Assessing Gender Differences in Autism Spectrum Disorder Using the Gendered Autism Behavioral Scale (GABS)

Background: Females with autism spectrum disorder (ASD) are diagnosed later than males with ASD (Lai et al., 2015). Some suggest gendered differences in ASD symptomatology, also known as the female autism phenotype, may contribute to delayed diagnoses for females with ASD (Bargiela et al., 2016). However, there are no standardized measures to assess the presence or absence of the female autism phenotype in individuals with ASD. The Gendered Autism Behavioral Scale (GABS) is a coding frame developed at University College London (UCL), measuring hypothesized components of the female autism phenotype in recordings of ADOS-2 administrations (Loomes, 2016). Preliminary GABS results were promising, with females scoring significantly higher than males (Loomes, Hull, Skuse, & Mandy, 2017). This study assessed the feasibility and validity of the GABS in an entirely new sample, the Rhode Island Consortium for Autism Research and Treatment (RI-CART).

Objectives: To assess the inter-rater reliability of the GABS, and replicate pilot findings that the GABS differentiates phenotypic expression of females versus males from video-recorded Module 3 and 4 ADOS-2 administrations.

Methods: A UCL researcher involved in piloting the GABS trained a RI-CART researcher to code the GABS. The first RI-CART researcher then trained a second, independent RI-CART researcher. During this second training, instances in which the RI-CART researchers could not reach consensus on a given item were resolved by the UCL researcher.

This study analyzed recorded ADOS-2 assessments from males (n = 40) and females (n = 20) aged 4-59 years enrolled in RI-CART, a state-wide, community-based sample. Participants had community diagnoses of ASD and a positive ADOS-2. Female participants were matched with two male participants of similar age, IQ, and/or Vineland ABC scores.

Results: Acceptable inter-rater reliability was achieved between the UCL and first RI-CART researcher (k = 0.69) as well as the first and second RI-CART researchers (k = 0.72). Chi-squares indicated male participants received significantly higher scores than female participants on GABS items C2 (externalising difficulties) X2 (1, n = 60) = 6.56, p = .010 and D1 (reported interests frequency/intensity) X2 (3, n = 60) = 8.81, p = .032, indicating male participants reported more externalizing symptoms and more intense restricted interests than female participants. However, the difference in overall GABS scores between males (M = 14.52, SD = 5.42) and females (M= 16.38, SD= 5.05) was not significant t(58)= 0.784, p = .076.

Conclusions: This study found acceptable inter-rater reliability between GABS assessors across UCL and RI-CART. Validity tests resulted in a partial replication of female vs. male differences on the GABS. This lack of differences may be attributable to an overly small and/or heterogeneous sample. The GABS could provide a means of extracting valuable data on the hypothesized female autism phenotype through a widely employed assessment, the ADOS-2. Future research should assess the GABS’ reliability and validity in larger samples, and consider additional tests of validity such as differentiating females with ASD and a negative ADOS-2 (i.e., false negatives) from females with concordant ADOS-2 and clinical diagnosis.