EXPLORING SEX DIFFERENCES IN THE STRUCTURE OF THE ADOS-2 IN AN EARLY INTERVENTION SAMPLE

Abstract

Autism Spectrum Disorder (ASD) is diagnosed using the same criteria for males and females (e.g., DSM-5, ICD-10). Our understanding of ASD, including its etiology, symptom presentation, and prevalence has evolved significantly over time motivating several changes to the diagnostic criteria and the tools with which symptoms are measured. One aspect of ASD prevalence and presentation that has remained consistent overtime and across the world is the consistently disproportionate male-to-female ratio that is reported. This ratio may reflect a true difference in prevalence or suggest that females with ASD are under- or mis-diagnosed, as evidenced by an emerging body of literature suggesting differences in symptom presentation and related abilities across sex. Studies have reached varied conclusions regarding sex differences as measured by standardized autism assessment tools, for example, finding no significant sex differences in Total Scores across measures or finding fewer parent reported repetitive and restricted behaviors (RRBs) in females (Duvekot et al., 2017). Limitations in the current body of research include small sample sizes, inconsistencies in participant ages, and the use of mixed sets of measures with a strong reliance on parent report. Most studies within this body of research use one of three iterations of the Autism Diagnostic Observation Schedule (ADOS). Indeed, the ADOS and its revisions are considered the gold-standard assessment tools for ASD symptoms and in many studies and clinical practice are used almost exclusively to determine whether a child meets diagnostic criteria. Previous studies have explored the factor structure of the ADOS in its current iteration (i.e., ADOS-2) by domain area and between ASD and typically developing or clinical samples of participants (Bishop et al., 2016; Gotham et al., 2007; Gotham et al., 2008). However, to date, no published studies have explored the factor structure of the ADOS-2 across sex with a preschool age sample. The current study examined the structure of the ADOS-2 through exploratory factor analysis to analyze algorithm items and determine if the Social Affect (SA) and Restricted and Repetitive Behavior (RRB) subscales demonstrate a different structure for males and females. A two-factor structure, relatively consistent with the tool itself, was identified for the combined and male samples. However, the female sample produced a three-factor structure, with some loadings inconsistent with the current algorithm of the ADOS-2. Further analysis through confirmatory factor analysis indicated a relatively well-fitting model for the combined, male, and female groups. However, despite adequate model fit, there were notable differences in the strength at which certain items loaded onto each subscale across sex. Similarly, correlation estimates between the two subscales were stronger for the combined and male groups, and relatively weak for the female group. Implications for the use of the tool with females are discussed, along with suggestions for future evaluation of the ADOS-2 algorithm across sex.