Adoption, Guardianship, and Post-Permanency Stability: Evidence from Linked Administrative Data across 12 States

Background

When foster care reunification is not viable, agencies seek permanency primarily through adoption and guardianship. Adoption establishes a new and legally permanent parental relationship through termination of parental rights. Guardianship allows for permanent living arrangements without terminating parental rights. Similar to guardianship is a living with other relatives (LWOR) permanency option, which also preserves parental ties but is less common. Although adoption does terminate parental rights, familial ties are not necessarily lost, as many children are adopted to relatives. In order for policymakers to use these options most effectively and weigh their benefits, it is critical to understand how they differ in terms of reinvolvement with the child welfare system (CWS). Despite the urgency of this knowledge, the literature remains inconclusive. This study leverages nationwide administrative data and asks: Accounting for child and foster care system experiences, how do adoption to a non-relative, adoption to a relative, guardianship and LWOR predict CPS reports and foster care reentry?

Methods
Using national longitudinal administrative data, we analyzed foster care exits from the Adoption and Foster Care Analysis and Reporting System (AFCARS) from FY2008–2021, linked to the National Child Abuse and Neglect Data System (NCANDS) through 2023. The sample included 428,893 children exiting to adoption (relative and non-relative), guardianship, and LWOR across 12 states with reliable linkage. Outcomes included (1) rereport to child protective services within two years and (2) foster care reentry within three years. Logistic regression models estimated odds ratios for each outcome, with and without controls for child demographics and foster care experiences, including age, length of stay, and placement history. Models included state and year fixed effects. Flexible parametric survival models assessed how risks varied over time, and state-specific estimates examined heterogeneity.

Results
Descriptive findings show that adoption—particularly non-relative adoption—the lowest rates of rereport (5.3%) and reentry (0.8%), with similar rates for relative adoption (7.3% and 0.7%, respectively). In contrast, guardianship (22.7% rereport; 5.7% reentry) and LWOR (29.0% rereport; 11.3% reentry) exhibit substantially higher rates. Adjusted models confirm these differences. Compared to non-relative adoption, relative adoption is associated with higher odds of rereport (OR=1.43) but lower odds of reentry (OR=0.87). Guardianship is associated with significantly elevated odds of rereport (OR=4.71) and reentry (OR=6.22), while LWOR shows the highest risks (OR=6.22 for rereport; OR=10.53 for reentry). Child characteristics also predict instability: older age and more placement settings are associated with increased odds of both outcomes. Survival analyses indicate that risks are highest shortly after exit—especially for reentry—and decline over time. ORs for relative adoption were not in a consistent direction across states. However, for both guardianship and LWOR, all states had significantly elevated ORs.

Conclusions
Adoption is consistently associated with substantially greater post-permanency stability than guardianship, even after accounting for observable differences in case characteristics. While differences emerge between non-relative and relative adoption, they are small in comparison to the differences between adoption and guardianship, suggesting the story may be less about kinship and more about the legal and structural mandates between the two.