Methods: Administrative data from families involved with Child Protective Services in Kentucky between 2010 – 2016 were analyzed. Propensity score matching was used to match the youngest child in families receiving START in five Kentucky counties to children in families receiving usual services (N = 1042). Eleven theoretically and empirically-supported matching variables were selected, including child-level factors (e.g., age; race), case characteristics (e.g., CPS investigation finding; year of maltreatment event), and family-level factors (e.g., substance use severity; poverty). Outcomes included: (1) placement in out-of-home care; (2) reunification for children who were placed in out-of-home care; and (3) subsequent maltreatment. Group differences were assessed using chi-square tests for categorical variables and t-tests for continuous variables.
Results: After matching, groups were statistically different on 2 of 11 variables used in the matching process; children in the comparison group were more likely to be White than children in START, and START families showed higher rates of substance use than matched comparison families. With regard to study outcomes, children in families receiving START were less likely to be placed in out-of-home care than children in families receiving usual services (20.7% vs. 34.2%; p < 0.001). Of children who were placed in out-of-home care, a higher percentage of those receiving START were reunified with their caregivers, though differences were not significant (50.0% vs. 41.6%; p = 0.26). Rates of subsequent maltreatment were higher among families receiving START than in families receiving usual services (12.3% vs. 6.1%; p < 0.001).
Conclusions: With regard to preventing out-of-home care placements, START outperformed usual services. Additionally, among children who were removed, reunification rates for children in families served by START were near national averages in spite of notable risk factors in START families (e.g., parental substance use and neglect; younger age of child). Finally, the unexpected finding of recurrent maltreatment being higher among children in families receiving START may result from limited specificity in the substance use variable used for matching. Whereas most parents receiving START meet criteria for substance use disorder, substance use severity among comparison families cannot be determined. Another factor that may have led to higher rates of maltreatment among START families is increased surveillance; over one-third of subsequent maltreatment reports for START children were made by social services personnel, which was twice as high as the matched comparison group and three times higher than the national average.