Association of Medicare Advantage Star Ratings With Racial, Ethnic, and Socioeconomic Disparities in Quality of Care

Abstract

IMPORTANCE Medicare Advantage (MA) plans, which disproportionately enroll racial/ethnic minorities and persons with socioeconomic disadvantage, receive bonus payments on the basis of overall performance on a 5-star rating scale. The association between plans' overall quality and disparities in quality is not well understood. OBJECTIVE To examine the association between MA star ratings and disparities in care for racial/ ethnic minorities and enrollees with lower income and educational attainment. DESIGN, SETTING, AND PARTICIPANTS This cross-sectional study included 1 578 564 MA enrollees from 454 contracts across the 2015 and 2016 calendar years. Data analyses were conducted between June 2019 and June 2020. EXPOSURES Self-reported race and ethnicity and low socioeconomic status (SES) (defined by low income or less than a high school education) vs high SES (neither low income nor low educational attainment). MAIN OUTCOMES AND MEASURES Performance on 22 measures of quality and satisfaction determined at the individual enrollee level, aggregated into simulated star ratings (scale, 2-5) stratified by SES and race/ethnicity. RESULTS A total of 1 578 564 enrollees were included in this analysis (55.8% female; mean [SD] age, 71.4 [11.3] years; 65.8% White; 12.3% Black; 14.6% Hispanic). Enrollees with low SES had simulatedstratified star ratings 0.5 stars lower (95% CI, 0.4-0.6 stars) than individuals with high SES in the same contract. Black and Hispanic enrollees had simulated star ratings that were 0.3 stars (95% CI, 0.2-0.4 stars) and 0.1 stars (95% CI, −0.04 to 0.2 stars) lower than White enrollees within the same contracts. Black enrollees had a 0.4-star lower rating (95% CI, 0.1-0.7 stars) in 4.5-to 5-star contracts and a no statistical difference in 2.0-to 2.5-star-rated contracts (difference, 0.3 stars; 95% CI, −0.02 to 0.7 stars). Hispanic enrollees had a 0.6-star lower simulated rating (95% CI, 0.2-1.0 stars) in 4.5to 5-star contracts and no statistical difference in 2-to 2.5-star contracts (difference, −0.01 stars; 95% CI, −0.5 to 0.4 stars). There was low correlation between simulated ratings for enrollees of low SES and high SES (difference, 0.2 stars; 95% CI, 0.03-0.4 stars) and between simulated ratings for White and Black enrollees (difference, 0.4 stars; 95% CI, 0.3-0.5 stars) and White and Hispanic enrollees (difference, 0.3 stars; 95% CI, 0.2-0.4 stars). As the proportion of Black and Hispanic enrollees increased in a contract, racial/ethnic disparities in ratings decreased. CONCLUSIONS AND RELEVANCEIn this cross-sectional study, simulated MA star ratings were only weakly correlated with those for enrollees of low SES in the same contract, and contracts with higher star ratings had larger disparities in quality. Measures of equity in MA plans' quality of care may be needed.

Medicare Advantage Enrollment Among Beneficiaries With End-Stage Renal Disease in the First Year of the 21st Century Cures Act

“…Similarly, compared with non-Hispanic White people, Hispanic or Latino individuals have higher prevalence rates of ESRD . There are well-documented racial and ethnic inequities in health outcomes among MA enrollees . For example, compared with TM, MA is associated with better outcomes and ambulatory care access for beneficiaries from racial and ethnic minority groups; however, these racial and ethnic minority beneficiaries reported worse outcomes for most measures compared with their White counterparts, irrespective of whether they were in MA or TM .…”

Provider discussion of health goals and psychosocial needs: Comparing older to younger veteran experience

“…We selected these variables as they have been shown to contribute to differences in patient experience on CAHPS. 10,[30][31][32] After fitting unadjusted logistic regression models for age differences on each outcome (model 1), models were adjusted by adding covariates successively, including sex (model 2), SES, education, and rurality (model 3), and self-rated physical and mental health and GCI (model 4). In all models, we expressed the effects of age as differences in the experience of outcome probabilities between the age 65+ group and each of the younger age groups.…”

“…We then tested three weighted adjusted models, each of which added successively more covariates, including sex, SES, education, rurality, self‐rated mental and physical health, and GCI. We selected these variables as they have been shown to contribute to differences in patient experience on CAHPS 10,30–32 . After fitting unadjusted logistic regression models for age differences on each outcome (model 1), models were adjusted by adding covariates successively, including sex (model 2), SES, education, and rurality (model 3), and self‐rated physical and mental health and GCI (model 4).…”

Hospital value-based payment programs and disparity in the United States: A review of current evidence and future perspectives

“…They are particularly concerned about the potential biases of the VBP programs and their implications for health care equity and disparity (46)(47)(48)(49)(50). Specifically, these programs are suspected to have penalized hospitals and medical practices that serve patients of predominantly lower socioeconomic status (SES) who tend to live in areas with higher concentrations of poverty and minority population groups (12,(51)(52)(53)(54)(55). Some of the biases arose from the budgetary constraints and incentives design inherent in the existing VBP programs, while others were the results of their implementation as providers made selfinterested decisions that might not be in the best interest of their patients.…”