Factors Associated With Disparities in Hospital Readmission Rates Among US Adults Dually Eligible for Medicare and Medicaid

Abstract

Importance: Low-income older adults who are dually eligible (DE) for Medicare and Medicaid often experience worse outcomes following hospitalization. Among other federal policies aimed at improving health for DE patients, Medicare has recently begun reporting disparities in within-hospital readmissions. The degree to which disparities for DE patients are owing to differences in community-level factors or, conversely, are amenable to hospital quality improvement, remains heavily debated.

Objective: To examine the extent to which within-hospital disparities in 30-day readmission rates for DE patients are ameliorated by state- and community-level factors.

Design setting and participants: In this retrospective cohort study, Centers for Medicare & Medicaid Services (CMS) Disparity Methods were used to calculate within-hospital disparities in 30-day risk-adjusted readmission rates for DE vs non-DE patients in US hospitals participating in Medicare. All analyses were performed in February and March 2019. The study included Medicare patients (aged ≥65 years) hospitalized for acute myocardial infarction (AMI), heart failure (HF), or pneumonia in 2014 to 2017.

Main outcomes and measures: Within-hospital disparities, as measured by the rate difference (RD) in 30-day readmission between DE vs non-DE patients following admission for AMI, HF, or pneumonia; variance across hospitals; and correlation of hospital RDs with and without adjustment for state Medicaid eligibility policies and community-level factors.

Results: The final sample included 475 444 patients admitted for AMI, 898 395 for HF, and 1 214 282 for pneumonia, of whom 13.2%, 17.4%, and 23.0% were DE patients, respectively. Dually eligible patients had higher 30-day readmission rates relative to non-DE patients (RD >0) in 99.0% (AMI), 99.4% (HF), and 97.5% (pneumonia) of US hospitals. Across hospitals, the mean (IQR) RD between DE vs non-DE was 1.00% (0.87%-1.10%) for AMI, 0.82% (0.73%-0.96%) for HF, and 0.53% (0.37%-0.71%) for pneumonia. The mean (IQR) RD after adjustment for community-level factors was 0.87% (0.73%-0.97%) for AMI, 0.67% (0.57%-0.80%) for HF, and 0.42% (0.29%-0.57%) for pneumonia. Relative hospital rankings of corresponding within-hospital disparities before and after community-level adjustment were highly correlated (Pearson coefficient, 0.98).

Conclusions and relevance: In this cohort study, within-hospital disparities in 30-day readmission for DE patients were modestly associated with differences in state Medicaid policies and community-level factors. This suggests that remaining variation in these disparities should be the focus of hospital efforts to improve the quality of care transitions at discharge for DE patients in efforts to advance equity.

Figure 1.. Within-Hospital Disparities in Risk-Standardized Readmission Rates Before and After Adjustment for Community-Level Indicators of Social Risk

AMI Indicates acute myocardial infarction; HF, heart failure; PN, pneumonia. Model 1 reflects the standard CMS approach. Model 2 adjusts the standard Centers for Medicare & Medicaid Services (CMS) approach for state Medicaid policy differences pertaining to eligibility and enrollment. Model 3 adjusts the standard CMS approach for state Medicaid policy differences pertaining to eligibility and enrollment, as well as health services availability; model 4 adjusts the standard CMS approach for state Medicaid policy differences pertaining to eligibility and enrollment, health services availability, as well as community-level social risk factors.

Figure 2.. Correlation in Within-Hospital Disparities in Risk-Standardized Readmission Rates Before and After Adjustment for Community-Level Indicators of Social Risk

^aWithin-hospital disparity reflects the difference in risk-standardized readmission rates between dual-eligible and all other patients. ^bFully-adjusted within-hospital disparities accounts for clinical risks (age, sex, clinical conditions), as well as state Medicaid policies, local health service availability, and community-level indicators of social risk. ^cOriginal approach to measuring within-hospital disparities accounts for differences in clinical risks (age, sex, clinical conditions).