Background: Diabetes costs account for 20% of all health care spending in the U.S. The Accountable Care Communities (ACC) Program was developed by a large U.S. insurer to provide practices with community data analytics for population management and nurses to support care improvement for Medicaid patients. We examined changes in costs and utilization for Medicaid patients with diabetes in ACC compared to non-ACC practices.

Methods: We used interrupted time series (ITS) analysis of mean costs per person-month between 2009-2015. We used data from patients with diabetes in practices with >1000 Medicaid patients. ACC was implemented in 14 states and comparison non-ACC practices were matched on state to control for state-level variation in Medicaid generosity. Separate ITS models were used for costs, ER visits and hospitalizations adjusted for age, gender, race/ethnicity, seasonality, and state by year fixed effects. Results are presented as differences-in-post-intervention monthly changes-in-expenditures (DID) by Medicaid category (TANF, SSI, Expansion and Duals).

Results: The sample included 899,273 ACC-person months from 54,721 Medicaid patients with diabetes. Over 29 months follow-up, ACC assignment was associated with significant increases in total costs at the time of implementation for all Medicaid patients except for those eligible for both Medicare/Medicaid (Duals); increases in total costs at implementation ranged from $65-143/person-month based on Medicaid category, p<0.05. There were no significant differences in ER use or hospitalizations over time for Medicaid patients with diabetes assigned to ACC practices.

Conclusions: Medicaid patients in ACC practices had increases in costs but not utilization over 2 years follow-up, compared to patients in non-ACC practices. Our findings suggest that this ACC program may not reduce costs of care or utilization for Medicaid patients with diabetes.


T. Moin: None. O. Duru: Advisory Panel; Self; Healthwise. C. Mangione: None.


Centers for Disease Control and Prevention; National Institutes of Health (U18DP006128)

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at