An Examination of Post-Permanency Discontinuity for Children in Adoption Vs. Guardianship Placements Using Propensity Score Matching

Background and Purpose: A growing number of maltreated children and youth now live outside the formal foster care system in adoptive and guardianship homes. For example, from 1998 to 2008 in the U.S., the percentage of children with previous child welfare involvement who were adopted grew from about 7% to 12%. This shift in the balance of substitute care from foster care to adoption and guardianship is generally heralded as a major child welfare achievement. However, a key concern is that placement stability may differ for children in adoption versus guardianship homes. Most post-permanency research to date has focused on a relatively short follow-up period (typically 18 to 24 months) and not tracked children through the age of 18.  This study provides a longitudinal analysis of 22,563 former foster children adopted or taken into guardianship, and followed for at least 10 years, or until the age of 18.

Methods:  Administrative data from Illinois’ child welfare system was used to track longitudinal outcomes.  The population included all children who had been adopted or taken into legal guardianship in Illinois between 1998 and 2002, had an open adoption or guardianship assistance case as of June 30, 2005, and were between the ages of 6 and 17 on June 30, 2006.  Long-term outcomes for these children were followed through December 31, 2012. Children were tracked to determine if they reentered foster care of if their adoption or guardianship subsidy payment ended prior to the child reaching the age of majority. 

For this study, post-permanency discontinuity was analyzed for children in adoption versus guardianship placements.  In addition, nearest neighbor within caliper propensity score matching was used to create matched adoption and guardianship groups (N=2,923 for each group), conditional on several covariates. Covariates included in the propensity score model were service factors (e.g., number of moves in foster care, kinship placement), child factors (e.g., race, gender) and caregiver/family factors (e.g., urban residence). Logistic regression was used to analyze outcomes using the matched sample. 

Results:  The majority of children in the full sample (88%) and in the matched sample (84%) did not reenter foster care after permanency.  Also, in the matched sample, adoption cases were about 33% less likely to experience discontinuity as compared to guardianship cases.

Implications:  Most children in adoptive and guardianship homes do not subsequently experience discontinuity.  The study provides some evidence that children in guardianship placements are at higher risk for discontinuity than children in adoption placements, even in a matched sample.  However, it is unclear whether the higher risk for guardianship discontinuity reflects a protective effect for adoption placement, or selection bias due to factors not included in the propensity score model (e.g., child maltreatment type or behavior problems of the child).