Latent Classes Among Substance-Involved Families in Child Welfare: Effects on Treatment Completion and Reunification

Background/Purpose: While often treated as a single group in previous studies, families involved with the child welfare system (CWS) because of parental substance use disorder (SUD) are not uniform. A growing line of research suggests these families simultaneously face variable co-occurring risk and need characteristics that affect their pathways into and through child protective services (CPS). Understanding these differential risk and need factors is important for effective case planning and service provision. The generalizability of existing research on this topic is often limited by geographic and/or demographic homogeneity, small sample sizes, or short observation windows.

Our study objectives were to examine latent risk/need profiles among families with parental SUD in child welfare and compare treatment and reunification outcomes across classes. We addressed the following research questions: 1) Do classes exist among substance-involved families with children in foster care?; 2) Do differences in likelihood of treatment completion exist for identified classes, controlling for demographic characteristics?; and 3) Do differences in likelihood of reunification exist for identified classes, controlling for demographic characteristics and treatment completion?

Methods: This study included N = 13,291 parent-child dyads (one focal child and one primary caregiver per family) served by the Regional Partnership Grant (RPG) program, which funded implementation of 53 demonstration projects from 29 states between 2007 and 2012. RPG projects aimed to increase well-being and permanency for children in CWS due to parental SUD. Grantees collected identical demographic, SUD treatment, and child welfare data.

Latent class analysis was used to identify groupings among the study sample on the basis of the following nine key parent-level risk and need factors: marital status, education, living arrangement, employment status, recent arrests, prior CWS involvement, treatment level of care, recent methamphetamine use, and recent opioid use. Bivariate tests explored differences in adult, child, treatment, and child welfare characteristics across classes. Generalized estimating equations (GEE) regressions examined likelihood of treatment completion and reunification according to latent classes, controlling for adult sex, adult race/ethnicity, adult age, child age, and treatment completion.

Results: Model fit statistics supported a four-class solution. The four classes were characterized by high SUD treatment need without CPS risk (class 1), high SUD treatment need with CPS risk (class 2), low SUD treatment need with high criminal risk (class 3), and low SUD treatment need with moderate CPS and criminal risk (class 4). The GEE model suggested classes 2 and 4 had higher odds of completing treatment compared to classes 1 and 3. Children in class 4 were more likely to be reunified than children in the other classes. No differences were found between classes 1, 2, and 3 regarding the likelihood of reunification.

Conclusions/Implications: Heterogeneity exists among families involved with the CWS due to parental substance use. The findings suggest, however, that they present with distinct patterns of risk/need characteristics. These combined characteristics can predict the likelihood that a parent completes SUD treatment and is reunified with their child. Comprehensive and collaborative service delivery must respond to the varied risks and needs of these families.