Patient Selection Under Quality Incentives: Evidence from Hospital Value-Based Purchasing Program

Historically hospitals treating Medicare patients are reimbursed on the basis of the quantity of services. Under a prospective payment scheme, hospitals incur the entire marginal cost of treatment and therefore face the tradeoff between intensity of treatment and profits. This payment system also incentivizes hospitals to strategically choose the extent of treatment, since more severe patients are more costly to treat. Ellis (1998)  shows that prospectively paid providers have the incentives to oversupply services to low cost patients and oversupply to high cost patients. In recent years, public policy makers and private insurers have become increasingly interested in the pay-for-performance scheme, which reward providers for providing better quality of care. With the passage of the Affordable Care Act, the focus now has been shifted toward containing cost growth and promoting better quality of care. In response, the Center for Medicare & Medicaid Services (CMS) inaugurated the Hospital Value-Based Purchasing Program (HVBP) as an effort to improve the quality of inpatient care.

In this study, we develop a theory model that incorporates the essential features of the HVBP and investigate how this change in financial incentives would impact provider behaviors. Specifically, we look at whether hospitals with different costs of improving quality would engage in patient selection under the quality incentives. In our model, capacity-constrained hospitals act as local monopolists and attract patients by setting two dimensions of quality: the clinical quality and patient satisfaction. Two types of  patients (low-severity versus high-severity) respond to the two dimensions differentially. Qualities are observable to the patients but not the payer. Hospitals are rewarded if the measured quality (by the payer) exceeds an exogenous threshold. Our model predicts that hospitals with high marginal cost and low marginally benefit of improving clinical quality (whose initial values at the higher end of its distribution) will be more likely to treat proportionally more low-severity patients. This implies that the quality incentives could lead to worse matching between hospitals and patients and thus have unintended welfare consequences.

We also seek to empirically test the predictions derived from our theory model. We put together data from several sources. We collect the 2008-2012 hospital performance data from the CMS website and compute measures of the two dimensions of quality using the CMS guidelines. We then link the quality measures to the comprehensive inpatient data compiled by the Pennsylvania Healthcare Cost Containment Council (PHC4). We also merge in the hospital characteristics obtained from the American Hospital Association (AHA). We use linear fixed-effects regression models to estimate the how hospital quality measures affect the severity of patients treated. Our preliminary results suggest that the introduction of HVBP is associated with a decrease in the severity of patients treated in hospitals with high clinical quality.

Our findings can be informative to policymakers. When designing quality-based reimbursement schemes, it is important to understand how the financial incentives would affect provider's behaviors. Our model suggests that taking into account provider heterogeneity in the cost of improving quality can mitigate the perverse incentives for patient selection.