Medicaid Accountable Care Model Designs and Maternal Health Measures.

Importance: There is significant heterogeneity in how Medicaid accountable care organizations (ACOs) are designed, but little is known about the association between different Medicaid ACO model designs and maternal health. In 2018, Massachusetts initiated a Medicaid ACO program with 2 distinct ACO models-model A (health system-managed care organization partnership) and model B (primary care practice [PCP]-led)-which may differentially affect coordination, integration, and quality for pregnant and postpartum patients. Objective: To estimate the differential association between Medicaid ACO model designs and maternal health care measures. Design, Setting, and Participants: This cohort study used the 2014 to 2020 Massachusetts All Payer Claims Database on Medicaid-covered deliveries among pregnant patients aged 18 years or older, indexed to the delivery-quarter. Data were analyzed September 2024 to March 2025. Exposure: Attribution to Medicaid ACO model A vs model B vs a non-ACO. Main Outcomes and Measures: Six quality of care-sensitive maternal health measures (severe maternal morbidity, preterm birth, cesarean delivery, timely postpartum care, postpartum depression screening, and postpartum glucose screening) and 2 health care utilization measures (office visits, emergency department [ED] visits) were measured during the prenatal period and at 60 days and 6 months post partum. Using a difference-in-differences (DID) approach, study measures were compared before (2016-2017) vs after (2018-2020) Medicaid ACO implementation among ACO model A vs model B vs non-ACO patients. Models adjusted for patient-level characteristics and included hospital, county, and delivery-month fixed effects. Results: Among 67 204 Medicaid-insured deliveries (mean [SD] maternal age, 28.1 [5.7] years), both ACO model A and model B were associated with increased probability of postpartum depression screening (DID for model A, 6.11 percentage points [pp]; 95% CI. 4.34-7.88 pp; DID for model B, 5.58 pp; 95% CI, 3.83-7.33 pp) compared with Medicaid non-ACO deliveries. Model A was associated with increased probability of a timely postpartum visit (DID, 5.18 pp, 95% CI, 4.34-7.88 pp) and decreased rates of prenatal ED visits (DID incidence rate ratio [IRR], 0.91, 95% CI, 0.84-0.98), whereas model B was associated with increased office visits during the prenatal and postpartum periods (eg, DID IRR for prenatal period, 1.10; 95% CI, 1.07-1.12). Changes in cesarean deliveries were inconclusive, with no other statistically significant changes in outcomes. Conclusions and Relevance: In this cohort study of Medicaid-insured deliveries, there was heterogeneity in the association between Medicaid ACOs and maternal health care measures by ACO model type. States should consider the differential implications of ACO design types and how to best optimize model designs to improve care and outcomes for pregnant and postpartum patients.