Hospital Behavior Heterogeneity in the Hospital Readmissions Reduction Program

The Hospital Readmissions Reduction Program (HRRP) incentivizes acute care hospitals to decrease their preventable readmissions by putting hospitals at risk of reduced Medicare reimbursements. This program motivates hospitals to reduce their excess readmissions below their expected rates for several conditions to avoid receiving a penalty. Understanding how hospitals respond to this incentive structure can help the program achieve its intended effects. Accordingly, we evaluate the extent to which the different incentives faced by hospitals explain changes in hospital readmission ratios over time.

Using CMS Hospital Compare and AHA Annual Survey data, we explore how well heterogeneity in hospital characteristics and hospital performance in the program explain hospital-level changes in readmission rates for Acute Myocardial Infarction (AMI), Heart Failure, and Pneumonia patients between FY 2014 and FY 2015 of the program. We examined the effect of the following characteristics on hospital program performance in subsequent years using linear and logistic regression analysis: percent Medicare discharges, lagged readmission ratios and program performance, and condition-specific diagnosis-related group (DRG) volume. Since the Medicare caseload for the incentivized diagnoses and the percentage of hospital discharges from Medicare patients affect the penalty implications on overall revenue as well as opportunities for improvement, we hypothesize that these characteristics will significantly modify hospital responsiveness to the program. We will also calculate the condition-specific marginal effect of reducing readmissions to identify which condition has the greatest potential to improve hospital performance in the program, and determine how well these marginal effects accurately predict readmission changes in FY 2016.

94% of hospitals that received a penalty in 2014 also received a penalty in 2015. In bivariate specifications, hospitals that received a penalty in 2014 had a significant decrease in their readmission rates compared to hospitals that did not receive a penalty (ex: drop in AMI readmissions ratio by 0.0150, p<0.001). Hospitals in the highest decile of AMI readmission ratios in FY 2014 had the largest drop in their readmission ratios (-.0510, p<0.001), while hospitals in the bottom decile of AMI readmissions in FY 2014 increased their readmission ratio (+.019, p<0.001). Percent Medicare discharges and condition-specific DRG volume were not significant predictors of changes in readmission ratios across all three conditions, though we will explore other specifications to assess whether these variables are truly insignificant.

We find some evidence that hospitals receiving penalties do seem to be responding to the program by reducing their readmission rates in subsequent performance periods, but they do not seem to be reducing their rates enough to avoid penalties. Surprisingly, hospital characteristics such as percent Medicare discharges and condition-specific DRG volume did not significantly modify hospital responsiveness to the HRRP. Once we are able to examine hospital performance in FY 2016, we will be able to determine whether hospitals are improving their readmission ratios in the conditions that have the greatest potential benefit, and whether the heterogeneity of hospital response to the incentive structure changes over three years.