Timing of Permanency Exits From out-of-Home Care

Purpose: In the child welfare system, how long a child is in substitute care before achieving permanency is a prime concern and is a barometer of performance. There is great variation in length of stay for children in out-of-home care, some of which is linked to type of permanency children achieve – reunification, adoption, or guardianship. Furthermore, children experience “competing risks” between these types of exits. We focus on how the characteristics of children, families, and the child welfare system may have varying effects on the timing of competing exit types. We also examine how the process may be different for children involved with the courts.

Method: We use eight years of administrative data on all first entries into out-of-home care for children in Washington State (N = 37,835). We follow children entering care in 2001 to 2007 through the end of 2008. Data come from the Case and Management Information System (CAMIS) provided by the Department of Social and Health Services (DSHS) and from the Administrator of the Courts (AOC). In event history models for competing risks, we examine the relationships between the characteristics of children and the child welfare system, and the processes of family reunification, adoption, and guardianship. We stratify the models by whether or not a dependency petition has been filed. Fixed covariates include the child's sex, age at entry, race, reasons for removal, and presence of siblings in the system. Time-varying covariates include the type of placement, number of moves, and the flow of children into the system in each county.

Results: For children who are not court-involved the incidence of reunification is over 90% several months into out-of-home placement. For court-involved children, however, reunification reaches only 30% by the end of year one, and is slightly less than 50% by the end of year two. Most covariates predicting time to reunification work similarly for children with and without a dependency petition. For example, children removed for neglect or parental drug abuse are slower to reunify. Two notable differences, however, are linked to dependency status. Court-involved African-American children appear to reunify more slowly than white children, but reunify at similar rates if they are not court involved. And court-involved children placed with an unlicensed relative appear to reunify at a similar rate to children in traditional foster homes, but reunify more slowly if they are not court involved. All children of color are slower to exit to adoption and guardianship.

Conclusion & Implications: Children with court involvement reunify on vastly different timelines than non-court-involved children. Whereas most covariates operate similarly, court-involved African-American children are slower to reunify, but reach permanency by other measures no more quickly, leading to disproportionality. Unlicensed relatives do not slow down reunification for court-involved children, but do for children who have yet to become entangled with courts. Possible explanations include that children are only placed with relatives when workers expect a lengthier stay or workers perceive less urgency to reunify.