Study 1
Brook and colleagues (2015) conducted a quasi-experimental design in which families assigned to Family Drug Court in Tulsa, Oklahoma, were matched (via propensity scores) with a comparison group of families in the adoption and foster care system.
A total of 226 treatment cases were available for study during the evaluation period. The full pool of potential comparison cases in Oklahoma’s Adoption and Foster Care Analysis and Reporting System (AFCARS) was reduced through the application of four criteria: 1) dates of children’s removal from their homes or their discharge from foster care; 2) reasons for alleged maltreatment must be due to parental alcohol or drug abuse; 3) age of the child at time of removal (must be under 17); and 4) only non-emancipated children. The remaining cases in the comparison pool were subjected to propensity score matching to the treatment cases on 14 child indicators. This resulted in a potential comparison pool of 3,827 cases. The final comparison group was composed of 418 treatment cases (204 FDC cases were matched with two comparison cases; an additional 10 were matched with only one comparison case, and 12 cases could not be matched and were excluded from further analysis).
After generating the comparison sample, life tables and Cox regression were used to assess whether FDC participants were more likely to reunify with their families of origin and the likelihood of reunification for the FDC group over time compared with the comparison group, after controlling for those covariates. The matched samples were found to have no significant between-group differences at baseline, except for removal time.
The treatment group sample was 49.5 percent female. The racial/ethnic diversity of the children in the treatment sample was 43.5 percent white, 10.3 percent American Indian, 5.6 percent African American, and 40.7 percent other races. Of the total treatment sample, 12.6 percent of the children were Hispanic. The comparison sample was similar, with mostly female children (44.5 percent). Of the comparison sample, 40.2 percent were white, 8.6 percent were African American, 7.4 percent were American Indian, and 43.8 percent were from other races. A total of 11.5 percent of the children in the comparison sample were Hispanic. The mean age of the child at removal was the same for the treatment and comparison cases (3.4 years old).
The primary outcome of interest was family reunification, which was measured though official records. Case outcomes were tracked from January 2011 through September 2013. A Cox regression was conducted to examine the likelihood of reunification over time for the FDC group, compared with the matched comparison group. The variables known to be predictive of reunification were used as control variables in the analysis.