Predictive risk modeling (PRM) family investigation risk assessment instrument - the texas department of family and protective services (dfps) investigates every report of child abuse and neglect in…

The Department of Health and Human Services Administration for Children and Families awarded the Texas Department of Family and Protective Services $599,998 through Cooperative Agreement 90CA1908 on September 30, 2026, under the Child Abuse and Neglect Discretionary Activities program (CFDA 93.670).

The award funds a comprehensive review and enhancement of DFPS's Predictive Risk Modeling (PRM) Family Investigation Risk Assessment Instrument, a scoring tool caseworkers use at the end of child abuse and neglect investigations to estimate family likelihood of future maltreatment and inform referrals to ongoing services. The tool has been in continuous use for more than a decade. The project will compare the tool's risk scores against families' actual later outcomes using statewide investigation records spanning many years; combine structured case records with caseworkers' written case notes using modern text-analysis methods to incorporate information the original tool could not access; identify where the tool's recommendations could better align with services DFPS can realistically provide; develop prioritized, region-specific recommendations for strengthening accuracy; and produce a roadmap for training staff and deploying an enhanced tool once the state's new case management system advances further. Caseworkers, supervisors, and people with lived experience in the child welfare system will help shape and review findings throughout.

DFPS will partner with the University of Texas at Arlington's Child Welfare Research Center on the analysis and recommendations.

Performance occurs in Texas with a period of performance from September 30, 2026 through September 29, 2029.

Predictive risk modeling (PRM) family investigation risk assessment instrument - the texas department of family and protective services (dfps) investigates every report of child abuse and neglect in the state. At the end of each investigation, caseworkers use a scoring tool to estimate a family's likelihood of future abuse or neglect, which helps inform whether the family is referred to ongoing services. This tool has been in continuous use for more than a decade. Since IT was built, the data available to caseworkers and the technology available to analyze IT have both advanced considerably, creating an opportunity to strengthen the tool using more complete information than was available when IT was originally developed. This project will conduct the tool's first comprehensive review since IT was put into practice, and use that review to guide specific enhancements. Dfps will partner with the university of texas at arlington's child welfare research center to compare the tool's risk scores against families' actual later outcomes, drawing on many years of statewide investigation records. The team will also combine structured case records with the written case notes caseworkers already produce, using modern text-analysis methods to draw on information the original tool had no way to use. Building on this expanded picture, the project will identify where the tool's recommendations could better align with the services dfps is realistically able to provide, and will develop specific, prioritized suggestions for strengthening its accuracy - accounting for differences across texas's regions. IT will also produce a roadmap for training staff and rolling out an enhanced tool once the state's new case management system is further along. Caseworkers, supervisors, and people with lived experience in the child welfare system will help shape and review the findings throughout, to ensure any recommended changes work well in practice. This project does not build or deploy a new tool. Its purpose is to modernize the evidence base behind an existing one, expand the information available for future improvements, and produce a tested set of recommendations for dfps to act on next.