Assessment of Clinical Symptoms and Maltreatment Risk in Child Welfare Service-Planning

Background: Assessment tools have become prominent in decision-making for social workers. In child welfare services, both clinical and actuarial risk assessments have been shown to predict caseworkers’ service-planning decisions (Mendoza, Rose, Geiger, & Cash, 2016). This study examines the relative effects of a clinical assessment of youth behavioral health symptoms and an actuarial assessment of maltreatment risk on caseworker decisions to refer youth to a behavioral health intervention. We hypothesized that clinical symptoms would be the primary determinant of referral, and maltreatment risk would be supplemental such that high risk would justify referral in cases with lower symptoms.

Method: Cognitive Behavioral Therapy – Plus (CBT+) was available to child welfare cases across one county of a Mid-Atlantic state. Caseworkers from the child protective services (CPS), foster care, and family preservation service units administered the Pediatric Screening Checklist-17 (PSC-17) assessment of behavior and trauma symptoms and an actuarial assessment of child- and caregiver-level maltreatment risk factors to inform their service-planning. Study participants include 258 youth who were referred to CBT+ (55%) or not referred because the caseworker did not believe their symptoms warranted the intervention (45%). Assessment and CBT+ referral data were linked with child welfare administrative data. We used binomial logistic regression to estimate the effects of the PSC-17 scores, maltreatment risk scores, and their interaction terms on CBT+ referral, controlling for child age and child welfare unit (CPS, foster care, or family preservation). All assessment scores were standardized for analysis to make their effect sizes comparable.

Results: Higher PSC-17 behavior (OR=8.16) and trauma (OR=4.72) scores were associated with greater odds of CBT+ referral (p<.001). Higher caregiver-level maltreatment risk was associated with greater odds of CBT+ referral that approached significance (OR=1.48, p=.065); child-level maltreatment risk (e.g., impairment of mental, physical, and social wellbeing) was not predictive (p=.933). Child-level maltreatment risk moderated the effect of trauma symptoms on CBT+ referral (OR=3.73, p=.038), whereby an increase in risk was more strongly associated with greater odds of referral among youth with higher trauma symptomology. No other risk-by-symptom interaction effects were observed (ps≥.139).

Discussion: As hypothesized, clinical assessment of behavior and trauma symptoms was the primary determinant of referral to CBT+, with behavior symptoms having the largest effect. Actuarial assessment of maltreatment risk had no significant main effects on referral. Contrary to our hypothesis, there were no interaction effects between clinical symptoms and caregiver maltreatment risk factors. Child-level maltreatment risk moderated the relationship between trauma symptoms and CBT+ referral in the opposite direction that we hypothesized. That is, child-level risk was positively associated with referral more so among youth with higher trauma symptoms than lower trauma symptoms. Thus, maltreatment risk was not a supplemental reason for referral when clinical symptomology was low, as we had predicted. Rather, as child risk factors include indicators of behavioral health impairment, child-level maltreatment risk may serve as a supplemental measure of symptomology to the clinical assessment for informing referral. Overall, findings from this study reflect social workers’ use of assessment tools for service-planning in a child welfare setting.