utilizing nationally representative survey data, the authors provide estimates for health care expenditures related to secondary conditions, obesity, and health behaviors among working-age people with disabilities. Furthermore, by computing the expenditures attributable to secondary conditions, obesity, and health behaviors, the authors emphasize the importance of disability-inclusive workplace health-promotion programs for employees with disabilities. Overall, the authors observed that the annual average health care expenditure among employed people with disabilities was US$4,524 (95% confidence interval [CI] = US$4,248–US$4,800) compared with US$1,325 (95% CI = US$1,299–US$1,351) for employed people without disabilities. Furthermore, obesity accounted for 27% to 41% of excess expenditures for people with various disability classifications compared with their nonobese peers with disabilities. Secondary conditions accounted for about 20% to 25% of higher health care expenditures among working people with various disability classifications, compared with their peers with disabilities who do not have secondary conditions. In addition, lack of exercise and alcohol consumption accounted for one fourth to over one third of excess health care expenditures among employed people with disabilities. The authors discuss implications of these findings for rehabilitation counselors and public health practitioners.

Keywords

workplace health promotion people with disabilities health care expenditures

Introduction

The American workforce is aging and a greater proportion of these seasoned workers are choosing to work. The changing landscape of retirement benefits from defined benefits to defined contribution plans (United States Bureau of Labor Statistics, 2011), the increase in Social Security’s full-retirement age, and increasing costs of health care (Johnson, 2007) have all contributed to this labor market trend. It is projected that by the year 2016, the number of workers in the age group of 55 and above will rise by 47%, which is nearly 5 times the projected growth in the overall U.S. workforce (Toosi, 2007). Beginning in 2007 and throughout the period of the Great Recession, more employers continued to employ older workers (Hill, 2011). From these data, it appears that employers continue to favor the experience mature workers bring to their respective roles. However, the higher likelihood of the onset of disabilities and chronic health conditions among older workers may lead to an increase in health care expenditures, ultimately causing concerns to employers (Garrett & Martini, 2007).

In general, overall health care costs have increased substantially in recent decades in the United States. Despite spending up to 16% of its gross domestic product (GDP) in providing health care, the United States lags behind in most benchmarks of health and well-being, in comparison with other industrialized countries (Banthin & Bernard, 2006). As employers pay about a third of total annual medical expenditures, these national trends substantially increase their costs of providing employee health benefits. Some employers, sometimes without specific information about causes of rising health care costs, have adopted discriminatory practices in hiring and retaining workers with a disability and older workers, thereby making them potentially susceptible to discrimination charges filed under the Americans with Disabilities Act (ADA; Bjelland et al., 2010). More precise information about the factors contributing to the largest proportional share of rising employer health care costs is needed for employers to make informed judgments about how to reduce costs, without needlessly discriminating against these protected populations.

Our increasingly unhealthy lifestyle as a country and the related rising epidemic of obesity have contributed significantly to the soaring health care costs (Thorpe & Howard, 2006). In the workplace contexts, Wright and colleagues (2004) demonstrated that the higher risk behavior accounts for 15% to 30% of health care expenditures among employees. Furthermore, Musich, Napier, and Edington (2001) indicated that such behaviors were attributable to almost 85% of worker’s compensation costs. Studies have noted up to a 228% increase in the cost of health care to employers for employees with high risk factors, including tobacco use, being overweight or obese, and so on (Goetzel et al., 1998). Furthermore, Anderson et al. (2000) observed that employee risk factors such as obesity, depression, smoking, alcohol consumption, and lack of physical activities contributed to almost one fourth of employer health care expenditures for their employees. By addressing these factors, employers also have a potential to leverage an untapped human resource—that is, a growing population of aging workers and people with disabilities (Morton, Foster, & Sedlar, 2005).

To cope with the higher costs of providing health care benefits, increasingly employers are shifting the costs of care to the employee. This strategy appears to address only the supply side of the excess health care costs equation (e.g., limiting the supply of health care services through high deductible plans). Furthermore, this strategy has been counterproductive to employers as employees who pay a higher share of their health care benefits forego necessary care, thereby affecting their workplace productivity (Vogenberg, Holland, & Liebeskind, 2007). Therefore, systematically addressing the demand side of excess health care costs (e.g., reducing factors leading to higher health care needs through prevention, risk reduction, and disease management) holds promise in the long term for employers in reducing the health care expenditures of their workers. Workplace health-promotion programs are designed to address these “modifiable risk factors” in the work environment and improve well-being among workers, especially for those with existing chronic health conditions and disabilities (Goetzel & Ozminkowski, 2008). Workplace health-promotion programs are employer-sponsored initiatives directed toward improving the health and well-being of employees for ensuring higher productivity and reducing the employers’ costs of health insurance premium costs by shrinking the high risk pools (Aldana, Merrill, Price, Hardy, & Hager, 2005). These programs range from the prevention and reduction of risk factors causing chronic diseases (e.g., weight and stress management programs, smoking cessation programs) to programs aimed at managing specific health conditions (e.g., asthma, depression, cardiovascular and high cholesterol-related diseases; Goetzel & Ozminkowski, 2008).

Workplace Health-Promotion Programs in the United States

In a 2011 survey of employers’ health benefits conducted by Kaiser Family Foundation and Health Research & Education Trust (Kaiser/HRET), 65% of employers offered some form of workplace wellness programs to their employees (Kaiser/HRET, 2011). These programs ranged from weight loss programs, gym membership discounts, smoking cessation programs, personal health coaching, classes in nutrition, web-based resources for healthy living to wellness newsletters. Larger firms (>200 workers) were more likely to provide an assortment of such programs, compared with smaller firms. The Kaiser/HRET survey further reported that nearly 14% of employers incentivized participation in wellness programs by offering gift cards and travel vouchers. Large-size firms were twice as likely to provide such incentives compared with small-size firms. A very small percentage of firms provided benefits such as contributions to health plan premiums, lowering deductibles, or health saving accounts.

Despite this high prevalence of offerings in wellness programs, employee participation continues to be low, costing the program impact on its intended health outcomes. One possible explanation for lower participation of employees in such programs could be access to these wellness programs. Nearly 87% of workplace wellness programs are provided through the employee health plans—a trend more common among small firms. Thus, employees without employer-paid health plans would not have access to such programs, leading to their low participation rates. This is a significant barrier for employed people with disabilities, as recent studies indicate that a greater proportion of employed people with disabilities either choose not to participate due to higher deductibles or are not offered employer-based health insurance (Crossley, 2005; Furrow, 2007).

In addition, it is likely that many employers offer only those wellness activities that come packaged with health plans, without much individualization of programs based on their employee needs or the work context. The findings of the National Worksite Health Promotion Survey support this assertion (Linnan et al., 2008), in which they found that only about 7% of employers in the United States provide comprehensive wellness programs. This could be one of the potential reasons for lack of participation in such programs, contributing toward their overall modest impact on reducing health care expenditures for employees observed in several demonstration studies.

Despite substantial variations in their structures and implementation, workplace health-promotion programs have been shown to create substantial savings to employers (Kuoppala, Lamminpaa, & Husman, 2008). Baicker, Cutler, and Song (2010), in their econometric analysis of workplace health care programs conducted by examining published estimates in the literature, estimated an average reduction for participating employees of up to US$360 per employee annually toward health care expenditures. Although such studies continue to demonstrate positive returns on investment to employers, research has yet to examine the impact of such programs on health care expenses of employees with a disability (Call, Gerdes, & Robinson, 2009).

In reviewing corporate wellness programs, Call et al. (2009) note a dearth of models that are inclusive of people with disabilities and indicate that effective workplace wellness programs have yet to be examined from the perspective of universal accessibility. As many workplace wellness programs are not accessible for people with disabilities (Rimmer & Rowland, 2008), employers may not be able to realize the benefits of implementing such programs due to lack of participation of employees who are situated in the highest bracket of health care expenses (Gulley, Rasch, & Chan, 2011). Furthermore, no research to date has documented the excess costs of health care attributable to high-risk behaviors and secondary health conditions including obesity among working people with disabilities. Such data are necessary to benchmark intervention impact studies aimed to improve health and wellness of workers, especially among those with a disability or a chronic health condition. In addition, such analyses can assist the disability community and disability advocates in better making the case to employers and policy makers for equitable access to workplace wellness programs (Call et al., 2009).

The research described in this article addresses this dearth of documentation to date and aims to inform about the potential of workplace health-promotion interventions by providing estimates of the impact of risk behaviors, workplace activity, and secondary health conditions on health care expenditures for working people with disabilities, using a nationally representative survey data set. This article specifically aims to examine the following research questions:

Research Question 1: What is the difference in estimated annual average health care expenditure for working people with disability and people without disabilities?

Research Question 2: To what extent do indicators of health and health behaviors (e.g., obesity, secondary conditions, work and leisure-time physical activities, smoking, alcohol use), sociodemographic factors (e.g., poverty, race/ethnicity, gender), and access/utilization of health care explain any observed differences in estimated annual average health care expenditures between working people with and without disabilities?

Research Question 3: To what extent can the observed differences in health care expenditures be attributed to the indicators of health and health behaviors among employed people with disabilities?

Method

Data Set

We utilized the National Health Interview Survey (NHIS) linked to the Medical Expenditure Panel Survey (MEPS) data set for the years 2004 through 2008 for the current analysis. The MEPS, conducted annually by the Agency for Health Care Quality, is a longitudinal panel survey providing nationally representative estimates of health care utilization/expenditures, sources of payments, and health insurance for the noninstitutionalized U.S. civilian population. We utilized the Household Component (MEPS-HC) of the MEPS survey data set. The MEPS-HC sample is a complex national probability sample of households that participated in the previous year’s NHIS conducted by the National Center on Health Statistics. Each year, the MEPS-HC sample is established to form a panel and everyone within the panel is interviewed 5 times over a period of 30 months. The MEPS-HC collects information on key sociodemographic factors, medical conditions/events, functional limitations, costs or expenditure for medical services including physician services, drugs, procedures, hospitalization and ambulatory care, as well as information on employment, hourly wages, access to health insurance, and workplace characteristics in every interview round. This establishes longitudinal data for 2 years on an individual basis.

The NHIS, initiated in 1957, is the prime resource providing national-level estimates on health care access, health care utilization, existing health conditions, health-related behaviors, and relevant sociodemographic information for the noninstitutionalized U.S. population. Each year since 1996, two of the NHIS secondary sample units or “panels” are set aside for the MEPS for follow-up data collection on related elements. The target sampling size for MEPS-HC is about 37,000 individuals, with oversampling for specific group quarters not targeted in the NHIS—that is, families with incomes 200% below the Federal Poverty Level (FPL). The most recent technical report documented a 65.6% overall response rate for the MEPS-HC sample that participated in the previous year’s NHIS study (Ezzati-Rice, Rhode, & Greenblatt, 2008).

Linking the NHIS sample to the MEPS not only provides an additional set of observations but also contributes important variables that are not tracked within the MEPS-HC survey (e.g., health behaviors such as exercise, smoking, alcohol consumption).

File Linkage Process

We linked the NHIS files for the years 2004 and 2005 to MEPS longitudinal data Panels 10 and 11 established in 2005 and 2006, respectively (i.e., MEPS file HC 106 and HC 114). The individuals in the MEPS Panels 10 and 11 were followed for a 2-year period. Adding an additional year of NHIS data linked to the MEPS data sets provided follow-up on individual-level data for a 3-year period (i.e., 2004–2007 for Panel 10, and 2005 through 2008 for Panel 11 MEPS data). Specific linkage files consisting of unique identification numbers for the MEPS and NHIS obtained from the Agency for Healthcare Research and Quality (AHRQ) research data center enabled the file linkage process for establishing the merged data set for analysis. Furthermore, we restricted the data analysis to individuals aged 18 to 65 years in the merged data sets to capture the working-age population. The final overall unweighted sample size for the combined NHIS/MEPS matched Panel 10 and 11 data sets consisted of 9,196 individuals providing a large-enough sample size for specific subgroup analysis. Longitudinal MEPS-HC weights were applied for estimating the variance for parameters accounting for the complex sample survey design.

Dependent Variable

The average annual health care expenditure was the dependent variable for this study, and it is the amount of direct payments made in a given year, including out-of-pocket payments and payments made by insurance (public and private) for services received for health conditions. This amount does not include payments for over-the-counter medications, alternative services and any charges resulting from phone contacts with physician offices.

The self-reported data on health care expenditures were validated by the Medical Provider Component (MPC) of the MEPS survey where a select sample of medical providers of the interviewed participants were contacted to collect accurate information on payments for services. The data emerging from the MC was used to statistically impute and adjust for any errors in self-reporting of medical expenditures for specific services and any missing information. Furthermore, the total health care expenditures for both panels are expressed in 2007 dollars using the Personal Health Care Price Index (Centers for Medicare and Medicaid Services, 2010).

Independent Variables

Disability classifications

We utilized the conceptual framework of the International Classification of Functioning, Disability and Health (ICF) developed by the World Health Organization (WHO) in developing disability classifications within this research (WHO, 2001). The ICF framework for disability recognizes disability as a multidimensional construct along the continuum of health and well-being (Kostanjsek, 2011). This framework recognizes disability as a function of interaction between individual health conditions and the environment. In brief, the ICF construct for disability includes the interaction between health conditions, impairment of body structures and functions, activities and participation restrictions, and personal and environmental factors.

Pollard, Johnston, and Dieppe (2011) distilled the ICF constructs for disabilities into three major categories of impairment in body structures and function, activity limitation, and participation restriction. The impairment of body structures and functions include specific chronic disease conditions—for example, multiple sclerosis affecting neuromuscular system. In MEPS-HC, the information on specific health conditions is coded using the International Classifications of Diseases–9 (ICD-9) criteria by professional coding experts based on the descriptions of the conditions provided by the participants. We converted these ICD-9 diagnostic codes into specific body systems impairments using the Clinical Classification Software (CCS) developed by AHRQ (Elixhauser, Steiner, & Palmer, 2011). Only chronic conditions were considered for inclusion in identifying impairments in body structures and functions.

The activity limitations construct includes functional limitations in the activities of daily living and instrumental activities of daily living measures, as well as other tasks such as bending, standing, walking, and grasping. Specifically, to construct the activity limitations domain, we included only those limitations lasting for more than 3 months in each of these areas in the MEPS-HC. The participation restriction includes limitations in going to work, school, and participation in social and leisure-time activities. We utilized specific question items from the MEPS-HC to construct the participation restriction domain. These constructs and their combinations informed development of disability classifications for this analysis. It must be noted that in our analysis, individuals with only impairment of body structures and function with no activity limitation and no participation restriction are not classified as having a disability. We chose to define disability in this way as many health conditions are not disabling conditions in their contextual settings.

Sociodemographic factors

Sociodemographic factors include variables such as gender, race/ethnicity, age, education levels, and family income. The family income is categorized in MEPS-HC based on FPL definitions into poor/near poor (<125% of FPL), low income (125%–200% of FPL), middle income (200%–400% of FPL), and high income (>400% of FPL).

Access to health care

The variables include coverage by health insurance (public or private during an interview round) and having access to a usual source of care¹ for preventive and curative services.

Employment and work-related physical activities

Each individual in NHIS-MEPS-HC was asked about his or her employment status, hourly wages earned, jobs changes, and type of work. Furthermore, to characterize potential physical activity involved in these occupations, summary-level metabolic equivalent values (MET values) developed by Tudor-Locke, Ainsworth, Washington, and Troiano (2009) and Tudor-Locke, Washington, Ainsworth, and Troiano (2011) were assigned to the occupational categories. The MET values essentially provide information on the excess of energy spent doing specific activities at work in comparison with the normal energy spent at rest. For individuals without employment, the MET values were retained as “1.”

Indicators of health and health behaviors

These include percentage body fat (PBF), secondary health conditions, and health behaviors such as smoking and alcohol consumption. We use PBF instead of the body mass index (BMI), a weight-to-height ratio, because the BMI underestimates obesity among people with disabilities (Burkhauser & Cawley, 2008; Rimmer, Wang, Yamaki, & Davis, 2010). The PBF, on the other hand, has been identified as a better indicator of adiposity in the body. Jackson et al. (2002) provided a formula to estimate adult PBF using BMI as follows:

Percentage body fat = (1.39 \times BMI) + (0.16 \times Age) - (10.34 \times gender) - 9

where gender is “0” for females and “1” for males. Furthermore, as suggested by Jackson and colleagues (2002), men with PBF more than 25% and women with PBF more than 33% were classified as being obese.

The secondary health conditions were identified as an incident report of health conditions among the participants across each round of the survey. Guided by methodology used by Rasch and colleagues (Rasch, Magder, Hochberg, Magaziner, & Altman, 2008), these conditions were further identified by examining the corresponding CCS codes to ensure that it is a new or an incident health condition across all body systems and functions to closely align with the definition of secondary conditions provided by the Institute of Medicine (IOM, 2007). Secondary conditions included all physical and mental health conditions as ascertained by the CCS codes.

Statistical Analysis

We conducted univariate descriptive analyses to study the characteristics of the population of people with disabilities. We additionally examined differences in total health care expenditure between people with disabilities compared with those with no disabilities in the study data set. We utilized the MEPS longitudinal weights for study samples for the purpose of reporting standard errors and estimates for our univariate analyses. We developed multiple regression analyses for studying the relationship between indicators of health and health behaviors and total health care expenditure among employed people with disabilities in comparison with their employed peers without disabilities, controlling simultaneously for other independent variables. In addition, as the information was longitudinally collected on the individual participants, we accounted for correlations within the outcome variable by using the generalized estimating equations (GEE) as described by Liang and Zeger (1986).

In brief, the GEE is an extension of the generalized linear models that accounts for correlations within individual response variables over time and calculates robust standard error estimates for the regression parameters for consistent statistical inference (Burton, Gurrin, & Sly, 1998). The correlation within subjects is treated as a nuisance variable. First, within GEE, a simple linear regression model is fit, assuming independence between discrete observations. Then a working correlation matrix is estimated using the residuals of the simple linear regression equation; the regression equation is then refit taking into account this correlation structure. These steps are followed iteratively to generate balanced parameters estimates.

The GEE method offers advantages in the case of missing data, as well as misspecification of the working correlation matrix. Furthermore, GEE allows the researcher to account for any correlations resulting from clustering due to the multistage sampling, as in the case of the MEPS-HC survey, in estimating the regression parameters. The GEE models provide marginal distribution effects, that is, group-level information for parameters estimated. In the case of this analysis, an exchangeable working correlation matrix was employed for GEE; this means that every observation within the subject was assumed to be equally correlated with their next observation. Estimation algorithms converged well with the assumption of this correlation structure.

It is also important to note that the outcome variable—average annual health care expenditure—was not normally distributed. As a result, a log transformation was applied to achieve normality of the outcome variable distribution.² This transformation affects the interpretation of regression parameters, where a regression coefficient is interpreted as a percentage increase or decrease in total health care expenditure, rather than using the absolute values of the parameters estimated.

We constructed the following model³:

\begin{array}{l} \log (Health Care Expenditures) = β_{0} + β_{1} (age group) + β_{2} (race) + β_{3} (gender) \\ + β_{4} (family income category) + β_{5} (disability type) + β_{6} (MET value) \\ + β_{7} (public insurance) + β_{8} (private insurance) + β_{9} (obese) + β_{10} (secondary conditions) + β_{11} (smoking) \\ + β_{12} (alcohol use) + β_{13} (exercise) + β_{14} (diagnosis of CVS) + β_{15} (diagnosis of RS) \end{array}

Furthermore, using the regression coefficients, we calculated the percentage excess expenditures attributable to indicators of health and health behaviors among specific subgroups for people with disabilities using the following equation:

Percent Excess Expenditures = P_{i} \times [(β_{5 i} - β_{5}) / β_{5 i}]

where, P_i is the prevalence of the indicator of health and health behaviors in a specific disability subgroup (e.g., obesity), β_5i is the percentage excess health care expenditures in the given subgroup having indicator of health and health behavior (e.g., percentage of excess health care expenditure among obese people with activity limitation), and β₅ is the percentage excess health care expenditures in the given subgroup who does not have the indicator of health and health behavior (e.g., percentage excess health care expenditure among nonobese people with activity limitations). We used this approach as each regression coefficient represented a proportion increase or decrease in costs, simulating the relative risk construct in epidemiological literature (Benichou, 2001). This approach computing excess expenditures attributable to specific indicators of health and health behaviors was adopted to demonstrate the potential for reduction in health care expenditures for specific subgroups of people with disabilities in the absence of such indicators.

Results

Study Population

About 14% of the overall population was identified as having a disability based on the ICF criteria of impairment, activity limitation, and participation restriction used in this research. The weighted population estimate indicates about 64 million people had a disability in the samples established in 2004 and 2005. In addition, about 2.6% had only activity limitation, 1.8% had only participation restriction, 1.3% had activity limitations and impairment, 1.2% had participation restriction and impairment, 3% had activity limitation and participation restriction, and almost 5% had all three constructs of disabilities (Figure 1).

Figure 1.

Percentage distribution of people with disabilities using ICF-based classification: NHIS/MEPS linked files for years 2004–2007

Table 1 illustrates the distribution of key demographic, health care utilization, health status, and health behavior variables across people with disabilities and those without disabilities. It can be observed that a higher proportion of people with disabilities were African Americans and lower proportion were Hispanics compared with their peers without disabilities (18% vs. 13%; 8% vs. 13%, respectively). In addition, more than one in five people with disabilities (23%) were less than high school educated, compared with 12% of people without disabilities. Similarly, more than one third of people with disabilities (36%) report a family income classified as “poor/near poor,” compared with 15% of people without disabilities. Almost twice the number of people without disabilities had their family income in the “high income” category, as compared with those with disabilities. About 41% of people with disabilities were employed, compared with 84% employment among people without disabilities. Interestingly, a greater proportion of people with disabilities (83%) reported being covered by some form of health insurance and having access to a regular source of care (87%) from a provider compared with their peers without disabilities. In addition, 19% of people with disabilities visited the emergency room (ER) one time for services, compared with 12% of people without disabilities over the period of data collection. Twelve percent of people with disabilities visited the ER 2 to 3 times, whereas only 4% of people without disabilities reported visiting ER 2 to 3 times over the same period of data collection (i.e., between 2005 and 2007).

Table 1.

Percentage Distribution of Key Demographic, Health Care Utilization, Health Status, and Health Behavior Variables: Comparison Between People With and Without Disabilities

Key variables	People with disabilities (%)	People without disabilities (%)	p value for χ²
Race/ethnicity
White	71	69	<.001
African American	18	13
Hispanic	8	13
Alaskan Native/American Indian	1	0.5
Asian	1	3
Multiple race	0.1	0.1
Others	1	2
Education level
<High school education	23	12	<.001
High school or GED	32	24
Some postsecondary education	45	64
Employment	41	84	<.001
Income level
Poor/near poor	36	15	<.001
Low	17	12
Middle	24	30
High	24	43
Receiving SSI due to disability	22	1	<.001
Has some form of health insurance	83	79	<.01
Covered by public insurance	46	12
Covered by private insurance	52	74
Has usual source of care	87	74	<0.001
Emergency room visits
0	62	82
1	19	12
2–3	12	4
4–5	4	1
>5	3	0.4
Smokes currently or in the past	58	40	<0.001
Consumes alcohol currently or in the past	48	60	<0.001
Exercise level
Some	47	67	<0.001
Moderate physical activity	41	57
Vigorous physical activities	22	48
Strength-building activities	17	29
No physical activities	53	33
Obesity	70	48	<0.001
Has secondary conditions	49	35	<0.001

Note: National Health Interview Survey–Medical Expenditure Panel Survey linked files 2004-2007. SSI = Supplemental Security Income from Social Security Administration

Fifty-eight percent of people with disabilities currently smoke or formerly smoked tobacco, compared with 40% of people without disabilities. More than one half of people with disabilities do not participate in physical activities, compared with 33% of people without disabilities. Similarly, using the PBF criteria to define obesity, 70% of people with disabilities were classified as obese, compared with 48% of people without a disability.⁴ Less than 50% of people with disabilities reported secondary health conditions. People with disabilities consumed less alcohol; 48% of people with disabilities currently consumed alcohol, compared with 60% of people without disabilities.

Health Care Expenditure

The annual average health care expenditures for people with disabilities amounted to US$5,702 (95% CI = US$5,531–US$5,873) and among people without disabilities, the average health care expenditure was US$1,446 (95% CI = US$1,419–US$1,471). Figure 2 represents the annual average health care expenditures across various disability classifications, as well as for people without disabilities. In general, an increase in the complexity of disabilities appears to be related to an increase in the average health care expenditures, ranging from US$3,277 (95% CI = US$3,054–US$3,501) for people with only activity limitations to US$7,224 (95% CI = US$6,920–US$7,528) for people with all three disabling constructs.

Figure 2.

Average annual health care expenditure in U.S. Dollars by ICF-based disability classification (adjusted to 2007 dollar amount using the PHE index): NHIS/MEPS linked files 2004-2007

The annual average health care expenditure among employed people with disabilities was US$4,524 (95% CI = US$4,248–US$4,800) compared with US$1,325 (95% CI = US$1,299–US$1,351) for employed people without disabilities. Table 2 provides a comparison of average health care costs between employed people with disabilities and their peers without disabilities across key variables. The difference in the annual average health care expenditure was US$3,447 higher for employed women with disabilities compared with their peers without disabilities. In addition, the average health care expenditures were higher among employed people with disabilities belonging to racial minorities (i.e., African Americans, American Indians, Asians), and those with less than a high school degree, compared with their counterparts without disabilities. It is important to note that the estimates of health care expenditures for American Indians and Asians had large standard errors due to low sample size (not reported in the table), potentially contributing to larger estimates. However, the estimates for Whites and other racial minorities had comparable standard errors.

Table 2.

Comparison of Average Health Care Costs Between Employed People With Disabilities and Their Peers Without Disabilities Across Key Variables

Variables	Employed people with disabilities	Employed people without disabilities	p value for t test
Gender
Male	US$3,865	US$1,011	<.001
Female	US$5,092	US$1,645
Race/ethnicity
White	US$4,711	US$1,432	<.001
African American	US$3,928	US$1,129
Hispanic	US$2,031	US$868
Alaskan Native/American Indian	US$13,276	US$1,775
Asian	US$7,604	US$1,140
Multiracial	US$225	US$9,696
Others	US$1,205	US$1,037
Education level
<High school	US$6,515	US$945	<.001
High school or GED	US$3,073	US$1,236
Postsecondary education	US$4,725	US$1,413
Health Insurance
Has some form	US$5,339	US$1,521	<.001
None	US$2,070	US$522
Usual source of care
Yes	US$5,120	US$1,624	<.001
No	US$1,918	US$531
Smoking
Currently or in the past	US$4,734	US$1,363	<.001
Never	US$4,258	US$1,300
Alcohol consumption
Consumes alcohol	US$3,934	US$1,413	<.001
Never	US$5,228	US$1,177
Physical activity
Vigorous	US$3,789	US$1,240	<.001
None	US$4,901	US$1,416
Obesity
Obese	US$5,346	US$1,623	<.001
Nonobese	US$2,889	US$1,070
Secondary conditions
Yes	US$4,577	US$1,691	<.001
No	US$3,189	US$973
Emergency room visits
1	US$5,463	US$1,521	<.001
2–3	US$10,684	US$1,842
3–4	US$10,500	US$2,667
>4	US$7,933	US$3,947
0	US$3,284	US$1,248
CVS diagnosis
Yes	US$6,881	US$2,203	<.001
No	US$2,983	US$1,124
Respiratory diagnosis
Yes	US$5,266	US$1,742	<.001
No	US$4,243	US$1,236

Note: GED = general education department. CVS = cardiovascular system. Adjusted to 2007 dollars using Personal Health Care Price Index; National Health Interview Survey–Medical Expenditure Panel Survey linked files 2004–2007.

Furthermore, employed people with disabilities who were obese had US$5,346 in total health expenditures compared with US$1,623 for obese individuals without disabilities. Similarly, employed people with disabilities who smoked and consumed alcohol had higher total health care expenditures, compared with their peers without disabilities. In addition, employed people with disabilities with secondary health conditions and instances of ER visits had higher annual health care expenditure, compared with their peers without disabilities.

Regression Analysis

Table 3 provides the regression coefficients as well as the percentage change values for easier interpretation of the main effects of the independent variables on the log transformed annual average health care expenditure. As the main focus of analysis was to examine the relationship between health status and risk behaviors among employed people with disabilities and compare them to those without disability, the regression model was constructed only for all employed individuals in the study data set. Several key sociodemographic variables, disability indicator variables, work-related physical exertion and other variables of health status, access to health care insurance, risk behaviors, and secondary conditions were included within the final model.

Table 3.

Regression Coefficients and Percentage Change Values for the GEE-Based Regression Model

Variables	Coefficients (B)	SE	p	% change
Age (years)
18–25	−0.61	0.167	<.001	−61.0
26–45	−0.69	0.133		−69.0
46–55	−0.23	0.136		−23.0
56–65	—	—	—	—
Gender
Female	1.05	0.08	<.0001	105.0
Male	—	—	—	—
Race/ethnicity
African American	−1.27	0.31	<.001	−127.0
Hispanic	0.85	0.3	<.01	85.0
Alaskan Native/American Indian	−0.53	0.23	.02	−53.0
Asian	−0.05	0.51	.9	−5.0
Others	−0.8	0.12	<.001	−80.0
Multiracial	−0.59	0.12	<.001	−59.0
White	—	—	—	—
Income level
Poor/near poor	−0.74	0.15	<.0001	−74.0
Low	−0.83	0.14	<.0001	−83.0
Middle	−0.31	0.1	<.001	−31.0
High	—	—	—	—
Activity/participation/impairment
Activity limitation	0.55	0.13	<.001	55.0
Participation restriction	0.76	0.15	<.001	76.0
Activity limitation and impairment	0.57	0.2	<.001	57.0
Participation restriction and impairment	0.97	0.32	<.01	97.0
Activity limitation and participation restriction	0.83	0.11	<.001	83.0
All domains	0.99	0.19	<.001	99.0
No disability	—	—	—	—
MET values	−0.21	0.06	.01	−21.0
Health insurance
None	−0.41	0.09	<0.0001	−41.0
Some form	—	—	—	—
Public insurance
Not covered	−0.27	0.11	<.01	−27.0
Covered	—	—	—	—
Private insurance
Not covered	−0.37	0.1	<.001	−37.0
Covered	—	—	—	—
Source of care
Has usual	0.58	0.11	<.0001	58.0
Does not have usual	—	—	—	—
Obesity (based on PBF)
Not obese	−0.33	0.09	<.001	−33.0
Obese	—	—	—	—
Secondary conditions
No	−0.42	0.03	<.0001	−42.0
Yes	—	—	—	—
Exercise
Some	−0.18	0.09	.04	−18.0
None	—	—	—	—
Smoking
Never	−0.04	0.08	.7	−4.0
Currently or in the past	—	—	—	—
Alcohol consumption
Never	−0.15	0.08	.08	−15.0
Currently or in past	—	—	—	—
CVS Diagnosis
No	−0.83	0.09	<.0001	−83.0
Yes	—	—	—	—
Respiratory disease diagnosis
No	−0.53	0.1	<.0001	−53.0
Yes	—	—	—	—
Wave of data collection	−0.09	0.02	<.001	−9.0

Note: GEE = generalized estimating equations; MET = metabolic equivalent values; PBF = percentage of body fat. Outcome variable: log transformed average annual health care expenditures. National Health Interview Survey–Medical Expenditure Panel Survey linked files 2004–2007.

Sociodemographic Variables and Health Care Expenditure

Among employed individuals, young people in the age group of 18 to 25 years had 61% lower health care expenditures compared with elderly people in the age group of 56 to 65 years holding all variables constant. Similarly, people in the age groups of 26 to 45 and 46 to 55 years have substantially lower annual average health care expenditures compared with those in elderly age groups of 56 to 65 years. Overall, employed women had a 105% higher average health care expenditure compared with employed men. The average health care expenses for employed racial minorities except Hispanics was significantly lower compared with employed Caucasians (e.g., African Americans had 126 lower health care expenses compared with Caucasians). Hispanic workers had 85% higher health care expenditure compared with their Caucasian peers. Furthermore, employed people with family incomes as “poor/near poor” and “low income” had 74% and 83% lower health care expenditures, respectively, compared with people with “high-income” designations for family income levels.

Employed individuals with activity limitations have 55% higher average health care expenditures compared with their peers without disabilities. Similarly, people with participation restriction had 76% higher average health care expenses compared with those without disabilities. It can be observed that the percentage difference in average expenditures increased with the increasing severity of disabilities, with people having all three constructs of disability reporting up to 97% higher health care expenditures compared with people without disabilities.

Health Status, Health Care Access/Utilization, Health Behaviors, and Health Care Expenditures

Work-related physical activity measured by the MET values indicate that one point increase in MET values is related to 21 percentage point reduction in average health care expenditures, with all else constant. Similarly, employed nonobese individuals (as defined by the PBF) have 33% lower health care expenditures, compared with their peers who were obese. Employed people who exercised had 18% lower average health care expenditures compared with employed people who did not exercise. In addition, employed people without any secondary health conditions have 42% lesser health care expenditures compared with their peers who had secondary health conditions. Similarly, people with a usual source of care have about 60% higher average health care expenditures compared with those who did not have access to a usual source of care. A clinical diagnosis of cardiovascular disease contributes to almost 17% higher health care expenditures, and those with chronic respiratory disease contribute to 47% higher health care expenditures among employed people in the study data set.

Excess Health Care Expenditures Attributable to Indicators of Health and Health Behaviors

Table 4 provides estimates for excess health care expenditures attributable to indicators of health and health behaviors among employed people with disabilities. Obesity contributed to about 41% excess costs for employed people with activity limitation compared with the estimates of their peers without obesity. Similarly, obesity accounts for 27% to 33% of excess expenditures for people with various disability classifications. In the similar vein, secondary conditions account for about 16% to 25% of excess health care expenditures among working people with various disability classifications. Lack of exercise and alcohol consumption accounted for one fourth to over one-third excess health care expenditures among employed people with disabilities. A diagnosis of cardiovascular condition contributed to 5% to 10% of excess health care expenditures among people with disabilities. Similarly, a diagnosis of chronic respiratory illness contributed to 7% to 13% excess health care expenses for working people with disabilities.

Table 4.

Percentage of Excess Health Care Expenditures Attributable to Indicators of Health Status and Health Behaviors Among People With Disabilities

ICF-based disability classification	Obesity (%)	Secondary conditions (%)	Lack of exercise (%)	Alcohol consumption (%)	Diagnosis of CVS (%)	Diagnosis of RD (%)
Activity limitation	41.23	21.71	33.08	34.00	9.39	12.60
Participation restriction	26.99	20.32	29.45	33.03	5.35	6.71
Activity limitation and impairment	32.70	26.20	32.96	34.48	9.29	12.98
Participation restriction and impairment	28.78	16.39	29.92	28.82	5.45	10.37
Activity limitation and participation restriction	32.12	20.01	27.15	23.32	7.26	9.33
All domains	26.96	19.64	25.36	24.29	7.61	11.75

Source: National Health Interview Survey–Medical Expenditure Panel Survey linked files 2004–2007.

Note: ICF = International Classification of Functioning, Disability and Health; CVS = cardiovascular system; RD = respiratory disease.

Discussion

With an increasingly aging workforce and prevalence of disabilities, soaring costs for providing health care, and the fiscal austerity imposed by the “global recession,” understanding factors contributing to higher health care expenditures for people with and without disabilities become crucial, especially for employers. Through the analysis of national data sets, this research estimates higher health care expenditures among working people with disabilities compared with their peers without disabilities. More importantly, the research also demonstrates that an individual’s health status, physical activity at work, secondary health conditions, and health-related behaviors substantially affect the overall health care expenditures in general and specifically among people with disabilities.

Targeting the “modifiable factors” such as employees’ health status and health-related behaviors, employers have a potential opportunity to design programs that can help employees maintain healthy lifestyles, thereby stemming the rising costs of providing health care benefits to their employees. This research, by generating estimates for change in average annual health care expenditures in relation to various “modifiable factors,” demonstrates the potential magnitude of reduction in health care expenses that can be achieved by removal of such factors. For example, based on this analysis, one can infer that a highly effective workplace intervention for reducing obesity has a potential to reduce average health care expenditures by 33%; similarly, engaging in exercise programs has a potential for saving 17% of annual average health care expenditures. More importantly, by estimating excess health care expenses attributable to health status and health behaviors for people with disabilities, this research provides compelling data for employers to consider investing their limited resources in workplace wellness programs, hopefully thereby stemming their health benefit costs. For example, our analysis indicates that obesity alone contributes to 41% of the excess health care expenses among working people with activity limitations. This means that any intervention preventing obesity among these individuals would potentially save 41% of their health care expenditures. However, one must note that these projections could be an overestimate of savings achieved as several programmatic and participation factors attenuate the potential impacts of workplace health-promotion programs. In addition, our calculation of excess health care expenditures assumes independence between the variables in the regression models, which may not hold well for all the independent variables in the regression model. However, these data provide a compelling picture for employers to consider offering comprehensive health-promotion programs at workplace with an intention to affect these “modifiable factors.”

Wellness Programs for People With Disabilities

Many health-promotion programs continue to be inaccessible for people with disabilities (Blanck, 2008). Despite the legal provisions for accessible environments for people with disabilities under the ADA, built environments pose barriers to participation of people with disabilities in these programs. The attitudinal and programmatic barriers further discourage their participation (Rimmer, 2005).

Rimmer and Rowland (2008) provide overarching strategies for developing an inclusive wellness program for people with disabilities through understanding of the interactions between personal factors (e.g., self-motivation, self-efficacy, level of functioning), environmental factors (e.g., physical accessibility to wellness centers), and interrelations between secondary conditions, associated conditions, and chronic conditions. They posit that most models of wellness are inaccessible for people with disabilities and introduce an Empowerment Health model consisting of the following: (a) management of associated conditions (e.g., self-care management of chronic diseases); (b) prevention or reduction in secondary conditions through health behavior modifications; and (d) eliminating environmental barriers that limit participation in wellness activities. Furthermore, the “Living Well With a Disability Program” developed by Ravesloot et al. (1998) has shown promising results in reducing health care expenditures for people with disabilities in community-based settings. Using a combination of social cognitive models and a consumer-directed approach for developing health-promotion programs, the Living Well With a Disability Program implemented in Centers for Independent Living resulted in an average savings of US$807 per program participant over 12 months (Ravesloot, Seekins, & White, 2005). Furthermore, advocating for the role of health promotion for individuals with disabilities seeking employment, Ipsen, Seekins, and Ravesloot (2010) demonstrated substantial reductions in secondary health conditions among Living Well With a Disability Program when comparing their data with Vocational Rehabilitation clients surveyed across 10 states.

As the recently signed Patient Protection and Affordability Care Act (Affordability Care Act) aims to extend support to employers to implement workplace wellness programs through tax credits, training and technical assistance, and seed money grants, it is critical for employers to consider inclusive approaches in health-promotion programs to ensure participation of people with disabilities. Comprehensive wellness programs grounded in theories of independent living and behavior change (e.g., Living Well with a Disability program and Empowerment Health Approach) needs more careful considerations by employers. Besides supporting employers in establishing and evaluating wellness programs, government funding should be used to examine ways to leverage employer support by utilizing the network of community-based providers of rehabilitation services to assist employers, especially small businesses.

Furthermore, community-based providers of rehabilitation services, especially in Vocational Rehabilitation, should actively partner with employers of their clients to explore ways of leveraging resources for seeking support for establishing universally accessible wellness programs. Deeper understanding of the corporate work culture and disability-context among rehabilitation counselors and researchers is critical for developing customized programs that can be embedded into workplace culture. Furthermore, rehabilitation counselors should work toward improving employers’ understanding of various disability-specific legislations (e.g., ADA, FMLA, HIPAA) and its relationship to worksite wellness programs for ensuring high compliance and minimizing unintended discriminatory practices.

Returns on Investment (ROI) in Workplace Health-Promotion Programs

Traditionally, the returns on investment in workplace health promotion have attempted to measure only the immediate benefits to employers in terms of reduced health care expenses and worker productivity. However, the contribution of wellness programs exceeds the immediate benefits, by continuing to engage working-age individuals in the labor market as well as delaying retirement due to increased health and well-being (Partners for Prevention, 2009). The returns on investment equations have yet to consider examining the potential costs saved due to continued labor market engagement and reduced transition of individuals on public benefits. In the contemporary economic environment, a comprehensive return on investment framework is especially important with respect to mature workers and people with disabilities. Such a framework, coupled with longer-term analysis of programs, will further help in demonstrating the broader impact of such wellness initiatives to employers and society alike. This also has a potential to increase stakeholders in worksite wellness programs, which could assist in leveraging limited employer resources and public health funding nationally.

In a recent press release, Health and Human Services Secretary Kathleen Sibelius announced US$10 million in funding for establishing and evaluating workplace health-promotion programs for companies of “all sizes” (HHS, 2011). Such funding tools must consider supporting studies that not only examine the structural and implementation factors impacting the program outcomes but also provide sufficient long-term follow-up time to provide a broader analyses, if impact of such programs.

Strengths

Previous research using the MEPS data set revealed higher health care expenditures for people with disabilities (Olin & Dougherty, 2006). However, the narrower definition identifying individuals with disabilities with only significant functional limitations limits the generalizability of results to the overall population of people with disabilities. The ICF-based framework for identifying people with disabilities in this research enables us to establish a more inclusive definition of people with disabilities, by explicitly incorporating information regarding specific impairments of body structures and functions, participation restrictions, and activity limitations contributing to the disabling experience. This was further reflected in the fact that this research overestimated the number of people with disabilities to 64 million for the years 2004 and 2005, in comparison with the more recent estimates of approximately 50 million people with disabilities using the American Community Survey (Erickson, Lee, & von Schrader, 2010). Furthermore, the ICF-based framework of disability is an inclusive model that places disability on the continuum of health and well-being, thus characterizing disability as part of the human experience and not a discreetly disadvantaged demographic group (Kostanjsek, 2011). From this perspective, this research has made a unique contribution in ways of identifying and classifying people with disabilities, as well as providing estimates of health care expenditures relating to the domains of ICF-based framework.

By merging the MEPS data with previous year’s NHIS data our analysis established a panel data set for longitudinal analysis. In addition, our use of GEE for constructing regression models helped in estimating relationships between health care expenditures and various independent variables. Furthermore, our analysis of excess health care costs attributable to specific health status and health behaviors across various disability classifications has enabled examination of potential savings one could realize in the absence of such factors. However, one must recognize that computation of such parameters assumes total independence between the variables in regression models, and it is likely that some interaction effects exists between explanatory variables (e.g., type of disability classification and obesity) in our regression models (Benichou, 2001).

Limitations

This research utilizes national data collected based on self-reporting of health care utilization, health status, and other sociodemographic variables. The fact that health conditions were self-reported could potentially lead to misclassification of conditions and has been shown to be prone to overreporting in some instances (Johnson & Sanchez, 1993). In addition, many individuals in the MEPS study were not aware of the amount paid by their health insurance toward specific services. To improve the quality of reporting on health expenditures, data were imputed based on a sample from the MEPS medical provider component. This method is potentially vulnerable to imputation errors and the analysis does not account for such imputed errors.

It is also important to understand the statistical consequences of log transformation of total health care expenditures for directly applying the estimates in this research as benchmarks for evaluation of health-promotion programs. Although the log-normal distribution helps in satisfying the statistical assumption of a linear relationship between the outcome and independent variables, from an actual dollar value perspective, the assumption of linearity may not hold true (Manning & Mullahy, 2001). Due to this difference in distribution patterns, it is likely that health-promotion programs with small-effect size may not see the actual desired drop in total health care expenses. In addition, it is likely that some independent variables could also be mediator variables, describing a relationship between total health care costs and disability classifications, for example, secondary conditions and obesity. Future research must include an investigation using techniques of mediation analysis and structural equation models to examine the interrelationship among the variables for studying causal pathways and mechanisms to observe relationships.

Implications for Rehabilitation Counseling

With an aging workforce, and an increased push in the hiring of people with disabilities, including returning veterans, employers will need to be increasing aware of ways to maximize productive participation of these groups in their workforce. Health care costs will continue to be a substantial component of operational overhead for employers, and strategies for reducing such costs have become crucial in the contemporary times of the global recession. Employers considering offering workplace wellness programs should examine comprehensive models that are inclusive for employed people with disabilities. In addition, principles of Universal Design should be considered to increase physical accessibility for employees with disabilities (North Carolina Office on Disability and Health, 2001). Attitudinal barriers toward participation of people with disabilities in health-promotion programs should also be simultaneously addressed to ensure barrier-free access to health-promotion programs for employees with disabilities to achieve maximum impact.

The results of this study have implications for rehabilitation counselor practice, education, and research. It is important for practitioners to be aware of the factors that might affect employers’ willingness to hire people with disabilities, so that they can inform the clients that they are working with and so they can do educational outreach to help employers in addressing these issues directly. The practice of rehabilitation counseling is often targeted to a particular service or outcome; however, among our interventions it should be at least a general awareness building among our clients of the importance of health factors in their long-term well-being. Also important is the identification of resources within the community and in their work setting that can help them to address efforts toward long-term health. In addition, building capacity among practitioners and the agency that they work for to provide consultation to employers for addressing the demand side of the equation that may affect higher health care expenditures for an aging workforce and employees with disabilities will be imperative to building a receptivity to hiring and retaining workers with disabilities.

Building awareness about these issues early on in the counselor preparation process is imperative. Consideration about health care factors and employer concerns about health care costs can be included in counseling and job development and placement courses, as well as included in public policy coursework. Furthermore, an understanding of health behavior theories and its emerging adaptation for people with disabilities will help counselors appreciate the contributions of health-related factors in employment for people with disabilities, as well as equip them with expertise to support employers concerns on strategies for containing costs of health care benefits for employees with chronic health conditions and disabilities. We increasingly are aware of the impact and sometimes disincentives that concerns about health care coverage can have for people with disabilities who are thinking of leaving the safety net of Social Security benefits to engage in full-time employment. We, therefore, are beginning to better prepare our students to more confidently and competently offer advisement to their clients on benefits planning. We seldom, however, educate our clients about the importance of lifelong health and well-being nor do we educate how to consult with employers about workplace program design that will facilitate health enhancement behaviors among people with disabilities and assure inclusion through addressing accessibility considerations.

This article also offers a number of ideas of fertile areas for future research. Having rehabilitation counselor education faculty partner with health economists, health educators, public health professionals, and others with expertise in the health area will afford our field the opportunity to insert our expertise and perspectives about inclusion into this very important area. Doing so will greatly heighten the likelihood that the needs of people with disabilities will be addressed not only at the workplace level but also more broadly in American public policy.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is supported by a grant to Cornell University from the U.S. Department of Education National Institute on Disability and Rehabilitation Research for a Rehabilitation Research and Training Center on Employer Practices Related to Employment Outcomes for Individuals With Disabilities (Grant No. H133B100017).

Notes

References

Aldana

S. G.

Merrill

R. M.

Price

Hardy

Hager

(2005). Financial impact of a comprehensive multisite workplace health promotion program. Preventive Medicine, 40, 131–137.

Anderson

D. R.

Whitmer

R. W.

Goetzel

R. Z.

Ozminkowski

R. J.

Dunn

R. L.

Wasserman

Health Enhancement Research Organization Research Committee. (2000). The relationship between modifiable health risks and group-level health care expenditures: Health Enhancement Research Organization (HERO) Research Committee. American Journal of Health Promotion, 15(1), 45–52.

Baicker

Cutler

Song

(2010). Workplace wellness programs can generate savings. Health Affairs, 29, 304–311.

Banthin

J. S.

Bernard

D. M.

(2006). Changes in financial burdens for health care: National estimates for the population younger than 65 years, 1996 to 2003. Journal of American Medical Association, 296, 2712–2719.

Benichou

(2001). A review of adjusted estimators of attributable risk. Statistical Methods in Medical Research, 10, 195–216.

Bjelland

Bruyere

S. M.

von Schrader

Houtenville

A. J.

Ruiz-Quintanilla

Weber

D. B.

(2010). Age and disability employment discrimination: Occupational rehabilitation implications. Journal of Occupational Rehabilitation, 20, 456–471.

Blanck

P. D.

(2008). The right to live in the world: Disability yesterday, today, and tomorrow. Texas Journal on Civil Liberties & Civil Rights, 13, 367–401.

Burkhauser

R. V.

Cawley

(2008). Beyond BMI: The value of more accurate measures of fatness and obesity in social science research. Journal of Health Economics, 27, 519–529.

Burton

Gurrin

Sly

(1998). Extending the simple linear regression model to account for correlated responses: An introduction to generalized estimating equations and multi-level mixed modeling. Statistics in Medicine, 17, 1261–1291.

10.

Call

Gerdes

Robinson

(2009). Corporate and worksite wellness programs: A research review focused on individuals with disabilities. Gaithersburg, MD: Social Dynamics LLC. Retrieved from http://www.dol.gov/odep/research/CorporateWellnessResearchLiteratureReview.pdf

11.

Centers for Medicare and Medicaid Services (2010). National health expenditure accounts: Methodology paper. Retrieved from https://www.cms.gov/NationalHealthExpendData/downloads/dsm-10.pdf.

12.

Crossley

M. A.

(2005). Discrimination against the unhealthy in health insurance. Kansas Law Review, 54, 1–93. Retrieved from http://law.bepress.com/cgi/viewcontent.cgi?article=1031&;context=pittlwps

13.

Elixhauser

Steiner

Palmer

(2011). Clinical classifications software (CCS): U.S. Agency for Healthcare Research and Quality. Retrieved from http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp

14.

Erickson

Lee

von Schrader

(2010). Disability statistics from the 2008 American Community Survey (ACS). Ithaca, NY: Cornell University Rehabilitation Research and Training Center on Disability Demographics and Statistics (StatsRRTC). Available from www.disabilitystatistics.org

15.

Ezzati-Rice

T. M.

Rhode

Greenblatt

(2008, March). Sample design of the Medical Expenditure Panel Survey household component, 1998–2007 (Methodology Report No. 22). Rockville, MD: Agency for Healthcare Research and Quality. Retrieved from http://www.meps.ahrq.gov/mepsweb/data_files/publications/mr22/mr22.pdf

16.

Furrow

(2007). Access to healthcare and political ideology: Wouldn’t you really rather have a pony? Western New England Law Review, 29, 405–420.

17.

Garrett

Martini

E. M.

(2007). The boomers are coming: A total cost of care model of the impact of population aging on the cost of chronic conditions in the United States. Disease Management, 10, 51–60.

18.

Goetzel

R. Z.

Anderson

D. R.

Whitmer

R. W.

Ozminkowski

R. J.

Dunn

R. L.

Wasserman

Health Enhancement Research Organization Research Committee. (1998). The relationship between modifiable health risks and health care expenditures: An analysis of the multi-employer HERO health risk and cost database. Journal of Occupational and Environmental Medicine, 40, 843–854.

19.

Goetzel

R. Z.

Ozminkowski

R. J.

(2008). The health and cost benefits of work site health-promotion programs. Annual Reviews of Public Health, 29, 303–323.

20.

Gulley

S. P.

Rasch

E. K.

Chan

(2011). The complex web of health: Relationships among chronic conditions, disability, and health services. Public Health Reports, 126, 495–507.

21.

Hill

(2011). The economic recession: Impact on employment trends and retirement patterns. New York, NY: Towers Watson. Retrieved from http://www.towerswatson.com/united-states/newsletters/insider/5086

22.

Institute of Medicine. (2007). The future of disability in America. Washington, DC: National Academies Press.

23.

Ipsen

Seekins

Ravesloot

(2010). Building the case for delivering health promotion services within the vocational rehabilitation system. Rehabilitation Counseling Bulletin, 53, 67–77.

24.

Jackson

A. S.

Stanforth

P. R.

Gagnon

Rankinen

Leon

A. S.

Rao

D. C.

. . . Wilmore

J. H.

(2002). The effect of sex, age and race on estimating percentage body fat from body mass index: The heritage family study. International Journal of Obesity and Related Metabolic Disorders, 26, 789–796.

25.

Johnson

Sanchez

(1993). Household and medical reports on medical conditions: National medical expenditure survey. Journal of Economic and Social Measurement, 19, 199–223.

26.

Johnson

R. W.

(2007). What happens to health benefits after retirement? Issue in brief. Chestnut Hill, MA: The Urban Institute. Retrieved from http://www.urban.org/publications/1001053.html

27.

Kaiser Family Foundation and Health Research Education Trust. (2011). Employer health benefits annual survey. Retrieved from http://ehbs.kff.org/

28.

Kostanjsek

(2011). Use of the international classification of functioning, disability and health (ICF) as a conceptual framework and common language for disability statistics and health information systems. BMC Public Health, 11(Suppl. 4), S3.

29.

Kuoppala

Lamminpaa

Husman

(2008). Work health promotion, job well-being, and sickness absences—A systematic review and meta-analysis. Journal of Occupational and Environmental Medicine, 50, 1216–1227.

30.

Liang

K. Y.

Zeger

S. L.

(1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 13–22.

31.

Linnan

Bowling

Childress

Lindsay

Blakey

Pronk

. . . Royall

(2008). Results of the 2004 National Worksite Health Promotion Survey. American Journal of Public Health, 98, 1503–1509.

32.

Manning

W. G.

Mullahy

(2001). Estimating log models: To transform or not to transform? Journal of Health Economics, 20, 461–494.

33.

Morton

Foster

Sedlar

(2005). Managing the mature workforce. New York, NY: The Conference Board.

34.

Musich

Napier

Edington

(2001). The association of health risks with workers’ compensation costs. Journal of Occupational and Environmental Medicine, 43, 534–541.

35.

North Carolina Office on Disability and Health. (2001). Removing barriers to health clubs and fitness facilities. Chapel Hill, NC: Frank Porter Graham Child Development Center.

36.

Olin

Dougherty

D. D.

(2006, March). Characteristics and medical expenses of adults 18 to 64-years old with functional limitations, combined years 1997-2002 (Working Paper No. 06002). Rockville, MD: Agency for Healthcare Research and Quality.

37.

Partners for Prevention. (2009). Healthy workforce 2010 and beyond. Washington, DC: Author. Retrieved from http://www.prevent.org/downloadStart.aspx?id=18

38.

Pollard

Johnston

Dieppe

(2011). Exploring relationships between International Classification of Functioning, Disability and Health (ICF) constructs of impairment, activity limitation, and participation restriction in people with osteoarthritis prior to joint replacement. BMC Musculoskeletal Disorders, 12, 97.

39.

Rasch

E. K.

Magder

Hochberg

M. C.

Magaziner

Altman

B. M.

(2008). Health of community-dwelling adults with mobility limitations in the United States: Incidence of secondary health conditions. Part II. Archives of Physical Medicine and Rehabilitation, 89, 219–230.

40.

Ravesloot

Seekins

White

(2005). Living well with a disability health promotion intervention: Improved health status for consumers and lower costs for healthcare for policymakers. Rehabilitation Psychology, 50, 239–245.

41.

Ravesloot

Young

Q.-R.

Norris

Szalda-Petree

Seekins

White

G. W.

. . . Golden

(1998). Living well with a disability: A workbook for promoting health and wellness (2nd ed.). Missoula: The University of Montana Rural Institute.

42.

Rimmer

J. H.

(2005). The conspicuous absence of people with disabilities in public fitness and recreation facilities: Lack of interest or lack of access? American Journal of Health Promotion, 19, 327–329.

43.

Rimmer

J. H.

Rowland

J. L.

(2008). Health promotion for people with disabilities: Implications for empowering the person and promoting disability friendly environments. American Journal of Lifestyle Medicine, 2, 409–420.

44.

Rimmer

J. H.

Wang

Yamaki

Davis

(2010). Documenting disparities in obesity and disability. Retrieved from http://www.uic-chp.org/ARTICLES/New%20Research/Focus.pdf

45.

Thorpe

K. E.

Howard

D. H.

(2006). The rise in spending among Medicare beneficiaries: The role of chronic disease prevalence and changes in treatment intensity. Health Affairs, 25, 378–388.

46.

Toosi

(2007). Labor force projections to 2016: More workers in their golden years. Monthly Labor Review, 130(11), 33–52. Retrieved from www.bls.gov/opub/mlr/2007/11/art3full.pdf

47.

Tudor-Locke

Ainsworth

B. E.

Washington

T. L.

Troiano

R. P.

(2011). Assigning metabolic equivalent values to the 2002 census occupational classification system. Journal of Physical Activity & Health, 8, 581–586.

48.

Tudor-Locke

Washington

T. L.

Ainsworth

B. E.

Troiano

R. P.

(2009). Linking the American Time Use Survey (ATUS) and the compendium of physical activities: Methods and rationale. Journal of Physical Activity & Health, 6, 347–353.

49.

U.S. Department of Health and Human Services. (2011). $10 Million in Affordable Care Act funds to help create workplace health programs [Press release]. Retrieved from http://www.hhs.gov/news/press/2011pres/06/20110623a.html

50.

Vogenberg

F. R.

Holland

J. P.

Liebeskind

(2007). Employer benefit design considerations for the era of biotech drugs. Journal of Occupational and Environmental Medicine, 49, 626–632.

51.

World Health Organization. (2001). International classification of functioning, disability and health. Geneva, Switzerland: World Health Organization.

52.

Wright

Adams

Beard

M. J.

Burton

W. N.

Hirschland

McDonald

. . . Edington

D. W.

(2004). Comparing excess costs across multiple corporate populations. Journal of Occupational and Environmental Medicine, 46, 937–945.