Background: Autism spectrum disorder (ASD) is characterized by notable phenotypic heterogeneity, which is often viewed as an obstacle to the study of its etiology, diagnosis, treatment, and prognosis. On the basis of empirical evidence, instead of three binary categories, the upcoming edition of the DSM 5 will use two dimensions – social communication deficits (SCD) and fixated interests and repetitive behaviors (FIRB) – for the ASD diagnostic criteria. Building on this proposed DSM 5 model, it would be useful to consider whether empirical data on the SCD and FIRB dimensions can be used within the novel methodological framework of Factor Mixture Modeling (FMM) to stratify children with ASD into more homogeneous subgroups.
Methods: The study sample consisted of 391 newly diagnosed children (mean age 38.3 months; 330 males) with ASD. To derive subgroups, data from the Autism Diagnostic Interview-Revised indexing SCD and FIRB were used in FMM; FMM allows the examination of continuous dimensions and latent classes (i.e., categories) using both factor analysis (FA) and latent class analysis (LCA) as part of a single analytic framework.
Results: Competing LCA, FA, and FMM models were fit to the data. On the basis of a set of goodness-of-fit criteria, a ‘two-factor/three-class’ factor mixture model provided the overall best fit to the data. This model describes ASD using three subgroups/classes (Class 1: 34%, Class 2: 10%, Class 3: 56% of the sample) based on differential severity gradients on the SCD and FIRB symptom dimensions. In addition to having different symptom severity levels, children from these subgroups were diagnosed at different ages and were functioning at different adaptive, language, and cognitive levels.
Conclusions: Study findings suggest that the two symptom dimensions of SCD and FIRB proposed for the DSM 5 can be used in FMM to stratify children with ASD empirically into three relatively homogeneous subgroups.