UC Berkeley

This dissertation is based on a unique dataset constructed by probabilistically linking records across three independent sources of data from California: 1) vital birth records, 2) administrative child protective service records, and 3) vital death records. The final dataset captures 4.3 million children born in California between 1999 - 2006 and includes maltreatment allegation information for over 500,000 children who were reported to child protective services (CPS), as well as death reports on 2,000 children who were fatally injured before age five. Three research questions were examined in the context of a prospective birth cohort analysis: 1) Is a referral to child protective services an independent risk factor for injury mortality? 2) Is allegation disposition associated with injury fatality risk? 3) Does injury fatality risk vary across maltreatment allegation types? To answer these three questions, a series of multivariate survival models were specified. Separate models were estimated for overall risk of injury death, risk of unintentional injury death, and risk of intentional injury death.

Findings indicate that an earlier report to CPS is the strongest predictor of a child's injury death during the first five years of life. Children previously reported for maltreatment died from accidental injuries at twice the rate of their unreported, demographically similar peers, and from intentional injuries at five times the rate. After adjusting for other characteristics, children whose report of maltreatment was evaluated out without an in-person investigation by CPS died of injuries at significantly higher rates than children who had never been reported. Children with a substantiated allegation of maltreatment and no foster care placement died of intentional injuries at over 10 times the rate of children who had not been reported. Placement in foster care for even a single day was protective. Children with a prior allegation of physical abuse died from injuries at rates that were notably higher than not only unreported children, but also children reported for reasons of sexual abuse, neglect, or other forms of maltreatment. When only intentional injury fatalities were modeled, a prior allegation of physical abuse was associated with a rate of death that was 38 times that of children who had not been reported.

This study represents the most rigorous longitudinal analysis of mortality outcomes following a report to CPS to date, with several key implications for practice and policy emerging. First, these data underscore that a child's report to CPS is not random, nor is it simply a function of poverty. Rather, a report to CPS signals a level of risk, including a risk of death, that is greater than sociodemographic factors would alone predict. A second and related point is that children evaluated out after a CPS hotline call reflect a group whose risk of injury death is far greater than their unreported sociodemographic peers. The decision to screen these children out without an investigation, under the logic that these children were assessed to be at no greater risk of harm than other demographically similar children, is not supported by the empirical evidence generated from this study. Third, these data highlight that although there has been a recent emphasis on the unmet service needs of children reported for neglect, it is young children reported for physical abuse who face the greatest risk of death. Given that physical abuse allegations represent a minority of reports received by CPS, these data suggest that a different protocol for investigating and intervening in cases in which physical abuse is alleged may be justified. Finally, the finding that a prior allegation of maltreatment is the single greatest predictor of not just intentional injury death, but also unintentional injury death, lends support to calls that have been made for child welfare services to be pursued under a broader, public health-oriented agenda, focused on the prevention of all manners of injury death.