Abstract 174: The Longitudinal Effect of Medicaid Expansion on 30-Day Readmissions for Heart Failure and Acute Myocardial Infarction - Insights from Hospital Compare

Background: The emergency department (ED) is a primary access point for hospital admissions/ readmissions. Recent analyses have found that state-level Medicaid expansion is associated with higher ED volumes and longer ED length of stay (LOS). Increased ED congestion has been proposed to lead to higher readmission rates, though this has not been empirically evaluated. We examined the association of Medicaid expansion with readmission rates for AMI and HF specifically given that these readmissions are prevalent, publicly reported, and subject to financial penalties. Methods: We utilized the Hospital Compare database to construct a panel of US acute care hospitals offering ED services over three time intervals: 10/01/2012-09/30/2013, before any Medicaid expansion effective coverage dates (“Period 1”); 10/1/2013-9/30/2014 and 10/1/2014-9/30/2015, after Medicaid expansion effective coverage dates (“Periods 2 and 3”). We used multivariate linear fixed-effects regression to evaluate the independent effect of Medicaid expansion on absolute unplanned 30-day case-mix adjusted readmission rates for AMI and HF. Fixed effects regression adjusts for time-invariant confounders by design and potential time-variant confounders were included as covariates, including year fixed effects to control for temporal variation. Standard errors were clustered on states. Sensitivity analysis was performed to evaluate the interaction between Medicaid expansion and ED LOS with respect to the primary outcome. Results: Our model examined 2,257 hospitals for AMI and 2,947 hospitals for HF. Overall, within-hospital 30-day unplanned readmission rates for AMI and HF decreased over the three defined periods. AMI readmission rates decreased by 1.37% (95% CI: -1.46% to -1.27%) from Period 1 to Period 3. HF readmission rates decreased by 1.07% (95% CI: -1.21% to -0.93%) from Period 1 to Period 3. All p-values were <0.001. Medicaid expansion effective coverage dates did not affect within-hospital readmissions rates for AMI (ARR -0.09%; 95% CI: -0.22% to 0.04%; p-value 0.162) or HF (ARR -0.03%; 95% CI: -0.20 to 0.15%; p-value 0.770). A sensitivity analysis showed no interaction between Medicaid expansion and ED LOS with respect to the primary outcome (p value = 0.190 for AMI, p-value = 0.255 for HF). Conclusion: Among hospitals with publicly reported outcomes in the Hospital Compare database, readmission rates for AMI and HF decreased from 2012-2015. Medicaid expansion was not associated with significant within-hospital differences in 30-day readmission rates for AMI and HF. There was no interaction by ED LOS. Our results offer evidence to quell concerns that Medicaid expansion, by increasing ED LOS, may adversely impact unplanned readmission rates. Further studies are needed to evaluate factors affecting within-hospital readmission rates.