Developing and Validating a Predictive Risk Model for Youth Placement in Residential Care to Support Decision-Making under the Family First Prevention Services Act

Abstract

ABSTRACT Under the Family First Prevention Services Act, federal Title IV-E claiming requirements for youth who are placed in Qualified Residential Treatment Programs (QRTPs) are more stringent than previous criteria for residential care. Independent Qualified Individuals are required to assess the appropriateness of QRTP placements and judicial reviews must formally approve these placements. Child welfare agencies can use administrative data and empirical methods to adapt to these changes from a prevention standpoint. The current study detailed an approach for developing and validating a Cox proportional-hazards model predicting time-to-first placement in residential care among 43,770 legal spells for youth under the care of one large Midwest child welfare agency. Internal validation using 200 bootstrap resamples yielded stable predictors consistent with those in the literature, strong discriminative ability, and high prediction accuracy. External validation applied the model to examine the risk of residential care placement in the first 90 days of youth’s legal spells. This proactive risk prediction encapsulates a preventive approach preceding the 90-day timeline of QRTP assessment and court review. Results suggest that robust prediction of existing practice of residential care placement has the potential to inform caseworker-level service and placement planning, and build system-level capacity for residential and community-based care.