Translational Child Welfare Research: Integrating PCIT Into Foster Care

Background/Purpose:  Children placed in foster care are more likely than their peers in the general population to exhibit externalizing behavior problems such as aggression. Externalizing behaviors are the most common basis for referring young children to mental health services, yet evidence-based interventions that have been shown to ameliorate these problems are seldom available for foster children.

This NIH-funded investigation (R15 HD067829-01A1) tests the efficacy of an innovative adaptation of Parent-Child Interaction Therapy (PCIT), an evidence-based clinical treatment for externalizing problems in young children. In standard PCIT a clinician works with parent-child dyads in a clinical milieu for approximately 12-14 weekly sessions. To address the lack of experiential training for foster parents and to increase foster children’s access to empirically supported services, we developed an adapted PCIT model for foster families that incorporates (a) intensive group-based training, (b) in-home activities, and (c) telephone counseling.

Methods: This study includes 82 children, ages 3-6, residing in a non-relative foster care placement licensed by the Bureau of Milwaukee Child Welfare (BMCW). Using a randomized controlled design, two groups of foster parent-child dyads were assigned to different doses of PCIT training (8 vs. 14 weeks). Both groups were compared to each other and to a third waitlist control group receiving child welfare services as usual.

We created measures of externalizing behaviors at baseline, 8 weeks post-baseline, and 14 weeks post-baseline using parent ratings on two well-validated instruments: (a) Eyberg Child Behavior Inventory (ECBI), and (b) Child Behavior Checklist (CBCL). Responses to the ECBI yielded an Intensity Scale indicating the frequency child problem behaviors and a Problem Scale indicating parental tolerance of the behaviors. We used CBCL responses to create an Externalizing Scale and a subscale for Aggressive Behavior.

We conducted descriptive analyses comparing the study sample to the BMCW child population, and we tested for baseline demographic and behavior differences between study arms using univariate ANOVAs. We then analyzed intervention effects under intention-to-treat assumptions using Multilevel Modeling (MLM), which accounts for the longitudinal data structure and repeated measures design. Multilevel models produced tests of interactions between dummy-coded vectors representing treatment comparisons with a contrast representing time.

Results:  Descriptive analyses showed that the sample and the BMCW child population were comparable demographically. Univariate ANOVAs confirmed that there were no significant differences between study arms at baseline. MLM analyses revealed significant group-by-time interaction effects indicating that both treatment groups improved relative to the control group on the ECBI Intensity Scale (p<.001) and Problem Scale (p=.046) as well as the CBCL Externalizing Scale (p=.008) and Aggressive Behavior subscale (p=.004). The two treatment groups did not differ significantly.

Conclusions/Implications: Findings suggest that, when compared to usual services, experiential PCIT training significantly reduced child behavior problems. These effects may promote short-run placement stability, and they may reduce long-term costs due to mental health and social services. More broadly, this study recommends investing in translational research that adapts and tests clinically validated treatments such as PCIT in community-based settings so that they may be integrated into child welfare services.