WITHDRAWN: Complex Households, Complex Decisions: Linking Administrative Child Welfare Data to Understand the Iterative Process of Maltreatment Assessment

Background: Public child welfare agencies generate vast bodies of administrative data that practitioners and scholars can integrate in order to generate evidence that guides preventive intervention. The data are complex however; a given client record may contain several iterations of a single assessment, updated as family circumstances evolve and workers gather information. Indeed, a more complicated case may require a recursive assessment process before a worker reaches a final safety determination. We hypothesize that presenting risk factors such as parental mental health, developmental, or cognitive problems (MH) and/or substance abuse (SA) may complicate the safety assessment process, resulting in multiple assessments and changes in safety determinations. Methods: San Francisco’s public child welfare agency administers all procedures involving child maltreatment referral, screening, and assessment, among other services. The agency uses the Structured Decision Making^® (SDM) System to organize information and guide decisions relating to maltreatment referrals. After linking SDM^® referral and assessment databases, we examined all households involving maltreating parents referred for the first time between January 2007 and February 2018 who received one or more SDM^® safety assessments within 10 days of referral, and whose final assessment was completed within 30 days of the first assessment (n=7,890). Of these households, 12% received at least one additional assessment. We used logistic regression to examine whether specific presenting risk factors, including MH and SA, were associated with higher odds of multiple safety assessments and safety determination reversals. Results: Parents with MH were twice as likely as their counterparts to receive multiple safety assessments (OR=2.03, p<0.001) and twice as likely to have a determination reversed from safe to unsafe (OR=2.19, p<0.05). Parents with SA were twice as likely as their counterparts to receive multiple safety assessments (OR=2.19, p<0.001) and nearly three times as likely to have a determination changed from safe to unsafe (OR=2.75, p<0.001). Implications: The analysis identifies a profile of presenting risk factors that increases the likelihood of an iterative safety assessment process in which households initially thought to be safe are eventually determined unsafe. Awareness of these risk factors may help guard against false negatives during safety assessment, thereby preventing maltreatment re-referral among children deemed safe in the home.