Investigating Gender-Specific Trajectories of Autistic Traits Across Childhood and Adolescence in a Large Birth Cohort

Background:

Autism spectrum disorder (henceforth ‘autism’) is a dimensional condition, which sits at the extreme of the continuum of autistic traits (ATs) that extends throughout the general population. There is mounting evidence that autism presents differently in males and females (Lai et al., 2015). To date, almost all studies of autism sex/gender differences have been cross-sectional, with a lack of longitudinal research. Therefore, we investigated gender differences in AT trajectories across childhood and adolescence. We tested the opposing predictions of two hypotheses about the development of the female autistic phenotype:

1. The female compensation hypothesis - females with autistic difficulties develop strategies to manage their social difficulties, such that their ATs diminish over time (e.g., Mandy et al., 2012).
2. The adolescent emergence hypothesis - ATs in females become more overt later in development, such that their ATs escalate during adolescence (e.g., Asperger, 1943).

Current diagnostic criteria, which are largely based on male cases, are relatively insensitive to the female autism phenotype. Therefore, we included participants across the full range of AT severity, not just those with an autism diagnosis, to avoid a systematic bias against participants with female-typical autistic difficulties.

Objectives:

To study gender differences in the trajectories of ATs in a large general population sample across childhood and adolescence.

Methods:

Participants (N=9744) were girls (n=4784) and boys (n=4960) from the Avon Longitudinal Study of Parents and Children (ALSPAC), a UK birth cohort study. ATs were assessed aged 7, 10, 13 and 16 years using the Social Communication Disorders Checklist (SCDC), a widely used and psychometrically sound parent-report AT measure. These longitudinal data were modelled using latent growth curve and growth mixture models.

Results:

We observed different AT trajectories for males and females (Figure 1). A latent growth curve model showed that, controlling for IQ, non-autistic psychopathology and socio-economic status: (a) aged 7, males had higher ATs than females (p<.001); (b) males and females both showed a decline in ATs between 7 and 10 years; (c) females, but not males (p<.001), showed a substantial escalation in ATs between 10 and 16 years.

This same pattern of findings occurred when we investigated gender ratios of those with very high ATs. For example, aged 7 years, males had much higher odds than females of scoring ≥ 99^th AT centile (OR = 6.15, 95% CI [5.44, 6.85]), but by 16 years females and males had similar odds of having ATs ≥ 99^th centile (OR=1.18, 95% CI [0.56, 1.80]).

Growth mixture modelling revealed three trajectory groups (Figure 2): Class 1 (26% female) who showed elevated ATs across childhood and adolescence; Class 2 (50% female) who showed persistently low ATs across childhood and adolescence; and Class 3 (52% female) who showed escalation of ATs during adolescence.

Conclusions:

We present the first empirical evidence for the adolescent emergence hypothesis: a sub-group of girls showed substantial escalation of ATs between 10 and 16 years. This may partly explain why females tend to be diagnosed with autism later than males.