We conducted a secondary analysis of the evaluations of 22 sites participating in four primary care redesign initiatives funded by the Centers for Medicare and Medicaid Services or the Center for Medicare and Medicaid Innovation. Our objectives were to determine the overall impact of the initiatives on Medicare expenditures and whether specific site-level program features influenced expenditure findings. Averaged over sites, the mean intervention effect was a statistically insignificant US$26 per beneficiary per year. Policy implications from meta-regression results suggest that funders should consider supporting technical assistance efforts and pay for performance incentives to increase savings. There was no evidence that paying for medical home transformation produced savings in total cost of care. We estimate that in future evaluations, data from 35 sites would be needed to detect feature effects of US$300 per beneficiary per year.
Keywords: Medicare expenditures; meta-analysis; meta-regression; primary care redesign.