Moneyball in Medicare: New Insights

The Center for Medicare and Medicaid Services (CMS) is moving away gradually from traditional fee-for-service reimbursement, and instead trying to reward hospitals that deliver high quality of care for lower cost. Over the last decade, CMS has increased the number of value-based purchasing programs, which provide financial incentives to hospitals to improve quality of care and reduce episode spending. These programs include the Hospital Value-Based Purchasing (HVBP) Program, the Hospital Readmission Reduction Program, and Hospital Acquired Condition Program. It remains unknown how well the financial incentives work at achieving the goals. Some early studies showed that the overall change in quality of care was modest or zero. However a recent study showed that hospitals that faced higher financial incentives at the margin for a particular quality measure were often more likely to improve their performance on that measure over the next year (Norton et al., 2018).

This paper updates the prior published study by using more recent data and exploring more carefully what the exact relationship is between the incentives and annual performance improvement. The prior study used a simple linear relationship to predict one-year improvement. Instead, the relationship might be non-linear, have lags, differ by hospital type, change over time, and spillover to other programs. This issue is not merely of academic interest. There are also important policy considerations. If this HVBP Program is to achieve its goals, it is important to know how hospitals respond, over what time period, and whether some hospitals respond more strongly. Understanding how hospitals respond to financial incentives is necessary to design program incentives that are effective.

This study builds on prior work to test the relationship between marginal financial incentives and year-to-year changes in specific measures for the HVBP program. We explore whether the relationship is linear or non-linear, has lags of up to three years, differs by for-profit status and integration status, and whether it has changed over the life of the program. Preliminary results show a strong nonlinear relationship that is stronger for for-profit hospitals but weakens over time.