JAMA Netw Open. 2026 Jan 2;9(1):e2554972. doi: 10.1001/jamanetworkopen.2025.54972.
ABSTRACT
IMPORTANCE: Since 2012, the Hospital Readmissions Reduction Program (HRRP) has penalized hospitals for excess, risk-adjusted 30 day readmissions among traditional Medicare (TM) beneficiaries. While risk adjustment may address observable differences in patient severity, it cannot account for unobservable differences. Medicare Advantage (MA) enrollment has continued to increase, and MA beneficiaries have been found to be both observably and unobservably healthier than their TM counterparts. Because relatively lower-severity patients are increasingly likely to enroll in MA, hospitals with higher MA penetration may have unobservably higher-severity TM patients, resulting in higher-than-estimated readmission risk and excessive HRRP penalties.
OBJECTIVE: To determine whether unobserved selection, as proxied by MA penetration, could be associated with distorted HRRP penalties and how associations may be moderated by peer grouping, which was incorporated into HRRP’s 2019 revision to penalty calculations.
DESIGN, SETTING, AND PARTICIPANTS: This retrospective cohort study included hospitals serving Medicare beneficiaries from fiscal years 2019 to 2022 for 6 HRRP-targeted conditions, including acute myocardial infarction, chronic obstructive pulmonary disease, heart failure, pneumonia, coronary artery bypass graft surgery, and elective primary total hip or knee arthroplasty. Data were analyzed from January 2024 to October 2025.
EXPOSURE: MA penetration at the hospital-year level.
MAIN OUTCOMES AND MEASURES: Excess readmission ratios (ERRs) and calculated HRRP penalties at the hospital-year level. To test whether HRRP penalties could have been distorted by unobserved selection, the association between the share of hospital admissions for MA patients (MA penetration) and excess readmission ratios (ERRs) for all patients was estimated, controlling for county-level variation and hospital-level covariates. The ERRs were rescaled by MA penetration to account for unobserved selection, and the rescaled ERRs were used to reestimate HRRP penalties under non-peer grouping and peer grouping paradigms.
RESULTS: This study included 3203 hospitals and 12 135 hospital-years. After adjusting for MA penetration, estimates indicated that hospitals in the first quintile of MA penetration would be penalized by a mean (SD) of $30 736 ($24 819.75) more, while hospitals in the fifth quintile would be penalized by a mean (SD) of approximately $26 915 ($42 017.23) less. Peer grouping does not mitigate these penalty distortions. Across hospitals, penalty redistributions would amount to $284 to $297 million annually.
CONCLUSIONS AND RELEVANCE: The findings of this study suggest that including MA penetration explicitly in risk adjustment or in peer group definitions may dampen distortions from unobservable patient severity in HRRP penalty calculations.
PMID:41569562 | DOI:10.1001/jamanetworkopen.2025.54972
