Background: Coordination of inpatient and outpatient care is an important challenge in improving population health but evidence examining the effectiveness of existing care coordination programs is mixed. The Comprehensive Care Physician (CCP) Program at the University of Chicago provides patients at increased risk of hospitalization the opportunity to receive inpatient and outpatient care from the same physician. We compared patient satisfaction, self-related health general and mental health status, and self–reported hospitalization rates of Medicare patients randomly assigned to the CCP program vs. standard care (SC) in which patients receive inpatient care from hospitalists and outpatient care from a primary care physician who does not care for them in the hospital.

Methods: Two-thousand Medicare patients with at least 1 hospitalization in the past year or in the emergency department at the time of recruitment were randomly assigned in equal proportions to CCP or SC between November 2012 and June 2016.Patients were surveyed every 3 months by telephone for a minimum of 1 year and maximum of 5 years to assess patient experience with their primary physician, general and mental health status, and hospitalization rate. Longitudinal outcomes were analyzed using mixed-effect regression models.

Results: At baseline, mean age was 63 years, 62% were female, 88% were black, and 45% were dual-eligible. There were no statistically significant differences in demographic or health measures between CCP and SC patients at baseline. Follow-up rates to 1 year were 95% for CCP and 85% for SC. Mean HCAHPS ratings of their physicians were 0.27 points higher for CCP vs. SC patients (p<0.0001,95%-CI:[0.16, 0.37]), corresponding to the difference between the 80th percentile and 95th percentile in such scores nationally. Mean self-rated health status measured from 1(poor) to 5(excellent), was not significantly different for CCP vs. SC for general health (DCCP-SC=-0.001, p=0.9701, 95%CI:[-0.06, 0.06]), but were 0.11 higher for CCP compared to SC mental health (p=0.0033,95%CI:[0.03, 0.18]). Using a zero-inflated Poisson mixed-model, the rate of hospitalization was 22% lower and statistically significant (p=0.030, event rate ratio 0.78, 95%CI:[0.62,0.98]) for CCP compared to the SC at the first 3-month follow-up wave and remained at least 15% below SC and statistically significant up to the minimum 1 year follow-up.

Conclusions: Patient-reported experience with their physician and mental health status were significantly higher, and patient-reported hospitalization rates were substantially lower for CCP patients vs. SC patients over the year following randomization to CCP vs. SC. Correlation of these findings with objective measures of utilization, such as claims data, and further follow-up is warranted. These findings suggest that the CCP model may improve patient experience and health status while substantially reducing utilization for patients at increased risk of hospitalization. Given the limited evidence supporting the effectiveness of existing care coordination programs in improving outcomes and reducing costs and given the need for effective approaches to population health management, the CCP model warrants further exploration through efforts to implement it in additional settings and rigorously evaluate is effects on outcomes and costs.