Policymakers have embraced pay-for-performance and shared savings/risk models to improve healthcare and moderate cost growth in publically funded healthcare programs. There is a strong commitment by both public and private payers to move from volume-based payment toward payment models rewarding value. The Affordable Care Act authorized Medicare to contract with accountable care organizations (ACOs), networks of providers responsible for the health care of a defined population. While implementation of these programs is moving forward rapidly, there is little evidence to guide whether these programs have the potential to yield real cost savings while improving quality. Under the R21 phase of this application, we will complete an evaluation of Medicare's Physician Group Practice Demonstration, a joint pay-for-performance and shared savings effort in 10 multi-specialty physician groups. The Physician Group Practice Demonstration reforms are similar in many respects to the currently legislated ACO reforms, and thus qualify as the best evidence we have to date of the effects of ACO incentives. Yet beyond bonus payments and quality scores publicly reported for each of the ten sites, little is known about how the incentive scheme affected overall costs, quality or outcomes. Our data analysis begins with a simple difference-in-difference strategy, with local Medicare beneficiaries as controls, to measure the effect of the Demonstration on Medicare payments and quality metrics beyond those measured for payment purposes. We will expand the comparison groups to similar physician practices beyond local controls and will focus on accounting for selection into the program and the role of risk adjustment in study of costs and outcomes. In addition, a major objective of our research will be to develop a structural model of Medicare payment reform incentive schemes, with parameters estimated using outcomes arising from the Physician Group Practice Demonstration, that we believe can explain some part of the heterogeneity observed in provider responses to the PGPD. The milestones for completion of the R21 phase will include development of a theoretical model of ACO incentive schemes, completion of appropriate control groups and estimation of a selection model, specification of appropriate quality and outcome measures, and adequate risk adjustment measures. In the R33 phase of the application, we will apply techniques and models developed in the R21 phase to the Pioneer and Medicare Shared Savings Programs, which will include a greater number of accountable care organization participants and more subtleties in incentive design.
Public and private payers are moving towards payment systems with pay-for-performance, shared savings, and risk incentives, yet little is known about how these programs affect the cost or outcomes for patients. The importance of our proposed research is to help policymakers design and prioritize their efforts in payment reform to reduce cost growth and protect against adverse consequences for patients.