(WITHDRAWN) A Cohort Study of Mortality Among Youth in Foster Care

Background and Purpose: Tracking mortality rates after child protective services (CPS) involvement has been identified as an important performance measure and research agenda. Our current study examined mortality for recent transition-age youth in-care in California (those in care between 2010 and 2012, and followed through 2017. Risk of death and manner of death were examined based on the youth’s placement status as a nonminor dependent, where state policies that now provide services for youth beyond age 18. We conducted generalized linear models (GLM) and multinomial logistic regression to examine risks associated with child welfare and likelihood of death. The data underscore disparate risk of violent deaths among Black and Hispanic youth and suggest that youth who remain in foster care as nonminor dependents were less likely to die in young adulthood in comparison to youth in care who exit care earlier.

Methods: The universe included in this analysis consists of all CPS youth with placement episodes in foster care in California from 2010 and 2012. Youth were between the ages of 10 and 18, allowing for a longitudinal examination this population of at-risk youth. Identified youths (n=31,477) were linked to vital death records using a probabilistic, machine learning linkage program called ChoiceMaker. Youth were followed prospectively through 2017 to identify transition age youth who died in young adulthood. Demographic and case characteristics of all youths and the nature of death among decedents were examined using χ² tests. We conducted GLM to identify characteristics related to death and multinomial logistic regression to stratify results by manner of death (e.g., accidental, homicide, other).

Results: Among decedents (n=260), 45% were coded homicide and 32% were accidents. Statistically significant variations in the manner of death emerged by race/ethnicity, Black and Hispanic youths made up the majority of homicide deaths. The GLM identified increased risk for males, children who were older at time of first placement, and had more placements. Youth with more allegations and who stayed in placements as nonminors had lower risk for death. Multinomial regression identified physical abuse allegations, total number of allegations, and number of placements as increasing risk of homicide; emotional and sexual abuse allegations decreased risk of accidents. Risk of accidents was higher for youth with later age of first placement and more placements.

Conclusions and Implications: This is the first study to examine the risk of death among a population of youth eligible for extended foster care. The data underscore disparate risk of death in young adulthood among Black and Hispanic youth. These results suggest youth who remain in foster care as nonminor dependents may be less likely to die in young adulthood, particularly for homicides. Given the number of decedents, results should be interpreted with caution. The current study is limited by left censored data in child protective service records for older youth. Findings should be considered in light of research suggesting important differences between youth who remain in-care after the 18^th birthday. Future research should include additional cohorts to better understand foster care and mortality risk.