Piloting the Linkage of Child Welfare and Medicaid Data

Facilitating future research on child and caregiver outcomes through the creation of a multi-state linked dataset

Parental substance use is associated with a greater risk of involvement with the child welfare system and placement of children in foster care. Referrals to the child welfare system with parental substance use increased between 2005 and 2018, expanding the need to provide parents with specialized inpatient and outpatient services for substance use disorders and comorbid mental illness.

The Family First Prevention Service Act was passed in February 2018 to help children remain at home by incentivizing states to expand access to mental health services, substance use treatment, and skill-based programs for caregivers. To support families effectively, a comprehensive picture is needed of the combined needs of children and their caregivers. Linking state child welfare and Medicaid datasets is a promising strategy, providing data that help agencies coordinate service delivery, identify gaps in services, and evaluate the impact of interventions on child and caregiver outcomes.

The Child and Caregiver Outcomes Using Linked Data (CCOULD) project, funded by the Office of the Assistant Secretary for Planning and Evaluation (ASPE) and the Administration for Children and Families (ACF), piloted the linkage of child welfare and Medicaid datasets in a few states to demonstrate feasibility. This proof-of-concept project was intended to strengthen the infrastructure of state-level data to

• help states link their child welfare and Medicaid data,
• determine the utility of a linked dataset, and
• provide state agencies and the research community with data to study the relationship between services and outcomes.  

Creating a Roadmap for Future Data Linkage Projects

After a robust state recruitment effort by RTI International, ASPE, and ACF, Florida and Kentucky signed on to participate in the pilot project. In Florida, the CCOULD project was managed externally by the University of Florida’s (UF’s) Department of Health Outcomes & Biomedical Informatics; in Kentucky, the project was coordinated internally by the Cabinet for Health and Family Services’ Office of Health Data and Analytics (OHDA).

RTI experts provided technical assistance to UF and OHDA to create their respective linked datasets using a deterministic matching approach. Before data linkage by UF and OHDA, RTI researchers developed a common data model with uniform variables and file structure, so that Florida and Kentucky’s databases could be combined. After data linkage, experts at RTI harmonized the data, correcting for any inconsistencies, and created a multi-state deidentified dataset.

The linked dataset covered the period 2016 to 2021 and included data from

• children with a child welfare report that was linked to a Medicaid record,
• caregivers listed in a child welfare report that was linked to a Medicaid record, and
• a comparison sample of children and adult Medicaid beneficiaries not in the child welfare system.

Building on the pilot project in Florida and Kentucky, RTI researchers developed resources to guide states that launch data linkage projects in the future. This includes a lessons learned report that details the methodology used by UF and OHDA, the results in each state, the challenges faced throughout the CCOULD project, and the solutions to those challenges.

Promoting Continued Research with a Multi-State Dataset

The CCOULD project dataset is housed on the National Data Archive on Child Abuse and Neglect, a well-established archive funded by the ACF’s Children’s Bureau that serves as the repository for several other large child welfare datasets. Through the data archive, researchers can access the deidentified dataset of Florida and Kentucky’s linked data to conduct analyses examining the relationships between Medicaid services, child welfare services, and child and caregiver outcomes.