Grella, C.E., Needell, B., Shi, Y., & Hser, Y.I. (2009). Do drug treatment services predict reunification outcomes of mothers and their children in child welfare? Journal of Substance Abuse Treatment, 36(3), 278-293.
Parental substance abuse has been associated with high rates of child maltreatment, and therefore a high likelihood of placement in out-of-home care. This study of the effect of mothers’ participation in substance abuse treatment on reunification with their children was part of the California Treatment Outcome Project (CalTOP). Data were provided by 13 counties that volunteered for the project (out of 58 within the state).
The study sample consisted of 1,115 mothers in 43 CalTOP programs; all had at least one child younger than 18 placed in out-of-home care prior to her admission into substance abuse treatment. Reunification was the binary outcome variable; a large number of possible predictors, previously identified in the literature, were used in the analysis:
- Mother’s demographics and scores on the Addiction Severity Index (ASI-Lite, administered at admission).
- Mother’s referral source and treatment modality (from CalTOP records).
- Type and number of services received by mothers (from a survey of program administrators).
- Child demographics and placement history (from Child Welfare Service records).
A correlational design was used; the elaborate data analysis involved hierarchical linear modeling. Three models were tested, all included the same mother, child and basic drug treatment variables, but different sets of service variables: 1) education/employment, 2) family/children, and 3) psychiatric/mental health.
A number of mother and child variables were found to be significant. For example, reunification was less likely for mothers with more employment and psychiatric problems, more likely for older children. Neither mother nor child ethnicity was a significant predictor. More importantly, there were significant treatment effects: completion of 90 or more days in treatment approximately doubled the chance of reunification. Programs with a “high” level of family-related or education/employment services were approximately twice as likely to lead to reunification as “low” level programs. No significant effects were found for psychiatric/mental health services. The authors point out the range of treatment needs for these mothers to help them achieve economic stability and improve parenting skills; they argue that substance abuse treatment can be a useful way to address these needs.
Strengths of this study were the large sample, data integrated from multiple sources, extensive list of control variables, and well-planned statistical design. Also worth noting was the linking of mothers’ CalTOP data with their children’s child welfare records; this was accomplished by a multivariate matching procedure called probabilistic record linkage. An important finding was the effectiveness of programs that provided a range of comprehensive services (i.e., “wraparound services”) in addition to the core elements of drug treatment. However, there were methodological weaknesses. The data about services offered in the various treatment programs was based on the self-report of programs administrators; it isn’t clear to what extent participants actually received the services listed. Also, in spite of the careful statistical controls, selection effects are always possible in a correlational design like this; an experimental design would have been much stronger.