Strategies for Researching the Impact of Pay for Performance Programs IV

Design Challenges in Field-based Research in Pay for Performance Meredith B. Rosenthal, Ph.D Harvard School of Public Health June 4, 2007 1 Observational Studies in Health Care Financing and Organization Unlike clinical interventions policy interventions are almost never studied in an RCT Even if we could convince a health plan or health system medical director to randomize (and I have tried) not clear what we would learn by studying interventions “in captivity” We need messy, real-world data 2 Case Example: Studying Pay for Performance in a Medicaid-focused HMO, Hudson Health Plan (HHP) HHP provides coverage for 62,000 Medicaid/SCHIP clients in New York State HHP membership is approximately 50% Hispanic, 25% African-American Network extends over 6 counties in Hudson Valley Network includes health centers and community physicians Mix of capitation, fee-for-service 3 HHP Pediatric Immunization and Screening Bonus Program Target: Complete and timely pediatric immunization and screening defined as: HEDIS plus lead and TB screening Bonus structure: Reports sent to practices monthly on patients who turned 2 in last month Practices must provide documentation of immunization/screening within 90 days to be eligible For each 2 year old fully immunized: $100 For each 2 year old fully and timely immunized: $200 4 Study Design Pre/post analysis at patient level Contemporaneous comparison group (other NYS Medicaid managed care plans) with aggregate data only Subgroup analyses by provider type, patient characteristics Physician survey of awareness, perceptions of program, benefits and risks of pay for performance 5 Key Research Questions Can pay for performance that rewards providers a fixed dollar bonus per child improve rates of timely immunization in a Medicaid/SCHIP managed care population? Which providers demonstrate most improvement in immunization under pay for performance? Does pay for performance diminish or increase racial/ethnic disparities? How do providers perceive the HHP pay-forperformance program and what changes do they report undertaking to improve quality? 6 What’s Great about this Project That Every Researcher Should Be So Lucky to Get HHP really cares not only about quality but about evidence-based policy and research The program really should work We have good data on patient demographics including race/ethnicity We have a plausible comparison group, although only aggregate data 7 Challenges, Issues that Arise in Projects Like This Finding a contemporaneous comparison group (creative solutions?) Instability of enrollment/eligibility What if we find nothing??? Environment in which so much else is going on 8