Patterns of out-of-home placement decision-making in child welfare

• Published in: Child abuse & neglect Vol. 37; no. 10; pp. 871 - 882
• Main Authors: Chor, Ka Ho Brian, McClelland, Gary M, Weiner, Dana A, Jordan, Neil, Lyons, John S
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.10.2013; Elsevier; Elsevier Science Ltd
• Subjects: Adolescent; Age Factors; Algorithms; Biological and medical sciences; Caregivers; Child; Child Welfare; Child, Preschool; Decision Making; Decision support algorithm; Decision Support Techniques; Female; Foster Home Care; Group Homes; Humans; Illinois - epidemiology; Infant; Infant, Newborn; Linear Models; Log-linear modeling; Male; Mathematics; Medical sciences; Out-of-home placements; Pediatrics; Prevalence; Psychiatry; Psychology. Psychoanalysis. Psychiatry; Psychopathology. Psychiatry; Team decision-making; United States - epidemiology; Victimology; United States; Illinois; Child welfare; Out-of-home placements; Decision support algorithm; Log-linear modeling; Team decision-making; Human; Decision making; Cognition; Victimology; Algorithm; Social assistance; Child abuse; Decision aid; Multidisciplinary
• ISSN: 0145-2134; 1873-7757
• DOI: 10.1016/j.chiabu.2013.04.016

Abstract Out-of-home placement decision-making in child welfare is founded on the best interest of the child in the least restrictive setting. After a child is removed from home, however, little is known about the mechanism of placement decision-making. This study aims to systematically examine the patterns of out-of-home placement decisions made in a state's child welfare system by comparing two models of placement decision-making: a multidisciplinary team decision-making model and a clinically based decision support algorithm. Based on records of 7816 placement decisions representing 6096 children over a 4-year period, hierarchical log-linear modeling characterized concordance or agreement, and discordance or disagreement when comparing the two models and accounting for age-appropriate placement options. Children aged below 16 had an overall concordance rate of 55.7%, most apparent in the least restrictive (20.4%) and the most restrictive placement (18.4%). Older youth showed greater discordant distributions (62.9%). Log-linear analysis confirmed the overall robustness of concordance (odd ratios [ORs] range: 2.9–442.0), though discordance was most evident from small deviations from the decision support algorithm, such as one-level under-placement in group home ( OR = 5.3) and one-level over-placement in residential treatment center ( OR = 4.8). Concordance should be further explored using child-level clinical and placement stability outcomes. Discordance might be explained by dynamic factors such as availability of placements, caregiver preferences, or policy changes and could be justified by positive child-level outcomes. Empirical placement decision-making is critical to a child's journey in child welfare and should be continuously improved to effect positive child welfare outcomes.