Early Impacts of the Comprehensive Primary Care Initiative: Effects on Medicare Fee-For Service Beneficiaries' Cost, Service Use, and Quality of Care
The Comprehensive Primary Care (CPC) initiative, which began in October 2012, tests the effects of substantial financial support, performance feedback, and technical assistance to primary care practices on practice transformation and patients’ Medicare expenditures, service use, and quality of care. The initiative was implemented in 497 primary care practices in 7 regions. Medicare and 29 other payers provided supplemental care management fees that were equivalent to 14 to 19 percent of practice revenue for the median practice over the first two years of CPC.
We estimated CPC’s effects over the first two years on Medicare Fee-For-Service (FFS) costs per beneficiary per month; annualized service utilization, including hospitalizations and outpatient emergency department (ED) use; and claims-based measures of quality of care processes and outcomes. We used a matched comparison group design and estimated difference-in-differences (DD) regressions to compare changes in outcomes between the period before CPC began (baseline) and the two years after CPC began for attributed Medicare FFS patients in CPC practices to changes over the same time period for beneficiaries attributed to matched comparison practices. All regressions control for baseline practice and patient characteristics.
The regression-adjusted average Medicare FFS expenditures (excluding CPC fees) increased by $105 (15%) per beneficiary per month from baseline to year 1 for the comparison group, and $91 (13%) for CPC practices. The DD impact estimate, therefore, was a $14 reduction in gross Medicare expenditures during year 1 (95% confidence interval -$27, -$3, p = 0.046). The total difference between CPC and comparison practices, including the care management fees averaging nearly $20 per month, was not statistically significant ($5, 95% confidence interval -$8, $19, p = 0.448). There were also favorable initiative-wide impacts on hospitalizations and outpatient ED visits with reductions of 2 and 3 percent respectively. Year 1 results did not vary by key baseline practice characteristics but did vary by region. The differences were driven mostly by the statistically significant effects on the subgroup of patients comprised of the 25 percent with the highest baseline risk score. Differences on most claims-based quality of care measures were not statistically significant. Although results from year 2 are available and will be presented at the conference, these cannot be disclosed until March 2016, when the results will be made public, as per restrictions imposed by the funding agency (Center for Medicare and Medicaid Innovation at the Centers for Medicare & Medicaid services). The results for Medicare expenditures were robust to alternative model specifications and were driven by a few CPC regions that also experienced differences in service use.
The findings for the first two years of CPC are promising. Additional research is needed to assess how the CPC initiative affects costs and quality of care beyond the first two years, how it affects patient and provider experience, and what types of practice changes are most strongly associated with the strongest performing practices.