Child Maltreatment (CM) is a major threat to children's health and development with lifetime (0-18) prevalence of investigation, substantiation and foster care placement in the US being about 37%, 12% and 6%, respectively. While not all CM is detected, children with CM reports have markedly worse outcomes in virtually all domains, extending into adulthood. Such a signal provides an opportunity for preventive intervention, but this requires advances in detection for targeting of scarce preventive resources and identifying key points of intervention to guide system response once CM occurs. This project takes advantage of recent advances in computer hardware and software, analysis and state data availability to achieve several aims. Child Welfare DATA SMART (Specification, Management, Analysis, Replication and Transfer) is a five-year, multi-state project that represents the first attempt to create standardized, replicable programming with linked longitudinal administrative data to address several priority areas of the Capstone RFA (HD-18-012) related to identifying CM risk, child welfare response, and longer term outcomes associated with CM. First, we develop similar longitudinal cross-sector datasets (Common Data Set ? Source: CDS-S) in five states (AK, CA, MD, MI, NC) with very different populations and policy structures. This will provide identical variable and data structures to allow replication of data management and analysis across sites.
The second aim i nvolves answering three sets of critical research questions. (1) Prediction cluster: We will explore the potential of Predictive Risk Modeling (PRM) to inform screening for CM risk at the population level as well as CM risk following a hotline. We will evaluate how well PRM's trained in one state or relative to a particular outcome function in other states or with other outcomes such as child injury. The data linkage required to build the PRMs provide the opportunity to address questions in two additional areas: (2) Rare and Understudied Cluster: We will look at rare and understudied populations, types of maltreatment and outcomes to better understand their predictors, occurrence and associated outcomes. (3) Race and Culture: We will look at potential areas of bias in CPS reporting, substantiation and placement relative to race, class and visibility. Programming and methods will be publicly ?transferred? as a national resource in Year 5. Consistent with the mission of the Center for Innovation in Child Maltreatment Policy, Research, and Training (CICM), this project will advance science related to the use of linked administrative data for screening and targeting of services and develop a set of creative dissemination and training products to prevent CM and promote healthy development of victims of CM.
