Incentives in Medicare's Payment for Episodes of Care

The Center for Medicare and Medicaid Services now pays hospitals in part on the entire episode of a patient’s care up to 30 days post-discharge as part of Hospital Value-Based Purchasing.  CMS rewards (or penalizes) hospitals financially for mortality, readmissions, clinical processes of care, patient satisfaction, and the total cost of care during an entire episode of care.  We estimate the magnitude of the financial incentives for each patient, and how those incentives differ across hospitals.  We then show whether hospitals responded to those incentives in the next year by making improvements in the areas where the incentives were strongest.  We hypothesize that hospitals will improve their performance in the areas where they have the highest incentives to improve care.

The Hospital Value-Based Purchasing (HVBP) program adjusts Medicare payments to hospitals by up by a predetermined percentage (up to 2.0% in FY 2017).  In other words, each patient’s episode of care ¾ including in a skilled nursing facility, outpatient rehabilitation, and home health care ¾ can affect future Medicare payments.  The HVBP program’s incentives have several implications.  First, hospitals have incentives to improve many aspects of quality of care and to reduce spending.  Second, the specific funding formula has implicit tradeoffs between reducing spending and lowering mortality (or improving other quality indicators).  Third, these tradeoffs differ across hospitals, depending on where the hospitals are in the distribution of spending and quality.  Fourth, because the tradeoffs differ across hospitals, different hospitals have vastly different incentives to make improvements in spending or in different quality domains.  Therefore, we expect hospitals to respond to their own incentives at the margin, and make improvements in different areas in the future.

After describing the HVBP program and incentives, the empirical work has three parts.  First, we quantify the overall tradeoff in the HVBP incentives between reducing episode spending and reducing mortality or improving other quality measures.  Second, we quantify the heterogeneity in those incentives across hospitals, by calculating the effect of marginal patients on their HVBP score.  Third, we regress changes between 2015 to 2016 in these different spending and quality domains on the marginal incentives faced by each hospital.  We predict that hospitals respond to those incentives by improving their score on the domains that have the greatest potential payoff.

We have patient-level data for all Medicare patients from FY 2015.  From those data we have computed the one spending and three quality domain scores and the resulting percentage change in Medicare payments for all 3,194 eligible hospitals in the US.  Soon we will obtain FY 2016 data and complete the analyses.  Using the new data we can see how hospitals changed their domain scores as a function of the marginal incentives.