An Exploratory Factor Analysis On the 3di Supports the Proposed DSM-5 Model

Background: In light of the newly proposed DSM-5 criteria, there is a growing interest in the underlying factor structure of autism spectrum disorders (ASD). A number of studies applied confirmatory factor analysis to investigate this structure on the basis of Autism Diagnostic Interview (- Revised) or Developmental, Dimensional and Diagnostic Interview (3di) domain scores. Yet, these studies are based on a priori theoretical perspectives, because they actively impose a model on the data by (1) using confirmatory factor analysis, and (2) basing the analysis on domain scores which represent DSM-IV-TR diagnostic criteria (operationalized in a number of questions/items).

Objectives: We aimed to explore the factor structure of ASD starting from 3di subscale scores without imposing any a priori theoretical restrictions. Doing this, we want to explore the internal consistency of DSM-domain content and the validity of the two-domain structure proposed in DSM-5.

Methods: The 3di was administered to parents of 275 Dutch speaking participants with ASD (59 girls, 216 boys; age range 4 to 22 years; 16 with intellectual disability, 269 high functioning). Thirty-seven subscales of the 3di were entered in an exploratory principal components factor analysis with varimax rotation: the 36 items used in the original diagnostic algorithm (based on DSM-IV-TR criteria) and one additional item targeting sensory sensitivity (which is proposed as an additional diagnostic criterion for ASD in DSM-5).

Results: Based on the scree rule, a two factor model was proposed. The first factor consists of subscales targeting non-verbal social behavior, sharing, offering comfort, imaginative play with peers, social overtures, conventional gestures, reciprocal conversation and imaginative play on one’s own and imitation. The second factor comprises the subscales targeting stereotyped, repetitive or idiosyncratic speech, ritualistic behavior, mannerisms, preoccupation with non-functional aspects of objects or their parts and auditory sensitivity.

Conclusions: This exploratory factor analysis is an empirically driven attempt to capture the underlying phenotypical dimensions of ASD by investigating the coherence between individual 3di subscales. The findings generally support the proposed DSM-5 two domain model: (1) a merging of scales targeting deficits in social communication and interaction into one factor, and (2) a pooling of scales targeting restricted, repetitive patterns of behavior, interests or activities, including idiosyncratic, stereotyped and repetitive speech, and some aspects of atypical sensory processing. Contrary to the DSM-5 proposal, our data suggest to incorporate imaginative play and imitation as aspects of the social communication and interaction domain.