**Background:** Maternal mortality in the U.S. has risen steadily from 2000 to 2020 and is over twice that of other high-income countries, with significant racial and ethnic disparities. Medicaid covers nearly half of all U.S. births, including over half of births by Black and Hispanic women. Medicaid claims data contain detailed longitudinal information on diagnoses and healthcare utilization not captured on birth certificates, while birth certificates contain key maternal and infant health characteristics not in claims data (e.g., parental education). Linking these data sources creates a rich resource for research on public health programs, clinical care models, and policy interventions. However, state-level linkage efforts have been fragmented with substantial variation in methods, data sources, and applications.
**Methods:** The study used a three-pronged approach: (1) A structured literature review of peer-reviewed articles published between February 28, 2017 and March 1, 2022, identifying 45 studies from 22 states that linked Medicaid or all-payer claims database (APCD) claims at the individual level with birth certificates or other data. (2) An inventory of state and territory linkage efforts through targeted searches of Medicaid agency websites, vital records websites, health department websites, and Title V MCH Block Grant reports for all 59 jurisdictions. (3) Semistructured discussions with representatives from 9 states with more than 50,000 live births in 2020 that perform ongoing data linkages. States were purposefully sampled by population size and census region. Discussions were audio-recorded, transcribed, and coded using conventional content analysis.
**Key Results:** The literature review identified 45 studies from 22 states. Birth certificates were the most commonly linked data type (75.6% of studies), followed by social services data (15.6%) and death certificates (13.3%). Of the studies, 37.8% linked claims with more than one category of data. Only 55.6% of studies specified linkage type (33.3% deterministic only, 6.7% probabilistic only, 15.6% both), and less than half provided match rates or validation analyses. The state inventory found that of 59 jurisdictions, 33 had in-scope data linkage efforts (28 mentioned in Title V reports, 5 through APCDs), totaling 39 jurisdictions when combined with literature review findings. Only 12.9% of jurisdictions with in-scope linkages publicly reported linkage type, fields used, or match rates. Group discussions revealed that most linkages used probabilistic approaches, were conducted annually, and commonly used mother/child name, date of birth, and Social Security numbers as linking variables. Match rates were almost always over 90%, and every discussion participant believed the linked datasets were of high enough quality for research. Key challenges included data characteristics (e.g., babies' claims billed under mother's Medicaid number, 'Baby Boy'/'Baby Girl' names on birth certificates, difficulty matching twins and out-of-state births), staff turnover, and interagency data access. Facilitators included strong interagency relationships, institutional champions, and dedicated funding.
**Clinical Implications:** Linked Medicaid-birth certificate data have been used extensively to evaluate policies such as Medicaid expansion and social services programs, assess maternal and infant health outcomes (e.g., neonatal abstinence syndrome, breastfeeding rates, low birth weight, maternal depression screening), and target programs to high-risk populations. Some states use linked data for individual patient-level outreach (e.g., contacting women with previous preterm delivery, COVID-19 vaccination follow-up). The study suggests that federal support for technical assistance, best practices, and interagency collaboration could help overcome barriers and enable a more coordinated approach across states, potentially improving maternal and infant health outcomes at a population level.