Using latent class analysis to identify clinical subgroups and pathways of youth in a therapeutic foster care program

Abstract

Behavioral health assessments often focus on the average status of group members and overlook unique individual needs and attributes. Therapeutic foster care (TFC), an alternative setting to residential care that serves the highest-need population in the child welfare system, is especially vulnerable to a one-size-fits-all approach. Research to date highlights the needs to identify the target population who benefits from TFC, clarify TFC eligibility criteria, and identify the youth’s clinical pathways in TFC. Using a sample of 175 children in a therapeutic foster care program, this study examined the Child and Adolescent Needs and Strengths (CANS) assessment to test a person-centered approach for identifying subgroups of children and to examine the children’s longitudinal behavioral health changes over time. Latent class analysis identified three classes of youth: (1) Some neglect, impulsivity, and medical issues; (2) Developmental and chronic medical issues; and (3) Complex needs with low strengths. A random forest regression model identified seven CANS items most predictive of unresolved needs. Each LCA class had a different distribution on three clinical pathways associated with their overall need reduction: (1) Moderate Improvement; (2) High Improvement; and (3) Low Improvement on the needs at initial assessment, reassessment, and most recent assessment. These findings have direct implications on tailoring TFC treatment planning at the clinician-level and help measure quality improvement at the program- and system-level.