Sex Differences in Autistic Traits: Is High Verbal IQ Protective against Social Impairments in Girls but Not Boys?

Background:

The increase in male to female ratio in autism spectrum conditions (ASCs) with increasing IQ may be an important clue to the aetiology of these disorders. Recent research (Skuse et al., 2009) suggests verbal IQ (VIQ) at the high extreme is protective against social communication impairments in girls only.

Objectives: The aim is to extend this finding to both verbal (VIQ) and nonverbal (NIQ) effects on the whole triad of autistic-like traits and to more clearly analyse the shape of the relationship between autistic traits and IQ for boys and girls separately.
Methods: IQ data at age 7 and parental ratings of autistic traits at age 8 came from 8250 twin pairs from the Twins Early Development Study (TEDS). The analyses are phenotypic and treat twins as individuals accounting for family membership and zygosity. Total autistic-trait scores (as measured by the Childhood Autism Spectrum Test, CAST - Scott, Baron-Cohen, Bolton & Bayne, 2002) with subscales for social, communication and repetitive, restricted behaviors and interests (RRBIs) were related to VIQ and NIQ using categorical (autistic traits in distinct IQ groups along the entire range of abilities) and continuous (structural equation modelling) approaches.
Results: Boys had higher mean autistic-like traits in all IQ groups. A continuous analysis revealed both linear and curvilinear effects of IQ on autistic-like traits with smallest effects of IQ on RRBIs. In VIQ sex differences occurred only in curvilinear effects on communication impairments in boys but not girls. This indicated that girls’ communication impairments decreased with increasing VIQ in a linear pattern whereas boys showed an upturn in communication impairments at the high end of VIQ (i.e. a significant curvilinear shape for boys only).  Effects of NIQ were smaller throughout with significantly curvier slopes in girls for the total CAST score which is the opposite pattern to the expected protective effects. Linear effects of NIQ originated entirely from the communication subscale. NIQ did not play a significant mediating role on RRBIs.
Conclusions: This study replicates recent findings that high VIQ is a protective factor for communication difficulties in girls only. The novel structural equation modeling approach used in this study allows more precise predictions about the number of autistic traits given a specific IQ. A sex difference in the slope between VIQ and communication traits indicates that with increasing VIQ girls show fewer autistic traits whereas boys have proportionally more autistic traits at the high end of VIQ. RRBIs appear to be least affected by level of IQ. When interpreting sex ratios in autistic traits at the high end of IQ abilities, our findings indicate that it is important to take into consideration the particular type of high IQ as well as which part of the triad is measured. The central role of communication traits, which are differently mediated by IQ according to sex, and the interpretation of possible protective effects of high VIQ are discussed.