Essays on How Hospitals Responded to Pay for Performance Incentives

My dissertation analyzes the effects of changes in Medicare reimbursement policies. The specific payment change I analyze is the Hospital Readmissions Reduction Program (HRRP), which is a prominent Pay for Performance (P4P) policy that penalizes hospitals for excess readmissions. In a body of work , I study the consequences of this policy on hospital resource use and patient outcomes for illnesses that are an explicit focus of the P4P policy and for illnesses not explicitly specified by the policy. Empirically, in two chapters, I use two, complementary quasi-experimental research designs, regression kink and difference-in-differences, combined with data on the entire Medicare inpatient population. The first chapter, develops a theoretical model of hospital behavior and from this makes predictions about the effect of the HRRP on hospital inpatient spending highlighting the potential for the P4P plan to reallocate resources in both intended and unintended ways. This model guides the empirical analyses. First, I use a regression kink design to obtain estimates of the effect of the HRRP on readmissions and potential mechanisms that hospitals may use to reduce readmissions, such as spending on inpatient care, discharge destination and patient selection. I also examine the effect of the HRRP on mortality. Estimates indicate that hospitals penalized for excess heart attack (AMI) readmissions decreased AMI readmissions by 30% and increased spending on AMI patients by 40%. This additional care had no impact on mortality. Interestingly, I find that hospitals penalized for AMI readmissions increased the quantity of care for patients with diagnoses not targeted by the HRRP. Thus the P4P incentives of the HRRP did not cause hospitals to reallocate resources away from non-targeted conditions. Hospitals penalized for excess readmissions for pneumonia or heart failure did not appear to respond to the HRRP incentives. Interestingly, I demonstrate using the conceptual model, that as the number of patients in the targeted condition rises, the marginal cost of reducing the penalty increases by relatively more than the marginal benefit. Since HF and PN admit a relatively larger number of patients, this could increase the cost associated with amending the process of care and reducing readmissions for these conditions. The analysis in chapter 1, assumes that the treatment is equal to the HRRP penalty (or revenue reduction) experienced at the start of the HRRP program. With this assumption, the empirical analysis is well-suited for the regression-kink research design because the HRRP penalty is zero until it reaches a threshold and then grows linearly from zero at the threshold to a maximum penalty. In a second chapter, I assume that hospitals form expectations about the probability of being penalized based on the relationship between past hospital performance with respect to readmissions and the HRRP penalty at the start of the program. In this approach, some hospitals that were not penalized at the start of the program still have a positive expectation of being penalized in the future. I use this expected penalty to define the treatment due to the HRRP and execute a difference in differences design to estimate the HRRP effects on all Medicare hospitals.I find the largest reduction in readmissions due to the HRRP to be for AMI (heart attack) patients. Specifically, hospitals reduced AMI readmission rates commensurately with expectations of future penalties. I find that only hospitals with the highest expected penalties reduced HF and PN readmissions. I also find evidence that hospitals focused the reductions in readmissions towards the condition with the highest share of the penalty. That is, hospitals mainly penalized due to AMI readmissions, intensified the response on AMI patients and did not reduce PN and HF readmissions.