31460
What Is the Optimal Structure of the Autism Phenotype: A Comprehensive Comparison of Dimensional, Categorical, and Hybrid Models
While initially conceptualized as a categorical construct, recent years have seen an emergence in clinical, research, and popular discourse about a continuous (dimensional) “autism spectrum.” More recently, a compelling new possibility has emerged that integrates both categorical and dimensional representations of ASD symptoms (i.e., hybrid model). Nevertheless, there was still a debate about how to best conceptualize ASD phenotypic symptom structure, with multiple categorical, dimensional, and categorical-dimensional hybrid models showing varying degrees of support (Frazier et al., 2012; Georgiades et al., 2007; James et al., 2016). Given this, adjudicating between these proposed structures of ASD symptoms is essential for effective diagnostic classification in the context of research as well as clinical practice.
Objectives:
We sought to delineate the optimal structure of the observable ASD symptom phenotype by comprehensively comparing categorical, dimensional, and categorical-dimensional hybrid models in two large, diverse samples of youth with and without ASD.
Methods:
The primary study sample comprised 3,825 youth, who were consecutive referrals to a university developmental disabilities clinic or a child psychiatric outpatient clinic. We analyzed the ASD symptom rating scale from the parent-report version of the Child and Adolescent Symptom Inventory-4R. A series of latent class analyses (LCA), exploratory and confirmatory factor analyses (EFA and CFA, respectively), and factor mixture analyses (FMA) were compared. Further, including only individuals whose ASD diagnoses were available, we conducted another subset of analyses where we further specified the three following category-based models: (a) diagnostically-driven categorization, (b) empirically-driven categorization, and (c) DSM-5-based-categorization. We also conducted a full replication analysis using a separate large heterogeneous and geographically diverse sample of clinic referrals (N=2,503).
Results:
Overall results indicated that the ASD symptom phenotype was best conceptualized as multi-dimensional rather than categorical (either diagnostically-based categories or empirically-based categories) or as a categorical-dimensional hybrid (Figure 1). ASD symptoms were best characterized as falling along three dimensions (i.e., social interaction, communication, and repetitive behavior). The results of the replication analysis were virtually identical to those of the analyses conducted using the primary dataset, supporting: (a) the superiority of the 3-factor dimensional model of ASD symptoms (Table 1-1), (b) the same top five models (Table 1-1), and (c) that the best fitting models for each approach were almost identical across the two subsets (i.e., the primary and replication analyses) of analyses (Table 1-2).
Conclusions:
In sum, our findings are consistent with the notions that ASD traits are widely distributed among the general population of clinic referrals; there are three core domains of symptoms (social, communication, repetitive behavior); and diagnostic models including their associated symptom clusters are better conceptualized as dimensional. Clinically, these results challenge traditional conceptualizations of autism, spectrum, and disorder, and have important implications for differential diagnosis and trans-diagnostic models of pathogenesis.