Implemented in 18 regions, Comprehensive Primary Care Plus (CPC+) was the largest US primary care delivery model ever tested. Understanding its association with health outcomes is critical in designing future transformation models.
To test whether CPC+ was associated with lower health care spending and utilization and improved quality of care.
Difference-in-differences regression models compared changes in outcomes between the year before CPC+ and 5 intervention years for Medicare fee-for-service beneficiaries attributed to CPC+ and comparison practices. Participants included 1373 track 1 (1 549 585 beneficiaries) and 1515 track 2 (5 347 499 beneficiaries) primary care practices that applied to start CPC+ in 2017 and met minimum care delivery and other eligibility requirements. Comparison groups included 5243 track 1 (5 347 499 beneficiaries) and 3783 track 2 (4 507 499 beneficiaries) practices, matched, and weighted to have similar beneficiary-, practice-, and market-level characteristics as CPC+ practices.
Two-track design involving enhanced (higher for track 2) and alternative payments (track 2 only), care delivery requirements (greater for track 2), data feedback, learning, and health information technology support.
The prespecified primary outcome was annualized Medicare Part A and B expenditures per beneficiary per month (PBPM). Secondary outcomes included expenditure categories, utilization (eg, hospitalizations), and claims-based quality-of-care process and outcome measures (eg, recommended tests for patients with diabetes and unplanned readmissions).
Among the CPC+ patients, 5% were Black, 3% were Hispanic, 87% were White, and 5% were of other races (including Asian/Other Pacific Islander and American Indian); 85% of CPC+ patients were older than 65 years and 58% were female. CPC+ was associated with no discernible changes in the total expenditures (track 1: $1.1 PBPM [90% CI, -$4.3 to $6.6], P = .74; track 2: $1.3 [90% CI, -$5 to $7.7], P = .73), and with increases in expenditures including enhanced payments (track 1: $13 [90% CI, $7 to $18], P < .001; track 2: $24 [90% CI, $18 to $31], P < .001). Among secondary outcomes, CPC+ was associated with decreases in emergency department visits starting in year 1, and in acute hospitalizations and acute inpatient expenditures in later years. Associations were more favorable for practices also participating in the Medicare Shared Savings Program and independent practices. CPC+ was not associated with meaningful changes in claims-based quality-of-care measures.
Although the timing of the associations of CPC+ with reduced utilization and acute inpatient expenditures was consistent with the theory of change and early focus on episodic care management of CPC+, CPC+ was not associated with a reduction in total expenditures over 5 years. Positive interaction between CPC+ and the Shared Savings Program suggests transformation models might be more successful when provider cost-reduction incentives are aligned across specialties. Further adaptations and testing of primary care transformation models, as well as consideration of the larger context in which they operate, are needed.