Exploring Multiple Autisms through the Lens of Personality: A Latent Profile Analysis

Background: Within the autism spectrum disorders (ASD) population, there is considerable variability in individuals’ symptom expression, verbal/intellectual ability, and comorbid symptoms. This heterogeneity can also be observed in the ASD genetics and neuroscience literature, which presents ASD as a neurodevelopmental syndrome that consists of multiple separable phenotypes with different etiological causes (Geschwind, 2011). Thus, a recent goal in ASD research has been to identify the possible subtypes that are represented under this broad clinical diagnosis. Although the relationship between personality traits and ASD symptomatology has been explored (Lodi-Smith et al., 2018), personality research has yet to be utilized in identifying underlying subtypes and etiological pathways, something which has been effective in other mental health fields. Comprehensive descriptions of personality such as the five-factor model incorporate most areas of human behavioral variability into their taxonomies, providing a unique lens for identifying meaningful subgroup differences with a frame that goes considerably beyond clinical symptoms and theoretical constructs such as social motivation. As such, the intersection between personality research methods and the “multiple autisms” model provides a promising direction in understanding the heterogeneity within the ASD population.

Objectives: To identify meaningful, homogeneous personality subgroups that may be representative of autism subtypes in the ASD population.

Methods: The current study utilized data from a randomized, controlled trial comparing personalized cognitive-behavioral therapy (CBT) to group CBT for school-aged youth with ASD (N=105). A latent profile analysis was conducted using the participants’ baseline personality measure scores (i.e., Hierarchical Personality Inventory for Children). A best-fitting model was determined by relative fit indices, including the Bayesian Information Criterion (BIC) and sample size-adjusted Bayesian Information Criterion (SS-BIC), as well as considerations for parsimony and interpretability. Concurrent validity was assessed by comparing the identified personality subgroups (i.e., classes) on measures of ASD symptomatology and comorbidities (SRS, CASI, CBCL) and measures of cognitive performance (WISC-IV, D-KEFS).

Results: A 4-class solution emerged as the best-fitting model with significant reductions in fit indices through four classes, while the 5-class solution presented an increase in BIC value. The additional class in the 5-class solution was deemed spurious given its similarity to another class and small class membership (representing less than 5% of the sample). The class with the largest membership (n=55) was characterized by low scores across all five personality factors. Another class (n=27) exhibited normative scores in Conscientiousness and Imagination, with low scores in Agreeableness, Extraversion, and Emotional Stability. The third class (n=14) presented very low Extraversion and Imagination scores, while the fourth class (n=9) presented high Imagination scores. The four classes were significantly different in SRS scores, as well as CASI and CBCL subscale scores and several WISC-IV and D-KEFS scores.

Conclusions: Results suggest that subgroups of children with ASD (IQ>70) seeking behavioral treatment may possess distinct personality profiles which may affect the autism symptom expression, severity level, cognitive features, and comorbid symptomatology which characterize them. Future research should determine the clinical significance of identified personality subgroups and whether or not they are identifiable on a neurobiological level as well.