Racial and Ethnic Health Equity in Medicare

Medicare is designed to be race-neutral. Payment rates, coverage rules, and beneficiary rights are uniform across racial and ethnic groups by statute. The outcomes are not uniform. Black Medicare beneficiaries are hospitalized for acute exacerbations of chronic disease at higher rates than white beneficiaries. Hispanic beneficiaries carry the highest uninsured rates and the most persistent cost-related barriers to care. American Indian and Alaska Native beneficiaries face access constraints that compound chronic disease burdens already three to five times the national average. The mechanisms driving these gaps are increasingly legible in CMS administrative data, in published research on HCC coding completeness, in OIG analyses of prior authorization denial rates, and in Health Affairs studies of MA network composition. What is changing in 2025 and 2026 is not the existence of these disparities but the systematic removal of the federal infrastructure that was designed to measure and address them.

HCC Capture Rate Disparities

#

The CMS-HCC risk adjustment model predicts healthcare costs for Medicare Advantage enrollees based on diagnosis codes documented in prior-year claims. The model is calibrated on fee-for-service Medicare spending data. When the model underpredicts costs for a subpopulation, the plans serving that population receive lower capitation payments than the actual cost of care warrants. When it overpredicts, the reverse occurs.

Avalere Health’s analysis of the CMS-HCC community model using 2018-2019 Medicare FFS claims data found that the model closely predicts costs on average for Black and non-Hispanic white beneficiaries but that the accuracy diverges as clinical complexity increases. For beneficiaries with higher HCC counts, overprediction increases for non-Hispanic white enrollees while underprediction emerges for Black and Hispanic enrollees. The pattern suggests that the HCC count variable mandated by the 21st Century Cures Act, which increased predicted costs for beneficiaries with more documented conditions, disproportionately benefits populations whose conditions are more completely coded. The model also underpredicts costs for American Indian and Alaska Native beneficiaries across the board.

The coding gap has identifiable drivers. Black and Hispanic Medicare beneficiaries have lower primary care utilization rates, shorter and less frequent provider encounters, and higher rates of receiving care in settings where documentation completeness is lower. Language barriers affect documentation for beneficiaries whose primary language is not English. Clinical documentation improvement programs, the chart review and retrospective coding operations that plans deploy to ensure diagnostic completeness, have historically been concentrated in markets with higher-income, higher-proportion-white populations where MA penetration rates and plan administrative capacity are greatest.

The V28 risk adjustment model transition, now fully phased in at 100 percent for 2025 dates of service, restructured the HCC classification using ICD-10 codes and updated the underlying data from 2014-2015 to 2018-2019. The transition from V24 to V28 reduced the number of eligible ICD-10 diagnosis codes by approximately 2,269 while expanding the number of payment HCCs from 85 to 115. The clinical reclassification improved specificity for conditions including heart failure, metabolic diseases, and substance use disorders. Whether the V28 model narrows or widens the racial coding gap depends on whether the conditions where coding variation is greatest overlap with the conditions where Black and Hispanic beneficiaries are most undercoded.

The encounter-based risk adjustment transition compounds the problem. Under chart review-based RA, plans could deploy CDI contractors to retrospectively capture diagnoses from any provider record. Plans with concentrated MA enrollment in higher-income white populations had better CDI infrastructure, wider provider network access for chart retrieval, and more resources to invest in retrospective coding. Under encounter-based RA, HCC capture requires documentation at the point of care by a treating provider. The access and documentation infrastructure gaps that drive the coding disparity become the binding constraint. If a beneficiary does not see a provider, or sees a provider in a setting where documentation protocols are weaker, the diagnosis is not captured regardless of what a retrospective chart review might have found.

MA Supplemental Benefit Access Disparities

#

Supplemental benefit availability in Medicare Advantage is not uniform across geographies. Counties with higher-income, higher-proportion-white beneficiary populations have historically received richer supplemental benefit offerings. The market logic is straightforward: plans offer richer benefits where they can attract healthier, lower-cost members whose expected spending falls below the capitation rate, generating the rebate dollars that fund supplemental benefits. The equity consequence is that the populations with the highest unmet need for services like dental, vision, hearing, transportation, and meal delivery are concentrated in markets where the fewest supplemental benefits are available.

Prior authorization patterns show a related disparity. OIG analyses of MA prior authorization denial rates have documented differential authorization patterns by race and ethnicity. The behavioral health PA disparity is the most pronounced: denial rates for mental health and substance use disorder services show the largest racial gaps in published OIG analyses. Black beneficiaries are denied behavioral health authorizations at higher rates than white beneficiaries, a pattern that intersects with the behavioral health coverage gaps documented across the MA program.

A Health Affairs study published in January 2025 examined the composition of MA physician networks by provider race and ethnicity. The analysis found that roughly 51 percent of white physicians in a given beneficiary’s county were included in their MA network, compared with approximately 43 percent of Black physicians and 44 percent of Hispanic physicians. About 20 percent of Black and Hispanic beneficiaries had no available Black or Hispanic physicians included in their MA network at all, and more than 40 percent of counties had no Black or Hispanic physicians in any MA network. The research literature consistently shows that racially concordant care is associated with greater use of preventive services and better health outcomes. When MA networks disproportionately exclude Black and Hispanic providers, the clinical consequences extend beyond the network adequacy question into the quality of care that beneficiaries receive.

The Current Administration’s Equity Infrastructure Dismantling

#

The CY 2027 proposed rule released November 25, 2025 formalized a set of policy changes that had been signaled throughout the year. CMS proposed not to implement the Health Equity Index reward, previously renamed the Excellent Health Outcomes for All reward, which had been designed to provide Star Ratings bonuses for plans that achieved high measure-level scores for enrollees with specified social risk factors. The HEI had been finalized for inclusion in the 2027 Star Ratings based on 2024 and 2025 measurement year data. Its removal eliminates the only quality-weighted financial incentive for plans to invest specifically in outcomes for socially at-risk populations.

The proposed rule also eliminated the requirement that MA utilization management committees include a health equity expert, conduct annual health equity analyses of prior authorization use, and publicly post the results. It rescinded the requirement that quality improvement programs include activities specifically addressing health disparities. It proposed to reduce cultural competency regulatory requirements.

These regulatory changes arrived alongside administrative actions that preceded them. Effective January 1, 2026, CMS redesignated the G0136 billing descriptor from social determinants of health assessment to physical activity and nutrition assessment, removing the reimbursement signal for providers to conduct standardized SDOH screening. DEI-focused contracting, data collection, and research programs within CMS were reduced or eliminated as part of the broader federal DEI executive order rollback.

The cumulative effect is the removal of multiple reinforcing mechanisms. The HEI created a financial incentive for plans. The UM committee health equity analysis created a transparency mechanism. The G0136 descriptor created a provider-level reimbursement signal. The DEI research infrastructure created the data and analytical capacity to measure disparities and design interventions. Each mechanism addressed a different point in the healthcare delivery chain. Removing them simultaneously does not eliminate racial health disparities. It eliminates the measurement, incentive, and accountability infrastructure that made those disparities visible to the organizations positioned to address them.

What Remains and What Can Be Built

#

Not everything has been removed. CMS continues to publish race and ethnicity enrollment data files that form the basis for independent research. The subset of ACO quality measures that survived the quality measure rationalization includes equity-relevant metrics. State-level equity programs in California, New York, Illinois, and other states operate under state authority independent of the federal framework.

Language access requirements for MA plans remain in effect. Plans are still required to provide translation services, multilingual materials, and interpreter access for beneficiaries with limited English proficiency. Enforcement of these requirements has been inconsistent, but the legal obligation has not been rescinded.

The culturally competent care investment case extends beyond regulatory compliance. Published evidence consistently shows that culturally appropriate care delivery reduces emergency department utilization, improves chronic disease management, and lowers total cost of care for minority populations. Plans that invest in culturally competent networks and care management are not doing so only for equity reasons; they are reducing costs that fall to the plan under capitated payment. The regulatory withdrawal does not change the underlying economics.

ACOs retain the capacity to build equity monitoring into operations voluntarily. The remaining ACO quality measures provide a framework for tracking outcomes by race and ethnicity even without a federal mandate to do so. Organizations with internal race and ethnicity data have an analytical advantage that the federal data withdrawal makes more valuable. When CMS stops producing equity analyses, the organizations that maintain their own capacity become the primary source of evidence on whether disparities are narrowing or widening.

The research community remains active. Academic and advocacy researchers drawing on CMS administrative data, MCBS survey data, and state-level data sources continue to publish analyses of Medicare disparities. The Health Affairs, JAMA, and NEJM pipelines for equity research have not slowed. What has changed is that the federal agency responsible for administering Medicare no longer treats the measurement of racial health disparities as a program priority. The data still exists. The analytical infrastructure still exists in universities and advocacy organizations. The signal from CMS about whether disparities matter to program administration has changed.

For plans, providers, and ACOs operating in 2026, the practical question is not whether federal equity infrastructure will return. It is whether the organizations that serve diverse Medicare populations will maintain their own measurement and intervention capacity in the absence of federal requirements to do so. The plans that do will be better positioned when the regulatory pendulum shifts again. The plans that treat the regulatory withdrawal as permission to stop measuring will discover that the disparities they stopped tracking did not stop growing.

Related Reading

#

MCR-03_03 Medicare Equity: What the HEI Reversal Signals and What Remains MCR-02_03 Three Years of HCC Reform: What the 2024 CMS-HCC Model Actually Changed MCR-00_03 The Medigap Market