Objective:
Autism spectrum disorder (ASD) diagnosis relies on clinical observation and documentation, but the presence of comorbidities can affect diagnostic validity across clinicians and exacerbate access to timely care. This study used latent class analysis to optimize subgroup identification based on functional level and associated comorbidities using the Behavior Assessment System for Children, Third Edition (BASC-3), and Vineland Adaptive Behavior Scales, Third Edition (Vineland-3), in a pediatric population referred for autism evaluation.
Methods:
This retrospective study reviewed clinical data extracted over a 3-year period (2018–2021). A latent class analysis was used to explore the presence of latent groups guided by the likelihood ratio test and fit indices. Additional analyses contrasted ASD and non-ASD groups on the BASC-3 and Vineland-3 variables.
Results:
There were 191 included participants (mean age 65.9 months, 76.4% male), of whom over half (60.7%) had an ASD diagnosis. Using 185 cases, the exploratory latent class analysis showed the emergence of 4 distinct subgroups. Composition of classes varied on ASD diagnosis, neurodevelopmental difficulties, behavioral health concerns, and intellectual disability. When contrasting ASD and non-ASD groups, significant between-group differences were observed across Vineland-3 variables and BASC-3 adaptive skills subscales indicating poorer social and adaptive functioning.
Conclusion:
Latent class analysis of commonly used behavioral and adaptive measures can help distinguish between subgroups of pediatric patients referred for ASD evaluations and assist in triage of cases based on severity.