Assessing the difference in racial and ethnic disparities in access to and use of care between Traditional Medicare and Medicare Advantage

Abstract

Objective

Test whether racial‐ethnic disparities in the access and use of care differ between Traditional Medicare (TM) and Medicare Advantage (MA).

Data Source

Secondary data from the 2015–2018 Medicare Current Beneficiary Survey (MCBS).

Study Design

Measure Black‐White and Hispanic‐White disparities in access to care and use of preventive services within TM, within MA, and assess the difference‐in‐disparities between the two programs with and without controls for factors that could influence enrollment, access, and use.

Data Collection/Extraction

Pool 2015–2018 MCBS data and restrict to non‐Hispanic Black, non‐Hispanic White, or Hispanic respondents.

Principal Findings

Black enrollees have worse access to care relative to White enrollees in TM and MA, particularly for cost‐related measures such as not having problems paying medical bills (11–13 pp. lower for Black enrollees; p < 0.05) and satisfaction with out‐of‐pocket costs (5–6 pp. lower; p < 0.05). We find no difference in Black‐White disparities between TM and MA. Hispanic enrollees have worse access to care relative to White enrollees in TM but similar access relative to White enrollees in MA. Hispanic‐White disparities in not delaying care due to cost and not reporting problems paying medical bills are narrower in MA relative to TM by about 4 pp (significant at the p < 0.05 level) each. We find no consistent evidence that Black‐White or Hispanic‐White differences in the use of preventive services differ between TM and MA.

Conclusions

Across the measures of access and use studied here, racial and ethnic disparities in MA are not substantially narrower than in TM for Black and Hispanic enrollees relative to White enrollees. For Black enrollees, this study suggests that system‐wide reforms are required to reduce existing disparities. For Hispanic enrollees, MA does narrow some disparities in access to care relative to White enrollees but, in part, because White enrollees do not do as well in MA as they do in TM.

What is known on this topic

• Medicare Advantage (MA) enrollment growth is greatest among non‐Hispanic Black and Hispanic enrollees, although evidence indicates that racial and ethnic minorities enroll in lower star‐rated plans relative to White enrollees.

• Within MA, studies provide conflicting evidence on the magnitude of racial and ethnic disparities in access to care and preventive care service use.

• Among studies comparing disparities between MA and Traditional Medicare (TM), findings are inconsistent, and some studies group all non‐White enrollees together, which masks challenges facing distinct racial and ethnic groups.

What this study adds

• Black enrollees have poorer access to care relative to White enrollees in both TM and MA, and there is no significant difference in disparities between TM and MA.

• Hispanic‐White disparities in not delaying care due to cost, not reporting problems paying medical bills, and satisfaction with the quality of care are significantly narrower in MA relative to TM.

• There is no systematic evidence that Black‐White or Hispanic‐White differences in the use of preventive services differ between TM and MA.

1. INTRODUCTION

In 2021, about 26 million people were enrolled in a Medicare Advantage (MA) plan, representing about 42 percent of the Medicare‐eligible population. ¹ While the previous decade saw MA enrollment growth across all enrollee groups, the largest growth rates were among Hispanic and non‐Hispanic Black enrollees who have, ² on average, lower incomes and may find MA more attractive than Traditional Medicare (TM) due to MA's supplemental benefits, care‐coordination, out‐of‐pocket spending caps, and often reduced cost‐sharing. ³ , ⁴ , ⁵ Recent research affirms long‐standing racial and ethnic disparities among Medicare enrollees in the access and use of care, ⁶ digital access to telemedicine, ⁷ and drug affordability under the Part D program. ⁸ At the same time, studies evaluating policies intended to improve the overall quality of care delivered to Medicare patients, such as the Hospital Readmissions and Hospital‐Acquired Conditions Reduction Programs, report worse outcomes for Black patients, in part, due to disproportionate penalties imposed on hospitals serving larger shares of Black patients. ⁹ , ¹⁰

Almost all these findings are based on the experiences of enrollees in TM, whose claims data are high‐quality and accessible. But by 2025, the Congressional Budget Office predicts that MA enrollees will represent half of the overall Medicare program, ¹¹ and thus evidence on racial‐ethnic disparities in TM only tells half the story. Efforts to narrow racial and ethnic disparities in health care access, use, and, ultimately, health among the Medicare population necessitate not only a clear assessment of existing disparities within TM and MA but also an assessment of the differences in disparities across these two parts of the programs. This study investigates Black‐White and Hispanic‐White disparities in access to affordable and quality care as well as the use of preventive services (e.g., screenings, tests, and flu shots) among enrollees in both TM and MA. We further explore the difference‐in‐disparities between these two parts of Medicare to assess whether Black or Hispanic enrollees in private Medicare plans are receiving better, worse, or comparable access and preventive service use relative to White enrollees than those in TM.

2. BACKGROUND

TM provides enrollees with access to a wide network of hospitals and physicians, with only about 1% of providers choosing to opt out of providing services to TM enrollees. ¹² MA plans, on the other hand, structure narrower provider networks in their service areas in ways that may leave certain providers out of their network entirely. Thus, enrollees in TM have wider flexibility to receive care from a setting that best fits their needs. Since enrollees have greater flexibility in selecting providers, we might expect that racial and ethnic disparities in TM could be narrower than in MA.

MA plans, however, have several advantages for enrollees relative to TM. Structured as a managed care model (often an HMO), MA plans can provide care coordination that may help patients better connect with the decentralized and fragmented U.S. health care system. In addition, MA plans that bid to offer Part A and B services at rates lower than a pre‐established benchmark based on per capita TM spending in the same service area can use some of the revenue difference between the benchmark and the bid to offer additional benefits not covered in the TM, such as dental, vision, and hearing benefits. MA plans can also create cost‐sharing arrangements for beneficiaries that are more generous than those in TM. Moreover, nearly two‐thirds of MA plans charge no additional premium for coverage, and another 20% of MA plans charge monthly premiums below $50, even though they often provide extra benefits. ⁵ Unlike TM, MA plans also include an out‐of‐pocket maximum. Thus, through enhanced care coordination, additional benefits, reduced cost‐sharing, and low‐to‐zero premiums, MA may make health care services easier and more affordable to access for lower‐income beneficiaries. Since non‐White Medicare enrollees have, on average, lower incomes than White enrollees, ⁴ these advantages of MA plans could generate narrower racial and ethnic disparities in measures of access and use than in TM. TM and MA each have distinct strengths and weaknesses for different beneficiaries, and it is not clear how these strengths and weaknesses will affect racial and ethnic disparities in access to and use of health care.

Several studies have examined racial and ethnic disparities in health care quality and access within MA. Using the 2011 Healthcare Effectiveness Data and Information Set (HEDIS) and Medicare Master Beneficiary Summary File (MBSF), Rivera‐Hernandez and colleagues found that Hispanic enrollees performed sometimes better and sometimes worse than White enrollees, depending on the outcome measure. ¹³ Similarly, analysis of the Medicare Health Outcome Survey before and after the 2012 implementation of the MA Quality Bonus Payment Demonstration revealed a mixture of better and worse performance on the likelihood of receiving preventive care services depending on race and ethnicity (Black, Asian, and Hispanic) and service use. ¹⁴ Findings varied within MA with respect to behavioral health quality as well; Black and Hispanic MA enrollees performed worse than White enrollees on antidepressant medication management, while these same groups performed better (Hispanic) and worse (Black) on follow‐up visits after hospitalization for mental illness. ¹⁵ Notably, Weech‐Maldonado et al. not only corroborates previous studies documenting racial and ethnic disparities in health care quality in MA using 2008 and 2009 Consumer Assessment of Healthcare Providers and Systems (CAHPS) and HEDIS measures but also find that measures of racial and ethnic disparities across outcomes within a plan are moderately correlated with one another. ¹⁶ Some of the observed racial and ethnic disparities in MA may be driven by differences in the quality of plans beneficiaries select. Park, Werner, and Coe show that while Black, Hispanic, Asian, and Pacific Islander enrollees tend to enroll in lower‐quality star‐rated plans relative to White enrollees; these enrollment differences are largely explained by differences in county‐level MA plan offerings. ¹⁷

Fewer studies have compared disparities in quality between MA and TM. Moreover, within the small literature that has examined this question, the findings have been inconsistent. A study using 2007 CAHPS data found larger negative disparities between Black and White beneficiaries in MA relative to TM for four out of eleven patient experience measures, ¹⁸ while another found greater negative disparities in mammography receipt in TM than MA for Black and Hispanic women. ¹⁹ Two studies investigated hospital readmission rates, one focusing on patients discharged to skilled nursing facilities and another examining overall readmission rates in New York following one of six major surgeries. The former found no significant differences in racial disparities between MA and TM for these readmission rates, while the latter found significant differences in disparities across the two programs, finding greater positive differences between Black and White patients in MA relative to TM. ²⁰ , ²¹ Finally, Johnston et al. used 2015–2018 data from the Medicare Current Beneficiary Survey to investigate racial and ethnic differences, by program type, in measures related to ambulatory care access and quality. ²² The authors find that, among all beneficiaries, MA enrollees reported significantly higher rates of influenza vaccinations, pneumonia vaccinations, and colon cancer screenings. Non‐white beneficiaries, as a group, had significantly better rates of access to a primary care physician, influenza vaccinations, pneumonia vaccinations, and colon cancer screenings, in MA relative to non‐White beneficiaries in TM. However, across both program types, non‐White enrollees had significantly lower rates of primary care physician access, specialist visits, influenza vaccinations, and pneumonia vaccinations relative to White enrollees. The authors find no significant difference‐in‐disparities between TM and MA for these selected measures.

This paper expands on the limited literature by investigating differences between TM and MA in racial and ethnic disparities by going beyond Johnston et al., ²² disaggregating non‐White enrollees into subgroups, and assessing differences in Black‐White and Hispanic‐White disparities. We consider measures not previously explored related to access to care, satisfaction with care, and use of preventive services. We also provide new evidence on racial and ethnic disparities between TM and MA in problems paying medical bills and whether respondents delayed care due to cost. By examining Black‐White and Hispanic‐White disparities separately, we identify unique challenges facing each group. Finally, given how few studies have examined this research question, this study contributes toward developing a scientific consensus on the differences‐in‐disparities experienced by Medicare enrollees in TM and MA.

3. DATA AND METHODS

We use data from the 2015–2018 Medicare Current Beneficiaries Survey (MCBS) to assess racial/ethnic differences in health care access and use in Traditional Medicare (TM) relative to Medicare Advantage (MA) during a 4‐year pooled time period. The MCBS is a national survey of Medicare enrollees that provides information on demographic characteristics, health care access and use, health status, supplementary coverage, and administratively reported cost and use data for medical services. We pool 4 years of MCBS data to increase sample sizes and to enable sufficiently powered statistical comparisons of disparities estimates across racial and ethnic groups and by Medicare program type. Importantly, both TM and MA enrollees are interviewed in the MCBS. MCBS respondents self‐report their racial and ethnic backgrounds; throughout this study, we restrict our analysis to respondents who identify as non‐Hispanic White, non‐Hispanic Black, or Hispanic. We investigate Black‐White and Hispanic‐White disparities and do not evaluate disparities for other racial and ethnic groups. The MCBS includes far fewer respondents in other racial and ethnic categories by program type. This would put us at risk of failing to observe significant differences in outcomes for these groups relative to White enrollees where one exists, or, where statistically significant differences are observed, inappropriately inflating the estimated disparities. Throughout this analysis, we exclude respondents residing in Puerto Rico (where a very high share of enrollees select MA). The 4‐year pooled sample contains 59,278 observations; 55,278 observations remain after restricting to non‐Hispanic White, non‐Hispanic Black, or Hispanic respondents alone; 55,206 observations remain after excluding respondents residing in Puerto Rico. However, sample sizes vary across outcomes based on question availability by years, item non‐response, and differences in skip patterns (e.g., by gender). We report our final sample sizes for each outcome and by each racial and ethnic group and program type in Appendix Table A3.

We use the MCBS to first establish racial and ethnic disparities in health care access and use among TM enrollees and, separately, among MA enrollees. We then investigate whether racial and ethnic disparities differ between MA and TM. We examine the following outcomes related to health care access: whether respondents identify having trouble getting care and whether that trouble was due to cost; whether they delayed care due to cost; whether they had problems paying medical bills; whether they were dissatisfied with the quality of medical care they received over the past year; and, whether they were dissatisfied with out‐of‐pocket costs of medical care over the past year.

For utilization measures, we examine whether respondents received the following screenings or preventive services in the past year: mammograms, pap smears, blood tests for prostate cancer, flu shots, blood cholesterol tests, or diabetes tests. These utilization measures are not recommended annually for all patients. For example, U.S. Preventive Services Task Force (USPSTF) guidelines suggest pap smears among women above age 65 are not necessary if a series of prior tests were normal or if the patient is not at high risk for cervical cancer. Though USPSTF guidelines do not differentiate recommendations based on patient race and ethnicity, rates of screening recommendations could differ by patient race and ethnicity if the patient background is associated with factors that raise the probability of recommendation.

We compare Black‐White and Hispanic‐White disparities in health care access and use. We use t‐tests to establish whether unadjusted differences in these disparities significantly differ between TM and MA. We emphasize that these differences‐in‐disparities are evaluated during a single pooled time period (2015–2018); testing whether these differences have widened or narrowed over time remains an important area of future research. We further assess whether estimates of these difference‐in‐disparities are associated with factors that could affect TM or MA enrollment decisions as well as health care access and use. We examine how age, gender, dual Medicare‐Medicaid enrollment status, health status based on fair or poor self‐reported health and the number of chronic conditions, income, education, census division, and county‐level MA penetration rates (for each analysis year from CMS) influence our estimates of disparities. Comparing estimates of the unadjusted difference‐in‐disparities to these adjusted ones provides evidence on whether immediate factors related to selection are likely to meaningfully bias our findings. Summary statistics of these controls by TM or MA enrollment and by enrollee race and ethnicity are presented in Appendix Table A1. To assess the role these factors may play in determining overall difference‐in‐disparities, we estimate multivariate linear probability models. To improve sample representativeness and help adjust for potential heteroskedasticity in errors, all models use weighted least‐squares methods using MCBS ever‐enrolled weights.

4. RESULTS

Table 1 presents averages of the selected outcome measures separately for enrollees with TM and MA. There are 30,536 TM respondents and 17,636 MA respondents in the pooled 2015–2018 MCBS dataset, however, the number of respondents varies based on outcome measure (sample sizes by the outcome are presented in Appendix Table A3). TM enrollees were significantly more likely to report not having trouble getting care (93.3%), not having to delay care due to cost (90.3%), and not having problems paying medical bills (90.4%) relative to MA enrollees (92.0%, 88.2%, and 87.5%, respectively). More than 95% of TM and MA enrollees reported being satisfied with the quality of medical care received over the past year. However, a slightly smaller share (83%) of TM and MA enrollees were satisfied with out‐of‐pocket costs incurred for that medical care in the past year.

TABLE 1.

Differences in access to health care between Traditional Medicare and Medicare Advantage enrollees, 2015–2018.

	Traditional Medicare	Medicare Advantage
Access measures
Respondents without trouble getting care	93.3%*	92.0%
Respondents without trouble getting care due to cost	98.2%*	98.0%
Respondents without a delay in care due to cost	90.3%*	88.2%
No problem paying medical bills	90.4%*	87.5%
Satisfaction with quality of medical care over past year	95.7%	95.5%
Satisfaction with out‐of‐pocket costs of medical care over past year	83.2%	83.0%
Utilization measures
Received mammogram or breast x‐ray in prior year	43.3%*	45.7%
Received pap smear test in prior year	16.9%*	15.9%
Received blood test for detection of prostate cancer in prior year	56.1%	57.1%
Received flu shot for current flu season	69.9%*	72.4%
Received blood cholesterol test in prior year	86.8%*	89.9%
Received blood test for diabetes in prior year	70.7%*	73.9%
Number of observations	30,536	17,636

Note: All estimates are weighted using MCBS ever‐enrolled weights. * indicates TM and MA differences are statistically significant at the p < 0.05 level. Analysis was restricted to non‐Hispanic White, non‐Hispanic Black, or Hispanic respondents. Analysis excludes respondents residing in Puerto Rico.

Source: 2015–2018 Medicare current beneficiary survey.

Female respondents with MA were more likely to have had a mammogram but less likely to have had a pap smear in the past year relative to those with TM. Rates of blood tests for detecting prostate cancer were similar between MA and TM male enrollees. However, for all other utilization measures, MA enrollees were significantly more likely to have had the screening or preventive service relative to TM enrollees. Relative to TM enrollees, MA enrollees were about 2.5% points more likely to receive a flu shot for the current flu season, 3.1% points more likely to receive a blood cholesterol test, and 3.2% points more likely to receive a blood test for diabetes. In Appendix Table A2, we further present the share of enrollees with MA coverage and averages of access and use outcomes separately by race and ethnicity. Notably, in our analytic sample, just under 1 in 3 non‐Hispanic White enrollees enrolled in MA, 43% of Black enrollees were enrolled in MA, and half of the Hispanic enrollees selected MA coverage.

Table 2 examines differences in access to affordable and quality care between Black and White enrollees within TM and within MA. A negative estimate of these differences indicates worse access for Black enrollees relative to White enrollees. Black enrollees report worse access to care relative to White enrollees in both TM and MA. For all six access measures, differences are negative, large, and statistically significant in both programs. Relative to White enrollees in TM, Black enrollees are 4.6% points less likely to report not having a delay in care due to cost; 11.4% points less likely to report not having a problem paying medical bills; and, 6.3% points less likely to report being satisfied with out‐of‐pocket costs associated with medical care. In MA, these Black‐White differences are −4.4% points, −13.0% points, and −5.1% points, respectively.

TABLE 2.

Black‐White differences in access to care in Traditional Medicare and Medicare Advantage.

	Traditional Medicare			Medicare Advantage			Traditional Medicare versus Medicare Advantage (TM–MA difference)
	Non‐Hispanic Black (%)	Non‐Hispanic White (%)	Difference (% points)	Non‐Hispanic Black (%)	Non‐Hispanic White (%)	Difference (% points)	Difference in Black‐White Disparity (% points)	Adjusted Difference in Black‐White Disparity (% points)
Respondents had no trouble getting care	91.6	93.7	−2.1*	90.6	92.5	−1.9*	−0.2	0.5
Respondents had no trouble getting care due to cost	97.0	98.4	−1.4*	96.4	98.3	−1.9*	0.5	0.7
Respondents had no delay in care due to cost	86.4	91.0	−4.6*	84.4	88.8	−4.4*	−0.2	0.6
No problem paying medical bills	80.4	91.8	−11.4*	76.5	89.4	−13.0*	1.6	2.4*
Satisfied with quality of medical care	94.4	96.0	−1.6*	94.7	95.7	−1.0*	−0.5	−0.1
Satisfied with out‐of‐pocket costs of medical care	77.6	83.9	−6.2*	78.6	83.7	−5.1*	−1.1	0.0

Source: 2015–2018 Medicare Current Beneficiary Survey. * indicates that the estimated difference is statistically significant at the p < 0.05 level. Adjusted difference includes controls for Age, gender, dual Medicare and Medicaid enrollment, income, and education. Analysis was restricted to non‐Hispanic White, non‐Hispanic Black, or Hispanic respondents. Analysis excludes respondents residing in Puerto Rico. See Table A3 for the number of observations in each race/ethnicity‐by‐coverage group for each outcome. Adjusted differences include indicators controlling for age, gender, income, education, Census division, fair or poor self‐reported health, the number of chronic conditions, and county‐level MA penetration.

The final two columns of Table 3 assess the differences‐in‐disparity between the two programs. For these estimates, a positive difference indicates that Black‐White disparities in TM are smaller than those in MA. Because Black‐White disparities in access to affordable and quality care are similar in magnitude in both TM and MA, we found no significant differences in Black‐White disparities between the two programs. That is, Black enrollees have uniformly worse access to care relative to White enrollees irrespective of which part of the Medicare program they enroll in. In the final column of Table 3, we assess adjusted differences in disparity measures between the two parts of the program, adjusting for demographic, socioeconomic, and geographic factors, and find qualitatively similar results as for the unadjusted differences, with just one of 6 estimates statistically significant.

TABLE 3.

Hispanic‐White differences in access to care in Traditional Medicare and Medicare Advantage.

	Traditional Medicare			Medicare Advantage			Traditional Medicare versus Medicare Advantage (TM–MA difference)
	Hispanic (%)	Non‐Hispanic White (%)	Difference (% points)	Hispanic (%)	Non‐Hispanic White (%)	Difference (% points)	Difference in Hispanic‐White Disparity (% points)	Adjusted Difference in Hispanic‐White Disparity (% points)
Respondents had no trouble getting care	90.3	93.7	−3.4*	90.2	92.5	−2.3*	−1.1	−0.2
Respondents had no trouble getting care due to cost	97.3	98.4	−1.1*	97.6	98.3	−0.7	−0.4	−0.1
Respondents had no delay in care due to cost	87.1	91.0	−3.9*	88.6	88.8	−0.2	−3.7*	−2.5*
No problem paying medical bills	86.2	91.8	−5.6*	88.1	89.4	−1.3	−4.3*	−3.1*
Satisfied with quality of medical care	93.6	96.0	−2.4*	94.9	95.7	−0.8	−1.6*	−1.3
Satisfied with out‐of‐pocket costs of medical care	81.9	83.9	−2*	83.6	83.7	−0.1	−1.9	−1.0

Source: 2015–2018 Medicare Current Beneficiary Survey. * indicates that estimated difference is statistically significant at the p < 0.05 level. Adjusted difference includes controls for Age, gender, dual Medicare and Medicaid enrollment, income, and education. Analysis was restricted to non‐Hispanic White, non‐Hispanic Black, or Hispanic respondents. Analysis excludes respondents residing in Puerto Rico. Adjusted differences include indicators controlling for age, gender, income, education, Census division, fair or poor self‐reported health, the number of chronic conditions, and county‐level MA penetration.

Table 3 shows similar analyses of disparities in access to affordable and quality care between Hispanic enrollees and White enrollees. We find that in TM, access to care is all worse for Hispanic enrollees relative to White enrollees, and these differences are all statistically significant. In MA, however, we observe that just 1 of 6 of the access measures (likelihood of reporting not having trouble getting care) is significantly lower for Hispanic enrollees relative to White enrollees. The rest of the estimated differences between Hispanic and White enrollees in MA are small in magnitude and not statistically different from zero.

When examining access and quality across both programs, we find that 3 of 6 significant difference‐in‐disparity estimates are negative, indicating narrower Hispanic‐White disparities in MA relative to TM. We find that Hispanic‐White disparities in the likelihood of not reporting a delay in care due to cost, not reporting problems paying medical bills, and being satisfied with the quality of care were all significantly narrower in MA relative to TM. For each of these three outcomes, we find that Hispanic enrollees had slightly higher reported access in MA relative to TM and that White enrollees had slightly lower access in MA relative to TM. The combination of these changes led to the narrowing of the Hispanic‐White disparity in MA relative to TM. Adjusted differences in the final column of Table 3 produce qualitatively similar findings, indicating that controlling for potentially related factors has little impact on the estimated difference‐in‐disparity.

Table 4 assesses Black‐White differences in the utilization of preventive screenings, tests, and flu shots within and across TM and MA. Black women use mammograms or breast x‐rays at higher rates than White women in both TM and MA (2.6% and 5.2% points, respectively), although this difference is only significant in the MA program. However, the last two columns in Table 4 show that the difference in these Black‐White rates of mammogram use in TM and MA is not significant. For two pap smear tests and blood cholesterol tests, Black enrollees have higher rates of use relative to White enrollees in both TM and MA. In fact, these Black‐White differences in TM and MA are similar in magnitude and as a result, there is no significant difference in these estimates between TM and MA.

TABLE 4.

Black‐White differences utilization measures in Traditional Medicare and Medicare Advantage.

	Traditional Medicare			Medicare Advantage			Traditional Medicare versus Medicare Advantage (TM–MA difference)
	Non‐Hispanic Black (%)	Non‐Hispanic White (%)	Difference (% points)	Non‐Hispanic Black (%)	Non‐Hispanic White (%)	Difference (% points)	Difference in Black‐White Disparity (% points)	Adjusted Difference in Black‐White Disparity (% points)
Received mammogram or breast x‐ray in prior year	45.6	43.0	2.6	49.5	44.3	5.2*	−2.6	−3.0
Received pap smear test in prior year	26.2	15.1	11.1*	22.7	13.5	9.2*	1.9	1.4
Received blood test for detection of prostate cancer in prior year	49.6	57.3	−7.7*	50.9	58.4	−7.5*	−0.2	1.8
Received flu shot for current flu season	56.0	71.4	−15.4*	62.9	74.6	−11.7*	−3.7*	−2.5
Received blood cholesterol test in prior year	88.5	86.6	1.9*	92.2	89.3	2.9*	−1.0	−0.9
Received blood test for diabetes in prior year	62.8	71.4	−8.6*	70.1	73.4	−3.3	−5.3*	−4.8*

Source: 2015–2018 Medicare Current Beneficiary Survey. * indicates that estimated difference is statistically significant at the p < 0.05 level. Adjusted difference includes controls for Age, gender, dual Medicare and Medicaid enrollment, income, and education. Analysis was restricted to non‐Hispanic White, non‐Hispanic Black, or Hispanic respondents. Analysis excludes respondents residing in Puerto Rico. See Table A3 for the number of observations in each race/ethnicity‐by‐coverage group for each outcome. Adjusted differences include indicators controlling for age, gender, income, education, Census division, fair or poor self‐reported health, the number of chronic conditions, and county‐level MA penetration.

Table 4 shows that Black men have much lower rates of blood tests used to detect prostate cancer relative to White enrollees in both TM and MA (−7.7% and −7.5% points, respectively). Again, since these disparities are similar, we observe no statistically significant difference in the measured disparity between the two parts of the Medicare program; that is, Black male enrollees are equally disadvantaged in both TM and MA. The likelihood of receiving a flu shot or a blood test for diabetes was lower for Black enrollees relative to White enrollees in TM, but this difference narrows in magnitude in MA. Analysis of the unadjusted difference‐in‐disparity suggests that these reductions in the Black‐White disparity in MA relative to TM are significant, although the differences are attenuated in our adjusted models. Thus, some of the difference‐in‐disparity may be explained by differences in the characteristics of TM and MA enrollees.

We assess Hispanic‐White differences in utilization in TM and MA in Table 5. In TM, rates of mammograms or breast x‐ray exams were similar between Hispanic and non‐Hispanic White enrollees. In MA, however, Hispanic women were 5.4% points more likely to receive these screenings relative to White women. The difference in these estimates between TM and MA is statistically significant (−4.8% points), and adjusted differences are similar in magnitude and significance (−5.2% points), suggesting that potentially related factors have minimal influence. Hispanic women had higher rates of pap smears in both TM and MA relative to non‐Hispanic White women (10.7% and 8.5% points, respectively). Given the similarity in the magnitude of these estimates in TM and MA, the estimated Hispanic‐White difference in use rates between the two programs is small and insignificant.

TABLE 5.

Hispanic‐White differences utilization measures in Traditional Medicare and Medicare Advantage.

	Traditional Medicare			Medicare Advantage			Traditional Medicare versus Medicare Advantage (TM–MA difference)
	Hispanic (%)	Non‐Hispanic White (%)	Difference (% points)	Hispanic (%)	Non‐Hispanic White (%)	Difference (% points)	Difference in Hispanic‐White Disparity (% points)	Adjusted Difference in Hispanic‐White Disparity (% points)
Received mammogram or breast x‐ray in prior year	43.6	43.0	0.6	49.7	44.3	5.4*	−4.8*	−5.1*
Received pap smear test in prior year	25.8	15.1	10.7*	22.0	13.5	8.5*	2.2	0.8
Received blood test for detection of prostate cancer in prior year	49.8	57.3	−7.5*	54.8	58.4	−3.6	−3.9	−3.8
Received flu shot for current flu season	69.6	71.4	−1.8	69.0	74.6	−5.6*	3.8*	4.8*
Received blood cholesterol test in prior year	86.5	86.6	−0.1	91.0	89.3	1.7*	−1.8	−1.2
Received blood test for diabetes in prior year	70.9	71.4	−0.5	80.8	73.4	7.4*	−7.9*	−7.2*

Source: 2015–2018 Medicare Current Beneficiary Survey. * indicates that estimated difference is statistically significant at the p < 0.05 level. Adjusted difference includes controls for Age, gender, dual Medicare and Medicaid enrollment, income, and education. Analysis was restricted to non‐Hispanic White, non‐Hispanic Black, or Hispanic respondents. Analysis excludes respondents residing in Puerto Rico. See Table A3 for the number of observations in each race/ethnicity‐by‐coverage group for each outcome. Adjusted differences include indicators controlling for age, gender, income, education, Census division, fair or poor self‐reported health, the number of chronic conditions, and county‐level MA penetration.

Table 5 also shows that rates of blood tests for the detection of prostate cancer were significantly lower for Hispanic enrollees relative to White enrollees in TM (a −7.5% point difference). This Hispanic‐White difference was smaller in magnitude and insignificant in MA. However, the −3.9 difference‐in‐disparity for this measure between TM and MA was not statistically significant. In TM, Hispanic and White enrollees had similar rates of using flu shots. In MA, on the other hand, Hispanic enrollees were 5.6% points less likely than White enrollees to receive a flu shot. The difference‐in‐disparity is positive and significant, indicating that relative to White enrollees, Hispanic enrollees had greater use of flu shots in TM compared to MA. However, we note that the difference in this disparity is driven not by an increase in the rates of flu shots among Hispanic enrollees in TM relative to MA (rather, these estimates are very similar at 69.6% and 69.0%) but by an increase in the likelihood of flu shot receipt among White enrollees in MA relative to TM. Rates of blood cholesterol tests and blood tests for diabetes were statistically similar between Hispanic and White enrollees in TM. For both outcomes, Hispanic enrollees had statistically greater rates of use in MA relative to White enrollees; however, the difference in these estimates between TM and MA is only significant for blood tests for diabetes.

5. DISCUSSION

In this study, we show that Black enrollees have markedly poorer access to care relative to White enrollees on all selected measures in both TM and MA and these estimated differences are similar in magnitude. As a result, there is little difference in the estimated Black‐White disparity between the two parts of the Medicare program. This evidence is consistent with Johnston et al. ^22's assessment of ambulatory care access and quality for Black and Hispanic enrollees and presents one potential mechanism of documented findings of worse downstream health outcomes for Black Medicare enrollees relative to White enrollees in both TM and MA. ²¹ Hispanic enrollees, on the other hand, have worse access to care relative to White enrollees in TM but have similar rates of access for most measures relative to White enrollees in MA. This provides some evidence that Hispanic‐White disparities in access to care are narrower in MA relative to TM. There was no discernable pattern for differences in Hispanic‐White disparities in preventive service use between TM and MA; Hispanic enrollees with MA had higher rates of use for some measures (use of mammograms and blood tests for diabetes) and lower rates of use for other measures (use of flu shots) relative to White enrollees, and these differences were significantly different than the experiences of Hispanic and non‐Hispanic White enrollees in TM.

Based on these findings, it does not appear that the increased growth in MA enrollment among Black enrollees relative to White enrollees has resulted in reduced disparities in access to and use of health care services. There may be other reasons for these enrollment patterns related to the broader benefits and lower cost sharing in MA plans compared to TM, but the potential for better care coordination has not yet resulted in substantial changes in disparities in satisfaction with financial burdens, care quality, or the use of the preventive services examined here, especially among Black enrollees. It is possible that the quality of MA plans available to Black enrollees is not as high as those available to White enrollees (as has been documented elsewhere) or that MA plans have limited incentives to reduce disparities in care relative to the broader health care system. CMS is currently exploring options that may change those incentives by incorporating measures of health equity into MA's quality star ratings program. ²³

For Black enrollees, system‐wide reforms are necessary to reduce disparities in health care access. Where our evidence indicates Black enrollees face large disparities in access and affordability measures relative to White enrollees in both TM and MA, this was especially true for cost‐related measures such as the likelihood of reporting not having problems paying medical bills (11%–13% points lower for Black enrollees relative to White enrollees) and respondent satisfaction with out‐of‐pocket costs for medical care (5%–6% points lower). One of the potential drivers of cost‐related disparities is the underlying differences in household income between Black and White Medicare enrollees (highlighted in Table A1). Although the existing Medicare Savings Programs and Low‐Income Subsidy Programs—designed to help low‐income beneficiaries ‐ may be effective at reducing income‐based disparities in beneficiary financial burdens, these programs have very low take‐up rates. ²⁴ At the same time, resource tests in most states prevent many low‐income beneficiaries from becoming eligible for these programs as well. ²⁵ Thus, some approaches to closing income‐based differences in access to care in both TM and MA are to expand eligibility to these programs by potentially reducing or eliminating asset tests and to expand beneficiary participation in these programs. Whether closing inequities in access and use by socioeconomic status translates into narrower disparities by beneficiary race and ethnicity remains an important area of future research.

While Hispanic enrollees reported similar access to care relative to White enrollees in MA but not in TM, it is important to note that the difference‐in‐disparities is attributed to a combination of better access measures among Hispanic MA enrollees (relative to Hispanic enrollees in TM) and worse access among non‐Hispanic White enrollees in MA (relative to White enrollees in TM). It remains unclear in our analysis why one population benefits markedly in TM, relative to MA, while another group fares worse. Results from our regression‐adjusted differences rule out first‐order selection explanations based on demographics, socioeconomic status, education, and geography. This study does not rule out the possibility that the strengths and weaknesses of TM and MA discussed earlier (e.g., wider provider network in TM, MA care coordination, and the expanded set of services often covered in many MA plans), may have heterogenous effects on beneficiaries in different race and ethnicity groups. Further research is needed to isolate the contribution of individual components of these two parts of the Medicare program to help determine the best path forward to promoting equitable access to affordable care. Finally, assessing differences in access and use experienced among TM enrollees with and without supplementary coverage (e.g., Medigap), or among MA enrollees enrolled in plans scored as high or low quality may also highlight potential differences‐in‐disparities within each part of the Medicare program and may also serve to highlight the mechanisms driving differences‐in‐disparities across the two parts of the program.

This study has several limitations. First, we cannot fully address selection in our research design, and there may be unobserved factors that affect both the decision to enroll in an MA or TM and our measures of access to and use of care. However, we believe that the role of such unobserved factors is minimal in this context since unadjusted difference‐in‐disparities measures are qualitatively similar to our adjusted estimates after accounting for a wide range of observed factors. Second, this study concentrates on differences‐in‐disparities estimates at a single pooled time period but does not examine whether racial‐ethnic disparities and differences‐in‐disparities in access to and use of care have narrowed over time, that is, possibly in response to growth in MA enrollment in recent years. Finally, there is substantial heterogeneity in MA plans in terms of plan comprehensiveness, supplemental benefits, and quality that are not accounted for in this analysis. Similarly, there are potential differences between TM enrollees with and without private supplemental coverage, such as Medigap, that could be meaningful drivers of differences‐in‐disparities between MA and TM. We emphasize these limitations of our study represent important areas of future research.

FUNDING INFORMATION

The authors report financial support from Arnold Ventures.

Supporting information

Appendix S1. Supporting information.

Gangopadhyaya A, Zuckerman S, Rao N. Assessing the difference in racial and ethnic disparities in access to and use of care between Traditional Medicare and Medicare Advantage. Health Serv Res. 2023;58(4):914‐923. doi: 10.1111/1475-6773.14150