Access to and receipt of health care in the United States is not distributed equally across population groups.1 There are significant variations in ability to obtain needed care in a timely manner. A substantial proportion of the population has no insurance, and these individuals encounter significant barriers to accessing health care.2–4 Medicaid expansion and implementation of health care exchanges under the Affordable Care Act (ACA) will help reduce uninsurance among people with disabilities and those in underserved racial and ethnic groups, at least in some states.5,6 However, current projections suggest that physician capacity will be inadequate to care for new Medicaid patients;7,8 hence, many of the newly insured may continue to lack a usual source of health care and remain at high risk of having unmet health care needs.
In the Behavioral Model of Health Services Use, having a usual source of care (USC) is an important indicator of potential health care access.9,10 Having a USC is positively associated with regular health maintenance visits and receipt of preventive services.11–13 Individuals without a USC are more likely to have unmet health care needs.7 Failure to receive health care when needed may lead to exacerbation of health problems, requiring utilization of more extensive and expensive care in the future.14 Although the Behavioral Model categorizes USC as an enabling factor similar to health insurance,9,10 USC is logically “farther down the chain” in that people with health insurance are much more likely to have a USC,7 which in turn is associated with increased receipt of care.
The Behavioral Model is designed to study utilization of health care (realized access), but greater utilization does not necessarily result in health care needs being better met. That concern suggests the importance of focusing instead on whether the health care received was adequate to meet perceived needs. Examining unmet need moves beyond realized access to address an aspect of patient satisfaction with care, which is a component of effective health care access.10
Both having a USC and unmet health care needs have been found to vary in relation to sociodemographic differences in the US population. For example, people in traditionally underserved racial and ethnic groups are less likely to have a USC, and less likely to receive regular checkups and other preventive care than non-Hispanic whites.7,12,15,16 People with disabilities (defined in this context as self-reported limitations in physical, sensory, cognitive, and/or emotional functioning) are more likely than those with no disabilities to have a USC, yet they are also much more likely to report unmet health care needs.17–21 Where there are disparities related to both disability and to race/ethnicity individually, one might anticipate that people in underserved racial and ethnic groups who also have a disability could experience magnified disparities. Further, people with disabilities in underserved racial and ethnic groups may experience a complex mix of advantages and disparities, particularly with regard to having a USC. However, very little is known about this intersection. Indeed, we were unable to locate any prior studies that have examined the interaction of disability status and race/ethnicity on access to health care and unmet health care needs. Similarly, there has been negligible attention in the literature to potential additive effects of race/ethnicity and disability.
The purpose of the present study was to determine how disability status in combination with membership in an underserved racial or ethnic group is associated with having a USC and having unmet health care needs while controlling for important predisposing and enabling factors related to health care access and receipt. Because the health insurance landscape is currently shifting, we were interested in differences in USC and unmet health care needs that might remain after accounting for differences in insurance status. Similarly, even if access to a USC improves, disparities in unmet health care needs may well continue to exist. The ACA does include provisions pertaining to culturally competent care for both underserved racial and ethnic groups and people with disabilities, but it is likely to be several years before substantial widespread reductions in disparities are apparent. Thus, we focused on disparities in effective receipt of health care that may require targeted intervention regardless of improvements in potential access to health care. Because most people over the age of 64 are covered by Medicare, we limited our analyses to working-age adults 18–64 years old. Our key research questions were:
- Are there interactions between race/ethnicity and disability such that disparities in USC and unmet health care needs are different for groups at the intersection than would be expected based solely on adding together the disparities associated with being in each group? (ie, are disability effects of consistent magnitude across racial and ethnic groups and vice versa?)
- Regardless of whether interactions are present, what is the additive effect of disability status and membership in an underserved racial or ethnic group on USC and unmet health care needs? Are there significantly greater disparities associated with having both a disability and being in an underserved racial or ethnic group than are associated with being in either group alone?
Our sample was drawn from the Medical Expenditure Panel Survey (MEPS) Household Component, which is conducted by the Agency for Healthcare Research and Quality. The MEPS includes data regarding demographics, health care utilization and expenditures, sources of payment, health insurance, and quality of care. It has an overlapping panel design with a new panel selected each year from the previous year’s National Health Interview Survey sample.22,23 Data are gathered via 5 in-person interviews over 2 years, and multistage stratified sampling is employed to provide a nationally representative sample of the US civilian noninstitutionalized population. Racial and ethnic minorities as well as low-income respondents are oversampled to increase precision of estimates for these groups.22 The Agency for Healthcare Research and Quality creates full-year consolidated files weighted to provide annualized US population estimates. We combined full-year files from 2002 to 2010 to provide adequate sample size for analyzing smaller racial and ethnic groups. Of the overall 294,513 MEPS respondents for 2002–2010, we identified 170,742 adults 18–64 years of age. After excluding 5714 observations with missing data on variables of interest, our final analytic sample included 165,028 respondents.
We assessed whether or not individuals reported having a USC (yes/no). A follow-up question asked those who answered yes what their source of care was. If the emergency department was the USC, we recoded respondents as having no USC. We analyzed 4 indicators of unmet health care needs in the past 12 months, each coded yes or no: (1) delay in receiving needed medical care; (2) unable to get needed medical care; (3) delay in getting prescription medication; (4) unable to get needed medication.
Primary Independent Variables
We created a 3-level variable indicating presence and severity of disability, based on the categories of basic and complex activity limitations described by Altman and Bernstein.17 Basic activity limitations were identified by affirmative responses to one or more MEPS survey questions about limitations in physical functions such as walking, lifting, standing, bending, reaching, or grasping; cognitive limitations including confusion or memory loss or difficulty making decisions; and difficulty with vision (while wearing glasses, if used) or hearing (with hearing aid, if used). We also included individuals who used assistive devices such as wheelchairs or walkers. Complex activity limitations were determined by an affirmative response to one or more MEPS items about needing assistance with activities of daily living (eg, bathing, dressing) or instrumental activities of daily living (eg, paying bills, going shopping), or having limitations in work, housework, social, or recreational activities. We grouped respondents into the following 3 categories: (1) no basic or complex activity limitations (nondisabled reference group); (2) limitations in basic activities only; and (3) limitations in complex activities. Race and ethnicity were grouped into a single variable with the following mutually exclusive categories: non-Hispanic white, non-Hispanic black, non-Hispanic American Indian/Alaskan Native (AIAN), non-Hispanic Asian/Native Hawaiian/Pacific Islander (AHPI), non-Hispanic Multiple races; or Hispanic of any race.
We included the following sociodemographic variables in our analyses: age (18–29, 30–39, 40–49, 50–59, and 60–64 y), sex, family income as percent of the Federal Poverty Line (≥400%, 200% to <400%, 125% to <200%, 100% to <125%, and <100%), employment status (employed or not employed), and education (bachelor’s degree and higher, other degree, general educational development [GED]/high school diploma [HS], or no GED/HS). A variable combining presence and type of insurance was also included as a key covariate impacting access to health care. The majority of individuals in our 18–64 age range who received public insurance all year were on Medicaid only (72.7%). Given the relatively small proportions who received Medicare only (12.0%) or were dually eligible (15.3%), we combined all individuals who were publicly insured all year into a single category. Final categories for our insurance variable were: (1) privately insured all year; (2) publicly insured all year; (3) uninsured part of the year; and (4) uninsured all year. “Privately insured all year” served as the reference category against which all other categories were compared. USC (yes/no) was included as an additional covariate in models of unmet health care needs.
We first calculated the prevalence of being without a USC and having each form of unmet health care need for each disability and racial/ethnic subgroup. We then performed logistic regression analyses examining each outcome while controlling for covariates. Our models examined the main effects of both disability and race/ethnicity. People with no disabilities and non-Hispanic whites served as the reference groups for the disability and race/ethnicity variables, respectively. In addition, we included an interaction term for disability and race/ethnicity as an independent variable. Interaction terms are essentially ratios of odds ratios that compare one main effect to another. Those that were significantly different from null described an interaction whereby the effect of race or ethnicity was different for people with disabilities compared to those without disabilities.
Finally, we conducted logistic regression analyses with a variable combining disability status and race/ethnicity to examine effects for each disability by race/ethnicity subgroup in comparison with nondisabled whites, while controlling for covariates. The 18 categories of the combined variable were: nondisabled white (reference), white with basic activity limitations, white with complex limitations, nondisabled black, black with basic activity limitations, black with complex activity limitations, nondisabled AIAN, AIAN with basic activity limitations, AIAN with complex activity limitations, nondisabled AHPI, AHPI with basic activity limitations, AHPI with complex activity limitations, nondisabled multiracial, multiracial with basic activity limitations, multiracial complex activity limitations, nondisabled Hispanic, Hispanic with basic activity limitations, and Hispanic with complex activity limitations.
Given that we were examining 5 dependent variables, we used a more conservative P value of 0.01 rather than the more typical 0.05 to reduce the chance of a type I error. Although we calculated estimates only for 18–64 year olds with nonmissing data on the variables in our study, all observations in the full dataset were used to calculate variance and SEs. This preserved the nationally representative nature of the data. Stata version 12.1 was used for all statistical analyses.24
Table 1 shows the demographic and socioeconomic characteristics of our analytic sample. The majority of the sample was nondisabled, non-Hispanic white, had a family income above the Federal Poverty Line, and at least a GED or HS.
Across racial and ethnic groups, lower proportions of people with disabilities were without a USC. However, people with disabilities had substantially higher proportions reporting each form of unmet health care need. This was especially true for people with complex activity limitations (Table 2). Regardless of disability status, the proportion of individuals without a USC was higher in all underserved racial and ethnic groups with the exception of AIAN. Hispanics were the group with the highest proportion having no USC (43.9%). Despite higher proportions without a USC, blacks, AHPI, and Hispanics each had similar or lower proportions with unmet health care needs, compared with whites. The racial groups with the highest proportions reporting unmet needs were multiracial individuals, followed by AIAN.
Disability status combined with membership in an underserved racial or ethnic group did not appear to have a compound impact for blacks, AHPI, or Hispanics. In other words, people with disabilities in these racial and ethnic groups had similar or lower proportions reporting unmet needs compared to whites with disabilities. AIAN and multiracial individuals with disabilities had higher proportions with unmet needs relative to both whites with disabilities and nondisabled members of their own racial groups. However, the confidence intervals for these estimates were quite wide (Table 2).
In regression models controlling for covariates, main effects revealed that people with disabilities had significantly lower odds of being without a USC, but greater odds of experiencing all forms of unmet health care need (Table 3). The magnitude of these effects was greatest for people with complex activity limitations. Main effects of race and ethnicity indicated that blacks, AHPI, and Hispanics had higher odds of having no USC compared with non-Hispanic whites. However, with USC included as a covariate, these same groups had significantly lower odds than whites of experiencing delayed or forgone medical care or prescription medications (Table 3). Multiracial individuals, in contrast, had significantly higher odds than whites of delays in getting needed prescription medication.
Most of the interactions between disability and race/ethnicity were not statistically significant, indicating that racial and ethnic disparities were similar for people both with and without disabilities and, correspondingly, disability-related disparities were similar across racial and ethnic groups. The exceptions were for forgone medical care (odds ratio [OR]=1.35; 99% confidence interval [CI], 1.01–1.80; P=0.01) and forgone prescription medication (OR=1.58; 99% CI, 1.12–2.24; P=0.001) among Hispanics with basic activity limitations. Given that Hispanics overall had reduced odds of reporting these unmet needs, the elevated odds ratios for the interactions imply that the disadvantage associated with having a basic activity limitation was of significantly greater magnitude for Hispanics than for whites.
Figures 1 to 3 provide a visual display of the combined effects of disability and race/ethnicity while controlling for covariates. In most racial and ethnic groups, there was a clear stair step progression with nondisabled individuals being the most likely to lack a USC, followed by individuals with basic activity limitations and then by those with complex activity limitations. Relative to whites, the stair step pattern was shifted to the right somewhat for some of the racial and ethnic categories, reflecting the overall racial/ethnic disparities for those categories and the fact that people with disabilities tend to share in those disparities to some extent (ie, the direction of racial disparities was consistent across disability groups). However, with the exception of Hispanics with basic activity limitations, none of the disability groups had significantly elevated odds of lacking a USC (Fig. 1).
With regard to delays in getting needed medical care, a comparable disability stair step pattern was apparent within most racial and ethnic groups. Consistent with the main effects described earlier, the steps were in the opposite direction from those seen for lacking a USC. In each racial/ethnic group, people with disabilities had greater odds of delays in getting needed care. Except among AIAN, the effect was magnified for those with complex activity limitations (Fig. 2). In all cases, people with disabilities were significantly more likely than the nondisabled white reference group to report delayed medical care, whereas none of the nondisabled underserved racial or ethnic groups had significantly elevated odds. In fact, blacks, AHPI, and Hispanics without disabilities had reduced odds of delayed medical care. ORs for people with disabilities in these racial and ethnic groups were also somewhat lower than for their counterparts with disabilities in other racial groups, although overlapping confidence intervals suggest that these within-disability group differences would not be statistically significant. At the other end of the spectrum, multiracial individuals with disabilities had higher odds of delayed care than whites with disabilities, but the confidence intervals for these estimates were quite wide and completely overlapped with those for whites.
Very similar patterns were apparent for not getting needed medical care (Fig. 2) and for delayed and forgone prescription medications (Fig. 3). Figure 3 suggests that the main effect of delayed prescription medication for multiracial individuals was driven by people with disabilities. Also of note, for forgone prescription medication, the confidence interval for AIAN with basic activity limitations did not overlap that for whites with basic activity limitations. Because our overall analysis only directly compared each group to the reference group of nondisabled whites, we conducted a follow-up test to compare AIAN and whites with basic activity limitations. In an unadjusted analysis, AIAN had significantly higher odds of going without needed medication (OR=1.91; 99% CI, 1.03–3.52; P=0.01). However, when controlling for covariates, the effect was no longer statistically significant (AOR=1.17; 99% CI, 0.64–2.13; P=0.50).
Figures 2 and 3 shed light on the interactions noted earlier for unmet needs among Hispanics with basic activity limitations. This group did not appear to be at a disadvantage relative to non-Hispanic whites with basic activity limitations; rather, they were closer to having similarly elevated odds of unmet needs than would be expected based on the otherwise low levels of unmet need reported by Hispanics. In other words, there was an interaction, but not an additive disparity.
Although considerable prior research has examined health care disparities impacting people in underserved racial and ethnic groups, and a newer body of literature has studied similar issues for people with disabilities, there has been very little attention to how these disparities combine. Ours is the first study we are aware of to study the interaction of race/ethnicity and disability in relation to health care disparities. We expected that people with disabilities in underserved racial and ethnic groups might have especially high unmet needs, experiencing barriers associated with membership in not one but two marginalized groups. However, we found limited evidence of either significant interaction between disability status and race/ethnicity or significant additive disparity related to both having a disability and belonging to an underserved racial or ethnic group.
With regard to having a USC, disability had a protective effect in all racial and ethnic groups, although not enough to eliminate disparities related to Hispanic ethnicity for those with basic activity limitations. Conversely, disability was a strong risk factor for unmet health care needs in every race/ethnicity group, even when controlling for other known risk factors. The lower levels of unmet need among blacks, AHPI, and Hispanics conferred somewhat of a protective effect for people with disabilities in these groups such that there was no elevated disparity associated with both having a disability and belonging to one of these racial or ethnic groups. However, the increased risk of forgone medical and prescription care associated with having a basic activity limitation was significantly greater for Hispanics than for non-Hispanic whites. Hispanics with disabilities may face a daunting combination of linguistic, cultural, physical, and attitudinal barriers to receiving needed care. Unfortunately, as noted by Peterson-Besse and colleagues (this issue), little is currently known about ways in which barriers associated with ethnicity and those related to disability may combine to impact receipt of needed care. Further attention is needed to understanding specific reasons for unmet health care needs, and to implementing strategies to address those problems. For example, Bogenschutz (this issue) found that care coordination and culturally competent care were important facilitators of health care access for immigrant families of individuals with intellectual and developmental disabilities. Both approaches are emphasized under the ACA and hold substantial promise for reducing disparities.25–27
There were some indications of potential additive disparities for AIAN and multiracial individuals with disabilities in that point estimates in the forest plots were typically highest for these groups, yet the confidence intervals were so wide that disparities did not appear to be statistically significantly different from those experienced by whites with disabilities. This was likely due to the relatively small size of these groups. A previous study with an even smaller sample of AIAN did find significantly greater unmet needs for this group compared with non-Hispanic whites, but racial groups in that study were not subdivided by disability status and the analyses did not control for covariates.12 Qualitative research may be an important mechanism to further explore health care experiences among AIAN and multiracial individuals with disabilities, given that these groups are only minimally represented in large-scale survey data.
Our analyses did confirm separate disparities for people with disabilities and for people in underserved racial and ethnic groups (main effects). As has been found in previous studies, people with disabilities were more likely to have an USC, but also more likely to have unmet health care needs.17–21 This situation may reflect the relatively high medical needs of the population of people with disabilities as a whole, particularly those with complex activity limitations. Although people with disabilities interact with physicians more frequently than do people without disabilities,19,28 that increased contact may still not be enough to address all of their needs for care. The disparities in unmet needs may presage later disparities in health outcomes and need for more intensive health care in the future.14 People with disabilities experience a “thinner margin of health”29,30 and are thus at high risk of adverse impacts when health care needs are not met promptly. The margin of health may be even narrower for people with disabilities in underserved racial and ethnic groups, particularly if they have limited financial and other resources for coping with health issues.
With regard to racial and ethnic disparities, our analyses did not suggest disparities in unmet needs for blacks, AHPI, or Hispanics, even in descriptive results. Although this is consistent with MEPS analyses presented in the National Healthcare Disparities Report,1 it was somewhat surprising given that these 3 groups were also the least likely to have a USC. As a sensitivity analysis, we did test whether USC could potentially be a mediator of unmet needs for these or any other racial/ethnic groups (results available from authors). With USC removed from the models, patterns of unmet needs were unchanged for all groups except multiracial. When no longer controlling for differences in having a USC, multiracial individuals had significantly elevated odds of delayed and forgone medical care and forgone prescription medications, in addition to the elevated odds of delays in getting prescription medications that we had seen in our previous models. Thus, USC does appear to have a mediating effect for multiracial individuals, suggesting that improving access to a USC could help reduce unmet needs in this population. Patient-centered medical homes, a model increasingly being implemented as part of health care transformation efforts, may be an especially valuable mechanism for meeting the health care needs of multiracial and other underserved groups, particularly those with elevated medical and care coordination needs.31,32
Other research indicates that blacks and Hispanics tend to have poorer health than whites, and thus could be expected to have greater, not lesser, health care needs.33 Yet Gulley and colleagues (this issue) found lower utilization of ambulatory care among blacks and Hispanics compared with whites. The fact that lower utilization in the context of greater anticipated need does not result in higher proportions reporting unmet needs suggests that perceived need and expectations for services differ in these racial and ethnic groups. Blacks and Hispanics, and perhaps AHPI as well, may have lower expectations of the health care system and thus report fewer unmet needs even though actual unmet need may be higher. There is some evidence of lower health care expectations among blacks and other underserved groups in the US and elsewhere.34,35 This topic bears further investigation among people both with and without disabilities.
We found disparities beyond what could be explained by differences in health insurance status and type. Lack of insurance has historically been a significant problem among underserved racial and ethnic groups3,5,33 and has also impacted substantial numbers of people with disabilities.21,25 The ACA aims to reduce health care disparities attributable to lack of insurance by expanding insurance coverage, but our analyses indicate health care disparities will remain even if insurance disparities are eliminated. Further, when they are insured, people with disabilities and individuals in underserved racial and ethnic groups tend to rely heavily on public insurance (especially Medicaid in the 18–64 age range), and Medicaid expansion is expected to result in even greater proportions of these groups being publicly insured.5,6 Prior research has found greater barriers to timely health care access among Medicaid beneficiaries than among privately insured individuals.36 Thus, Medicaid expansion alone is not a panacea for addressing health care disparities. It remains to be seen whether other components of the ACA, such as implementation of Accountable Care Organizations and increased funding of community health centers, can combine with expanded insurance coverage to effectively reduce health care barriers for vulnerable groups.
This study has certain limitations. Although pooling several years of data allowed us to analyze the combined impact of disability and race in some smaller population groups (AIAN; multiracial), we did not have sufficient sample size to more fully disaggregate racial and ethnic groups. For example, research in the general population has found important differences among native Hawaiians, Pacific Islanders, and Asians from different countries,37 and among Hispanics from different countries38 that we were not able to examine within the population of people with disabilities. Further, we did not have sufficient sample size to analyze different types of disability (eg, physical, hearing, vision, cognitive) separately within each racial and ethnic group. Recent research has indicated differences in unmet health care needs by disability type when controlling for race and ethnicity,39 but how disability type interacts with race and ethnicity has not yet been examined. We did attempt to account for some of the heterogeneity in the disability population by subdividing people with disabilities into those with and without complex activity limitations. For some of the smaller racial groups, this subdivision may have diluted our statistical power for discerning overall disability effects.
Considerable research has examined racial and ethnic health care disparities. Prior research has also established disparities between people with and without disabilities. However, there has been little attention to ways in which the combination of race/ethnicity and disability status impacts access to and receipt of needed health care. Our findings indicate that, among working age adults, Hispanics with basic activity limitations share in disparities but experience less of the protective effects associated with either being Hispanic or having a disability. We found little other evidence of interaction or additive effects of disability and race/ethnicity but did confirm separate disparities for each. Ongoing research is needed to track both disability-related and racial/ethnic disparities, to determine whether increased insurance coverage, provider training, care coordination, and other efforts under the ACA lead to reductions in disparities.
The authors thank Glenn T. Fujiura, two anonymous reviewers, and Robert Weech-Maldonado, MBA, PhD for valuable feedback on previous versions of this manuscript.
1. US Department of Health and Human Services. National Healthcare Disparities Report 2012. 2013.Rockville, MD: Agency for Healthcare Research and Quality.
2. Hoffman C, Paradise J. Health insurance and access to health care in the United States. Ann N Y Acad Sci. 2008;1136:149–160.
3. Institute of Medicine. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. 2003.Washington, DC: National Academy Press.
4. Jerant A, Fiscella K, Tancredi DJ, et al.. Health insurance is associated with preventive care but not personal health behaviors. J Am Board Fam Med. 2013;26:759–767.
5. Clemans-Cope L, Kenney GM, Buettgens M, et al.. The Affordable Care Act’s coverage expansions will reduce differences in uninsurance rates by race and ethnicity
. Health Aff. 2012;31:920–930.
6. Gettens J, Henry AD, Himmelstein J. Assessing health care reform: potential effects on insurance coverage among persons with disabilities. J Disabil Policy Stud. 2012;23:3–13.
7. DeVoe JE, Tillotson CJ, Lesko SE, et al.. The case for synergy between a usual source of care and health insurance coverage. J Gen Intern Med. 2011;26:1059–1066.
8. Sabik LM, Ohri Gandhi S. Impact of changes in Medicaid coverage on physician provision of safety net care. Med Care. 2013;51:978–984.
9. Aday LA, Andersen R. A framework for the study of access to medical care. Health Serv Res. 1974;9:208–220.
10. Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36:1–10.
11. DeVoe JE, Fryer GE, Phillips R, et al.. Receipt of preventive care among adults: insurance status and usual source of care. Am J Public Health. 2003;93:786–791.
12. Kim G, Bryant AN, Goins T, et al.. Disparities in health status and health care access and use among older American Indians and Alaska Natives and Non-Hispanic Whites in California. J Aging Health. 2012;24:799–811.
13. Peek ME, Han JH. Disparities in screening mammography. Current status, interventions and implications. J Gen Intern Med. 2004;19:184–194.
14. Long SK, King J, Coughlin TA. The implications of unmet need for future health care use: findings for a sample of disabled Medicaid beneficiaries in New York. Inquiry. 2005-2006;42:413–420.
15. Abdus S, Selden TMl. Preventive services for adults: how have differences across subgroups changed over the past decade? Med Care. 2013;51:999–1007.
16. Dubay LC, Lebrun LA. Health, behavior, and health care disparities: disentangling the effects of income and race in the United States. Int J Health Serv. 2012;42:607–625.
17. Altman B, Bernstein A. Disability and Health in the United States, 2001–2005. 2008.Hyattsville, MD: National Center for Health Statistics.
18. Chevarley FM, Thierry JM, Gill CJ, et al.. Health, preventive health care, and health care access among women with disabilities in the 1994-1995 National Health Interview Survey, supplement on disability. Womens Health Issues. 2006;16:297–312.
19. Gulley SP, Altman BM. Disability in two health care systems: access, quality, satisfaction, and physician contacts among working-age Canadians and Americans with disabilities. Disabil Health J. 2008;1:196–208.
20. Henning-Smith C, McAlpine D, Shippee T, et al.. Delayed and unmet need for medical care among publicly insured adults with disabilities. Med Care. 2013;51:1015–1019.
21. Iezzoni LI, Frakt AB, Pizer SD. Uninsured persons with disability confront substantial barriers to health care services. Disabil Health J. 2011;4:238–244.
22. Ezzati-Rice TM, Rohde F, Greenblatt J. Sample Design of the Medical Expenditure Panel Survey Household Component, 1998-2007. 2008.Rockville, MD: Agency for Healthcare Research and Quality (AHRQ).
23. Cohen JW, Cohen SB, Banthin JS. The Medical Expenditure Panel Survey: a national information resource to support healthcare cost research and inform policy and practice. Med Care. 2009;477 suppl 1S44–S50.
24. Stata [computer program]. Version 12.1. 2011.College Station, TX: StataCorp.
25. Betancourt JR, Corbett J, Bondaryk MR. Addressing disparities and achieving equity: cultural competence, ethics, and health-care transformation. Chest. 2014;145:143–148.
26. Clancy GP, Duffy FD. Going “all in” to transform the Tulsa community’s health and health care workforce. Acad Med. 2013;88:1844–1848.
27. Misky GJ, Carlson T, Thompson E, et al.. Implementation of an acute venous thromboembolism clinical pathway reduces healthcare utilization and mitigates health disparities. J Hosp Med. 2014;9:430–435.
28. Gulley SP, Rasch EK, Chan L. Ongoing coverage for ongoing care: access, utilization, and out-of-pocket spending among uninsured working-aged adults with chronic health care needs. Am J Public Health. 2011;101:368–375.
29. Dejong G, Palsbo SE, Beatty PW, et al.. The organization and financing of health services for persons with disabilities. Milbank Q. 2002;80:261–301.
30. Institute of Medicine. Disability in America: Toward a National Agenda for Prevention. 1991.Washington, DC: National Academy Press.
31. Lee K, Palacio C, Alexandraki I, et al.. Increasing access to health care providers through medical home model may abolish racial disparity in diabetes care: evidence from a cross-sectional study. J Natl Med Assoc. 2011;103:250–256.
32. Pourat N, Lavarreda SA, Snyder S. Patient-Centered Medical Homes Improve Care for Adults With Chronic Conditions. 2013.Los Angeles, CA: UCLA Center for Health Policy Research.
33. Kirby JB, Kaneda T. “Double jeopardy” measure suggests blacks and Hispanics face more severe disparities than previously indicated. Health Aff. 2013;32:1766–1772.
34. Bowling A, Rowe G, McKee M. Patients’ experiences of their healthcare in relation to their expectations and satisfaction: a population survey. J R Soc Med. 2013;106:143–149.
35. Groeneveld PW, Kwoh CK, Mor MK, et al.. Racial differences in expectations of joint replacement surgery outcomes. Arthritis Rheum. 2008;59:730–737.
36. Cheung PT, Wiler JL, Lowe RA, et al.. National study of barriers to timely primary care and emergency department utilization among Medicaid beneficiaries. Ann Emerg Med. 2012;60:4–10.
37. Moy E, Mau MK, Raetzman S, et al.. Ethnic differences in potentially preventable hospitalizations among Asian Americans, Native Hawaiians, and other Pacific Islanders: implications for reducing health care disparities. Ethn Dis. 2013;23Winter6–11.
38. Vargas Bustamante A, Chen J, Rodriguez HP, et al.. Use of preventive care services among Latino subgroups. Am J Prev Med. 2010;38:610–619.
39. Horner-Johnson W, Dobbertin K, Lee JC, et al.. Disparities in healthcare access and receipt of preventive services by disability type: analysis of the Medical Expenditure Panel Survey. Health Serv Res. [Epub ahead of print]. doi 10.1111/1475-6773.12195.