Health Outcomes of Dually Eligible Beneficiaries Under Different Medicare Payment Arrangements

ABSTRACT

Objectives: Dually eligible beneficiaries (hereafter, Duals) qualify for Medicare and Medicaid due to low income and/or disability. Duals comprise 19% of Medicare beneficiaries but consume 35% of Medicare spending. Identifying high-quality, efficient care arrangements may improve outcomes and reduce costs for Duals. This study evaluated the effect of different payment arrangements on Duals’ quality and efficiency outcomes.

Study Design: Retrospective, cross-sectional analysis using CMS data and health plan contract information from 17 participating physician groups (n = 15,488 primary care physicians).

Methods: We identified Duals within the same physician groups treated under at-risk Medicare Advantage (MA), traditional Medicare (TM), and fee-for-service (FSS) MA payment arrangements. We then compared the 3 cohorts across 20 health outcome metrics for the 2016-2019 period.

Results: The sample comprised 1,980,691 person-years (at-risk MA, 15.4%; TM, 48.3%; and FFS MA, 36.4%). Duals in at-risk MA had better outcomes in 17 of 20 measures compared with TM, with avoidable hospital and emergency department (ED) measures showing 9.0% to 32.7% higher quality and efficiency. Compared with FFS MA, at-risk MA had better outcomes in 18 of 20 measures, with avoidable hospital and ED measures showing 7.7% to 15.3% higher quality and efficiency. FFS MA had better outcomes than TM for 17 of 20 measures; 1 measure favored TM.

Conclusions: At-risk MA was associated with higher quality and lower health resource utilization for Duals compared with TM and FFS MA. The CMS goal of accountable care arrangements should include at-risk MA for Duals due to these beneficiaries’ increasing health care utilization and costs.

Am J Manag Care. 2026;32(5):In Press

Takeaway Points

• In this retrospective cohort analysis of dually eligible beneficiaries, at-risk Medicare Advantage (MA) was associated with higher quality and avoidable health care utilization compared with both fee-for-service (FFS) MA and traditional Medicare (TM).
• Within the MA Dual Eligible Special Needs Plan (D-SNP) population, quality was also higher and avoidable health care utilization lower in at-risk MA compared with FFS MA.
• The greatest differences were seen when comparing at-risk MA D-SNP with TM.

In 2024, 12.8 million dual-eligible beneficiaries (hereafter, Duals) were enrolled in both Medicare and Medicaid.¹ Duals may participate in traditional Medicare (TM), or they may enroll in Medicare Advantage (MA), including Dual Eligible Special Needs Plans (D-SNPs). Care costs for Duals are rising faster than for other Medicare beneficiaries. Duals totaled 19% of the Medicare population in 2024 but accounted for 35% of Medicare spending.² Thus, there is strong interest in studying both the quality and efficiency of their care.

There is a paucity of data comparing MA with TM with respect to health outcomes in the Duals population. A key consideration is that MA contracts most commonly pay providers via fee-for-service (FFS) arrangements; however, many plans contract with physician groups under delegated 2-sided risk arrangements where the financial risk of providing health care is transferred to the group (at-risk MA). Most TM arrangements are FFS; in 2019, the measurement year of this study, only 8% of Medicare accountable care organizations (ACOs) were accepting some 2-sided risk in the Medicare Shared Savings Program (MSSP).³

Increasing evidence indicates that at-risk MA is associated with improvements in quality and health resource utilization, suggesting better value-based care outcomes relative to both FFS MA and TM.^4,5 However, no prior studies have examined these outcomes specifically in the Duals population. Here, we extend to the Duals population our prior research on health outcomes as a function of Medicare payment arrangements.^4-6 We compared 20 measures of quality and efficiency in 3 cohorts of Duals (at-risk MA, TM, or FFS MA), all cared for by the same physicians within the same physician groups.

METHODS

Study Design

This cross-sectional study examined the association of different Medicare payment arrangements with quality and efficiency in the Duals population treated by the same physician groups. All groups participated with MA in full risk or full professional risk contracts.

This study, approved by Solutions IRB, involved a retrospective analysis of preexisting deidentified data, qualifying as non–human subjects research and exempt from further review. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines.

Study Data

The study utilized deidentified TM claims and MA encounters from 2016 to 2019 housed in the CMS Virtual Research Data Center (VRDC). The National Provider Identifiers (NPIs) from a nonpublic data set of participating physician groups that provided contract data were linked to the NPIs of physicians who cared for Duals in the VRDC. Claims and encounter data associated with health resource utilization were available for all 3 Medicare groups.

Sample and Cohorts

Cohen et al detailed the methodology used to create the sample and cohort.⁴ The cohort was built using data from 2016 to 2019 to avoid health care disruptions experienced during the COVID-19 pandemic. We restricted beneficiary-year combinations to individuals who were enrolled in Medicare Parts A and B for all 12 months of those years and were dually eligible. Dual status was defined as being enrolled for at least 1 month in a calendar year with full or partial eligibility for both Medicare and Medicaid benefits and included beneficiaries younger than and at least 65 years. We removed beneficiaries who switched from MA to TM or from TM to MA during a calendar year and limited the sample to beneficiaries who successfully aligned to primary care physician (PCP) NPI flags for participating physician groups. We removed beneficiaries with missing Hierarchical Condition Category (HCC) scores. Three distinct dual-eligible cohorts were constructed for each calendar year: at-risk MA, TM, and FFS MA (eAppendix Table 1 and eAppendix Figure [eAppendix available at ajmc.com]). The final sample was 17 provider groups, comprising 15,488 PCPs (eAppendix Table 2).

Outcomes

We selected 20 quality and efficiency measures. Outcomes were defined at an individual claim or encounter level and aggregated to a person-year level. Hospital care was reflected by acute inpatient admission rates and 30-day all-cause readmission rates. We tracked emergency department (ED) visit volume and granular measures for ED utilization. We tracked avoidance of disease-specific admissions based on the Agency for Healthcare Research and Quality Prevention Quality Indicator (PQI) definitions, including avoidable admissions for acute and/or chronic complications for diabetes, chronic obstructive pulmonary disease, hypertension, heart failure, bacterial pneumonia, and urinary tract infections (eAppendix Table 3). Outpatient quality was tracked via pharmacy measures of high-risk medication use and medication adherence using the Healthcare Effectiveness Data and Information Set and via office visit volume.

Statistical Analysis

Cohort characteristics were described for the overall sample and by payment arrangement (Table 1). Continuous variables were summarized as means and SDs and categorical variables as frequencies and percentages. We summarized unadjusted outcome measures as mean event rates per 1000 persons (eAppendix Table 4). Using multivariable logistic regressions, we modeled outcomes as binary indicators given the relatively low odds and relevance of 0 values to compare Duals in at-risk MA, TM, and FFS MA cohorts. To mitigate potential confounding from patient-mix differences, all outcome models were adjusted for age, sex, self-reported race and ethnicity (based on Research Triangle Institute race code), composite HCC version 24 risk adjustment factor score, and prevalence of high-level disease categories based on high-level HCC groupings (Table 1). Calendar year was adjusted as fixed effects in all regression models. We included an indicator for the provider group of the attributed PCP to account for potential confounding from provider differences within a specific payment arrangement. Associations are reported as change in average marginal risk (AMR) or equivalent marginal risk differences (MRDs) (Figure 1 and Figure 2).⁷ Statistical analyses were performed from June 1, 2024, to January 29, 2025, and were conducted using SAS Enterprise Guide 7.15 (SAS Institute Inc). To account for multiple comparisons, we applied the Holm-Bonferroni correction to control the family-wise error rate, adjusting P values in a stepwise manner to allow rigorous control over type I error while minimizing loss of statistical power.⁸ A 2-sided P less than .05 by Wald χ² indicated significance for regression estimates. Results are reported as MRDs.

Subgroup and Sensitivity Analyses

We also assessed differences in effect size across Duals in MA with and without D-SNP enrollment (eAppendix Table 5). In a separate analysis, we compared D-SNP and non–D-SNP subgroups with the TM group (eAppendix Table 6). As a robustness test to assess the sensitivity of associations to coding intensity, we ran models adjusting for HCC version 28 scores and groupings in place of those using version 24 (eAppendix Table 7).

RESULTS

The final study cohort comprised 1,980,691 person-years: 15.4% were in at-risk MA, 48.3% were in TM, and 36.4% were in FFS MA. The mean ages were 70.7, 68.3, and 69.4,respectively. In the at-risk MA, TM, and FFS MA groups, women comprised 61.6%, 61.4%, and 63.0% of each group, respectively, and non-Hispanic White beneficiaries constituted 26.3%, 37.3%, and 28.4%. The Pacific region had the greatest number of beneficiaries in the sample, with 39.7%, 53.4%, and 29.3%, respectively. The mean HCC version 24 scores were 1.7, 1.8, and 1.6, respectively, with differences statistically significant in pairwise comparisons (P < .0001).

Duals in at-risk MA compared with those in TM were observed to have more favorable outcomes in 17 of 20 measures of quality and health resource utilization across 4 domains of patient care (acute hospital care, avoidance of unnecessary ED use, avoidance of disease-specific inpatient admissions, and outpatient care) (Figure 1). With respect to measures of hospital use, the decrease in AMR for hospital admission was 46.97 per 1000 (95% CI, –49.46 to –44.49; P < .0001) and for 30-day readmission was 11.83 per 1000 (95% CI, –13.03 to –10.63; P < .0001), reducing unnecessary hospital use by 23.8% and 32.7%, respectively. For ED use measures, the decrease in AMR ranged from 9.04 per 1000 (95% CI, –10.63 to –7.45; P < .0001) for avoidable ED visits to 36.84 per 1000 (95% CI, –40.45 to –33.22; P < .0001) for overall ED visits, representing a 9.0% to 28.4% reduction in ED usage. With respect to measures reflecting avoidance of disease-specific admissions in 9 categories, significant reduction of avoidable utilization in the at-risk cohort was observed for 8 measures. The decreases in AMR of admissions in at-risk MA compared with TM ranged from 1.48 per 1000 (95% CI, –1.88 to –1.07; P < .0001) for hypertension-related admissions to 10.38 per 1000 (95% CI, –11.63 to –9.13; P < .0001) for the composite of chronic disease admissions. Overall, avoidable admission reductions ranged from 2.0% to 45.5%. For the outpatient measure domain, the decrease in AMR of high-risk medication use was 40.77 per 1000 (95% CI, –43.43 to –38.10; P < .0001), translating to a 26.1% reduction in the at-risk MA cohort (Figure 1). The at-risk MA cohort also was observed to have increased medication adherence by 50.70 per 1000 (95% CI, 45.88-55.52; P < .0001) for antihypertensive (renin-angiotensin system [RAS]) drugs and 36.59 per 1000 (95% CI, 31.82-41.37; P < .0001) for statins, corresponding to 6.4% and 4.6% higher adherence, respectively (Figure 1). The AMR of an office visit occurring was 20.81 per 1000 (95% CI, 19.42-22.20; P < .0001), representing a 2.2% increase in the at-risk MA cohort. The differences for admissions for diabetes-related lower-extremity amputations, diabetes composite admissions, and adherence to diabetes medications were not statistically significant between at-risk MA and TM.

Comparing Duals in at-risk MA to Duals in FFS MA, the former group was observed to have higher-quality outcomes in 18 of 20 measures (Figure 2). With respect to measures of hospital use, the decrease in AMR of hospital admission was 14.58 per 1000 (95% CI, –15.88 to –13.29; P < .0001) and of 30-day readmission was 4.10 per 1000 (95% CI, –4.74 to –3.45; P < .0001), translating to decreases of 8.8% and 14.4%, respectively, for Duals in at-risk MA. For the measures reflecting ED use, the decrease in AMR ranged from 6.51 per 1000 (95% CI, –7.41 to –5.60; P < .0001) for avoidable ED visits to 31.05 per 1000 (95% CI, –33.01 to –29.09; P < .0001) for overall ED visits, resulting in a 7.7% to 15.3% reduction in ED usage. For avoidance of disease-specific admissions, 8 of the 9 measures were significant. The at-risk MA cohort observed an AMR decrease for 8 of the measures that ranged from 0.38 per 1000 (95% CI, –0.57 to –0.19; P < .0001) for admissions related to hypertension to 4.51 per 1000 (95% CI, –5.19 to –3.82; P < .0001) for the composite of chronic disease admissions, corresponding to a reduction in avoidable admissions of 8.3% to 26.0%. There was statistical equivalence for the diabetes composite (PQI 93) measure. An increase in AMR of 0.25 per 1000 (95% CI, 0.06-0.45; P < .009), or 22.7%, was observed in admissions for diabetes-related lower extremity amputations in the at-risk MA cohort vs the FFS MA cohort. With respect to the outpatient domain, the reduction in AMR of high-risk medication use was 23.37 per 1000 (95% CI, –24.86 to –21.89; P < .0001), representing a 16.8% reduction in the at-risk MA cohort (Figure 2). Increased medication adherence AMR ranged from 12.85 per 1000 (95% CI, 8.36-17.35; P < .0001) for adherence to diabetes medications to 22.36 per 1000 (95% CI, 19.91-24.80; P < .0001) for adherence to antihypertensive (RAS) drugs, translating to increases of 1.8% and 2.7%, respectively, in the at-risk MA cohort (Figure 2). The increase in AMR of an office visit occurring was 3.06 per 1000 (95% CI, 2.43-3.69; P < .0001), corresponding to 0.3% more visits in the at-risk MA cohort.

We also compared Duals in FFS MA and TM (eAppendix Table 8). FFS MA was observed to have higher quality and efficiency on 17 of 20 measures relative to TM; TM exceeded FFS MA on 1 measure. Although highly significant, these outcome differences were somewhat smaller than the differences noted in the comparison of at-risk MA and TM.

As a subanalysis, heterogeneity of effect was examined for Duals in MA with and without D-SNP enrollment (eAppendix Table 5). Compared with D-SNP beneficiaries in FFS MA, D-SNP beneficiaries in at-risk MA showed higher quality and lower health resource utilization that was statistically significant and consistent with the overall analysis. For 15 of the 20 measures, performance gains in at-risk MA relative to FFS MA were even greater for those with D-SNP coverage than for those without it. In our separate subanalysis looking across D-SNP and non–D-SNP subgroups compared with TM, the greatest differences in performance were seen between at-risk MA D-SNPs and TM, with beneficiaries in at-risk MA D-SNPs showing the most favorable outcomes overall (eAppendix Table 6).

DISCUSSION

In this study, at-risk MA payment arrangements for Duals were associated with significantly higher outcomes in both quality and efficiency compared with both TM and FFS MA. The greatest differences were in the comparison of Duals in at-risk MA with those in TM. Importantly, the differences also remained significant when comparing Duals in at-risk MA with those in FFS MA. These differences were clinically meaningful and important drivers of both quality and cost of care. Given that Duals have a greater frequency of hospitalization than non-Duals, the measures of preventable hospitalizations are of particular importance because these are measures of outpatient care quality.⁹ Across all 3 cohorts, the 3 diabetes measures showed the smallest differences, often rendering these measures statistically equivalent among the cohorts. We do not have an explanation for this finding, and it warrants further study.

Prior study findings have suggested improved clinical and economic outcomes for the broad Medicare program when physicians are practicing in at-risk MA payment arrangements.^4,5 This study examined the impact of payment arrangements specifically on the Duals population in at-risk MA compared with both TM and FFS MA, which had not previously been studied.Our findings align with those of the prior studies that examined payment arrangements in the broad Medicare program,^4,5 also showing higher quality and efficiency when Duals are cared by physicians in at-risk arrangements.

There are several possible explanations for the differences we observed, all of which may be operating interdependently. Our results may reflect the care management infrastructure built by the physician groups and the MA health plans caring for Duals in MA. Because this infrastructure is funded by MA, it did not exist for most patients in TM during this study period. This infrastructure may include health-related social needs (HRSNs) interventions specifically adopted to help manage a Duals population, such as providing enhanced access to care, assisting with transportation, and addressing social isolation and food insecurity, because HRSNs are known to increase health care spending.⁹ Additional types of care management interventions are a feature of MA D-SNP plans, which offer more supplemental benefits than other MA plans, including Special Supplemental Benefits for the Chronically Ill. This care management infrastructure is primarily built and deployed by the physician groups in at-risk MA and the health plans in FFS MA. Although both create similar infrastructures, the primary differences when the physician group develops the infrastructure may be tighter provider/patient integration, access to patient-level electronic health record data, and interventions that are more tailored to the specific patient population. To the extent that the MA Duals population is enrolled in D-SNPs, this may also contribute to the observed differences relative to TM due to infrastructure elements specific to D-SNPs. D-SNP beneficiaries cared for under at-risk MA arrangements also showed overall higher quality and avoidable health resource utilization compared with D-SNP beneficiaries cared for under FFS MA arrangements. Additionally, the practice infrastructure built by these physician groups for use in the broader at-risk MA population is also deployed across the at-risk Duals population. This includes comprehensive case management and disease management, social work services, integrated behavioral health care, and pharmacy management, among others.

Lastly, our results could also reflect differences in access to care among the 3 cohorts, as Duals in MA experience less difficulty accessing and affording care than those in TM.¹⁰ However, access to care has not previously been studied specifically in Duals in at-risk MA. Our study found that Duals in both at-risk and FFS MA were more likely to have office visits than Duals in TM, with at-risk MA Duals having the highest rate. Finally, we note that because physicians tend to treat patients similarly despite differences in their health plans, this spillover effect would likely attenuate the results seen in this study.⁵

At the policy level, there is interest in Duals because of concerns about the large and growing expenditures for this relatively small group of individuals.¹¹ Given the rising percentage of Duals who choose MA over TM, the current study results suggest that the improvements in quality and efficiency would be expected to affect a greater percentage of the Duals population over time because both at-risk MA and FFS MA were associated with care of higher quality and efficiency. If these beneficiaries are cared for increasingly in at-risk payment arrangements, the benefits should be even greater when compared with those who remain in FFS MA. Given the CMS goal of having all Medicare and Medicaid beneficiaries in accountable care arrangements by 2030, this push should include encouraging Duals’ participation in at-risk MA in addition to at-risk models within MSSP and the ACO Realizing Equity, Access, and Community Health Model.

Limitations

To minimize any potential differences between groups, we adjusted for differences in populations across payment arrangements by adjusting for demographic, observable health, and clinical risk measures. We also accounted for differences in patient mix across the 3 payment models by using demographic and health risk score controls. Further, we minimized the impact of differences in physician mix because all 3 beneficiary cohorts were cared for by the same physicians within the same physician groups. It is possible that despite incorporating the above factors, there could remain residual unobservable differences among populations. We could not control for differences in the benefit design of the 35 MA health plans in this study because that level of detail was not available.

Prior studies comparing at-risk MA with TM have been criticized due to potential coding intensity in the MA population. The mean HCC scores in our 3 cohorts using version 24 were closely matched at 1.7 in the at-risk MA cohort, 1.8 in the TM cohort, and 1.6 in the FFS MA cohort, and we did not include chart reviews in our analysis. Coding intensity was therefore unlikely to have confounded the results. To further address potential coding and reporting differences between MA and TM, we conducted a sensitivity analysis adjusting for risk using HCC version 28 instead of HCC version 24 (eAppendix Table 7). The mean HCC scores in the 3 cohorts using version 28 were 1.3 in the at-risk MA cohort, 1.6 in the TM cohort, and 1.4 in the FFS MA cohort. The effects on outcomes remained strong and statistically significant, although slightly reduced compared with version 24 results.

Lastly, although we found associations between MA risk payment arrangements and differences in outcomes across the 3 cohorts, it cannot be determined from this study whether the risk payment arrangement is causally related to the outcome differences observed.

CONCLUSIONS

In this cross-sectional analysis, at-risk MA payment arrangements, when applied to the Duals population, were associated with care of higher quality and efficiency across 4 domains of care relative to Duals in TM and FFS MA. Although the study does not assert causality, the results support the current CMS goal of having all TM and Medicaid beneficiaries in fully accountable care arrangements by 2030. Further, they suggest that this goal should be broadened to include at-risk accountable care arrangements within MA and should be a particular focus for Duals given their rapidly increasing health care utilization and overall cost of care. 

Author Affiliations: Optum Center for Research and Innovation (KCo, OA, KCa, MSJ, JS), Minnetonka, MN; Department of Health Care Policy, Harvard Medical School (BV), Boston, MA; America’s Physician Groups (JP, SD), Washington, DC; CareJourney by Arcadia (NS), Arlington, VA.

Source of Funding: Funding was provided by Optum Health.

Author Disclosures: Dr Cohen is an employee of Optum Health, which participates in both Medicare Advantage and traditional Medicare; he owns stock in UnitedHealth Group and is a member of the America’s Physician Groups board. Dr Vabson received funding from Optum to fund his research time on this manuscript and, after completion of this manuscript, began serving as a senior adviser to CMS. Ms Podulka is employed by America’s Physician Groups. Dr Ameli, Dr Catlett, and Ms Sullivan are employees of Optum Health and own stock in UnitedHealth Group. Ms Jarvis is employed by Optum. Ms Dentzer is employed as president and CEO of America’s Physician Groups. Dr Smith reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (KCo, BV, JP, OA, KCa, NS, MSJ, SD); acquisition of data (KCo, BV, JP, NS); analysis and interpretation of data (KCo, JP, OA, KCa, NS, MSJ, JS, SD); drafting of the manuscript (KCo, BV, JP, KCa, MSJ, JS, SD); critical revision of the manuscript for important intellectual content (KCo, JP, OA, KCa, SD); statistical analysis (NS); administrative, technical, or logistic support (KCo, KCa, JS); and supervision (KCo).

Address Correspondence to: Kenneth Cohen, MD, Optum Health, 11000 Optum Circle, Eden Prairie, MN 33554. Email: ken.cohen@optum.com.

REFERENCES

1. Sutton JP, Jacobson G, Leonard F. The health care experiences of people dually eligible for Medicare and Medicaid: comparing traditional Medicare and Medicare Advantage. The Commonwealth Fund. June 27, 2024. Accessed February 13, 2025. https://www.commonwealthfund.org/publications/2024/jun/health-care-experiences-people-dually-eligible-medicare-medicaid

2. Medicare Payment Advisory Commission; Medicaid and CHIP Payment and Access Commission. Data Book: Beneficiaries Dually Eligible for Medicare and Medicaid. Medicaid and CHIP Payment and Access Commission. January 2024. Accessed March 7, 2025. https://www.macpac.gov/wp-content/uploads/2024/01/Jan24_MedPAC_MACPAC_DualsDataBook-508.pdf

3. Top 5 takeaways: CMS’s proposed overhaul of the MSSP ACOs. McDermott+. Accessed August 12, 2025. https://www.mcdermottplus.com/insights/top-5-takeaways-cmss-proposed-overhaul-of-the-mssp-acos/

4. Cohen KR, Vabson B, Podulka J, et al. Medicare risk arrangement and use and outcomes among physician groups. JAMA Netw Open. 2025;8(1):e2456074. doi:10.1001/jamanetworkopen.2024.56074

5. Vabson B, Cohen K, Ameli O, et al. Potential spillover effects on traditional Medicare when physicians bear Medicare Advantage risk. Am J Manag Care. 2025;31(8):390-396. doi:10.37765/ajmc.2025.89686

6. Cohen K, Ameli O, Chaisson CE, et al. Comparison of care quality metrics in 2-sided risk Medicare Advantage vs fee-for-service Medicare programs. JAMA Netw Open. 2022;5(12):e2246064. doi:10.1001/jamanetworkopen.2022.46064

7. Norton EC, Dowd BE, Maciejewski ML. Marginal effects—quantifying the effect of changes in risk factors in logistic regression models. JAMA. 2019;321(13):1304-1305. doi:10.1001/jama.2019.1954

8. Holm S. A simple sequentially rejective multiple test procedure. Scand J Statist. 1979;6(2):65-70.

9. Canterberry M, Figueroa JF, Long CL, et al. Association between self-reported health-related social needs and acute care utilization among older adults enrolled in Medicare Advantage. JAMA Health Forum. 2022;3(7):e221874. doi:10.1001/jamahealthforum.2022.1874

10. Archibald N, Soper M, Smith L, Kruse A, Wiener J. Integrating Care Through Dual Eligible Special Needs Plans (D-SNPs): Opportunities and Challenges. HHS Office of the Assistant Secretary for Planning and Evaluation; April 2019. Accessed April 10, 2025. https://aspe.hhs.gov/sites/default/files/migrated_legacy_files//188071/MMI-DSNP.pdf

11. Medicare Payment Advisory Commission; Medicaid and CHIP Payment and Access Commission. Data Book: Beneficiaries Dually Eligible for Medicare and Medicaid. Medicaid and CHIP Payment and Access Commission; February 2022. Accessed February 13, 2025. https://www.macpac.gov/wp-content/uploads/2022/02/Beneficiaries-Dually-Eligible-for-Medicare-and-Medicaid-February-2022.pdf