Self-Reported Sex Differences in High-Functioning Adults with Autism: A Meta-Analysis

Background: Sex differences in autistic symptomatology are believed to contribute to the mis- and missed diagnosis of many girls and women with an autism spectrum condition (ASC). Whilst recent years have seen the emergence of clinical and empirical reports delineating the profile of young autistic girls, recognition of sex differences in adulthood is far more limited. We chose to investigate these as they emerge in self-report screening instrument, the Ritvo Autism Asperger Diagnostic Scale-Revised (RAADS-R), which is recommended by the National Institute for Health and Care Excellence (NICE, 2016) for the identification of autistic adults in Great Britain.

Objectives: To compare autistic symptomatology, in autistic men and women, on RAADS-R domains of Social Relatedness, Circumscribed Interests, Sensory Motor (henceforth Sensorimotor), and Language symptoms.

Methods: The analysis employed a two-factorial design: by comparing autistic men and women to each other as well as to typically-developing (TD)men and women, it was possible to tease out normative sex differences in cognition which might or might not be present in autism. To supplement data gathered by our research group, we conducted a meta-analysis of studies which had used the RAADS-R, and obtained a total 961 datasets. Eventually, datasets from 137 TD men and 136 TD women (age-matched to autistic groups), 118 autistic men and 136 autistic women (age-matched to each other) were analysed. Analysis of variance (ANOVA) searched for the presence of main effects of Sex and Diagnosis and for interactions between these factors.

Results: In social relatedness and circumscribed interests, main effects of Diagnosis revealed that as expected, autistic adults reported significantly greater lifetime prevalence of symptoms in these domains; an effect of Sex, in circumscribed interests, also suggested that males generally reported more prevalent symptoms than females. An interaction of Sex and Diagnosis in language symptomatology revealed that a normative sex difference in language difficulties was attenuated in autism. An interaction of Sex and Diagnosis in the sensorimotor domain revealed the opposite picture: a lack of sex differences between typically-developing men and women and a greater prevalence of sensorimotor symptoms in autistic women than autistic men.

Conclusions: Not all childhood sex differences, such as the male overrepresentation in repetitive behaviours and interests and the female advantage in social skills, were reflected in adult self-reports. Where childhood sex differences failed to emerge in RAADS-R scores, several interpretations exist; it may be that sex differences are attenuated with age, but an inherent sampling bias may mean that only autistic women most similar to the male presentation are diagnosed and thus included in studies such as ours. The finding that sensorimotor symptomatology is more highly reported by autistic women is a finding requiring objective confirmation, given its potential importance in diagnosis.