Using latent class analysis to identify treatment-use subgroups among parents of children with autism spectrum disorder

Among parents of 2,582 children (ages 4-17 years old) with autism spectrum disorder (ASD), we used latent class analysis to identify subgroups and profiles of treatment users and included annual household income in the specification of the models, then described characteristics of each subgroup. Based on three indicators of fit (Akaike's Information Criterion, Bayesian Information Criterion, and Lo-Mendell-Rubin), six latent classes of treatment users emerged. Subgroups included users of: (a) mostly private and school speech and occupational therapies; (b) nearly all treatment types; (c) mostly speech and occupational therapies, plus intensive behavioral and "other" treatments, but little medication use; (d) private therapies almost exclusively; (e) primarily psychotropic medications; and (f) mostly school-based therapies. Income significantly predicted class differences for all but one latent class. Probabilities of families' lifetime use of nine treatment types varied depending on latent classification. Proportions of families reporting having observed children's developmental regression were largest in those with the highest overall treatment use, and these children also had the lowest cognitive and adaptive-functioning scores and the highest ASD symptom scores. Understanding patterns of treatment use among families of children with ASD is an important first step in enhancing treatment-related selection and implementation. Autism Research 2019, 12: 843-854. (c) 2019 International Society for Autism Research, Wiley Periodicals, Inc.