Abstract 10400: The Role of Patient, Community, and Healthcare System Factors in Variability of Hospital Readmissions After Myocardial Infarction

Abstract

Introduction: The Hospital Readmissions Reduction Program (HRRP) penalizes hospitals by reducing Medicare (CMS) reimbursement for those institutions with higher rates of readmissions following hospitalizations for myocardial infarction (MI). Understanding the contributors to readmission following MI hospitalization can identify meaningful areas for intervention. We examine whether community and patient or healthcare system factors contribute more to variability in MI readmissions. Methods: We compiled CMS hospital data on MI readmissions, CMS physician reimbursement data, and 2020-21 County Health Rankings (Credit: University of Wisconsin). The outcome was the ratio of a hospital’s MI readmissions to MI discharges. Physician/healthcare system factors included total number of cardiac subspecialists, presence of all subspecialties, and the mean proportion of reimbursed charges. Patient and community factors included mean Hierarchical Condition Category (HCC) risk score (quantifies expected healthcare costs for CMS beneficiaries based on ICD-10 coding), median household income, the proportion of residents reporting no physical activity, and access to food stores. Models were created using linear regression. Model of best fit was determined using R 2 , AIC, BIC in model training data and mean absolute prediction error (MAPE) and root mean square prediction error (RMSE) in test data. Results: The linear regression model incorporating community and patient factors explained greater variability in MI readmission rates and performed with greater accuracy in the test data (R 2 0.21; AIC -40.2; BIC -5.1; MAPE 0.030; RMSE 0.042) than the model incorporating physician/healthcare system factors (R 2 0.04; AIC 170.0; BIC 195.0; MAPE 0.033; RMSE 0.045). HCC score, level of physical activity, median household income, and number of cardiac specialists were associated with significant effect sizes on the outcome. Conclusions: At a county level, community and patient factors appear to contribute more to variability in hospitals’ MI readmission rates. Further analyses should explore other community, patient, and provider factors that contribute. The HRRP does not account for community or patient factors, indicating its limitations.