Prevalence and Placement Patterns of Children in Foster Care With Autism Spectrum Disorder

Background and Purpose:  This study examines the prevalence of children with autism spectrum disorder (ASD) entering one large state child welfare system, and focuses attention on their pathways through the child welfare system.  Autism spectrum disorder is a chronic developmental disorder estimated to affect 1 in every 88 children in the U.S.  Less is known about the prevalence of ASD among children served in the child welfare system, or the ways in which large service systems respond to their needs.  To identify placement patterns at the front end of entry, the study uses latent class analysis (LCA).  Previous research has used a combination of methods to identify common patterns of placement.  This paper will be among the first to apply LCA to a placement pattern analysis. 

Methods:  The study uses linked administrative data in the Integrated Database on Children’s Services (IDB) in Illinois (IL) (Goerge, Van Voorhis, & Bong Joo Lee, 1994).  The IDB is a relational database that combines administrative data from multiple state agencies.  This study relies on linked data on children and their placement experiences from the IL DCFS and paid medical claims data for Medicaid services from the IL Department of Healthcare and Family Services.  The study population (n=9,853) includes all children (3-18) entering foster care for the first time between FY 2006-2010 as a result of substantiated abuse or neglect.  The LCA is based on the subpopulation that remained in care for at least 18 months (n=5,423).  Categorical indicators used in the LCA include first type of placement, predominant placement type, restriction-level, stability-level, and any adverse events.  The presence of a single latent class structure for children with and without ASD is assessed both statistically and through visual inspection. 

Results:  The data do not support a single LCA for both the ASD and non-ASD groups.  Goodness of fit statistics support the presence of a 3-class model for children with ASD, and a 4-class model for children without ASD.  The hypothesis of measurement invariance for a 3-class model across groups was also rejected.  Class 1 (High need; 34%) consists of children initially placed in congregate care, with the highest rates of adverse events, and few less restrictive placements.  Class 2 (Step-up; 37%) consists of children initially placed in family settings and stepped-up to more restrictive care where stability is reached.  Class 3 (Stable; 29%) consists of children placed in family settings who achieve stability early and remain in less restrictive settings.  The analysis of covariates reveal that children ages 11 and older and children living in Cook County with ASD were less likely to be in the ‘Step-up’ or the ‘Stable classes compared to the ‘High Need’ class. 

Conclusions and Implications:  Children with ASD were unique in terms of their “High Need” class when compared to the overall population of children in foster care.  Both “High Need” and “Step-up” children with ASD warrant special attention from the child welfare system.