Longitudinal Change Profiles of Repetitive Behaviours in ASD Children, Adolescents and Adults over Two Time Points

Background:

Restricted, repetitive behaviours and interests (RRB) are core features of autism spectrum disorder (ASD), some of which cause significant impairment and disrupt daily life (learning/socialisation). RRB are multi-faceted, ranging from self-injurious behaviour to circumscribed interests. Some evidence suggests that RRB symptoms decrease with age. Besides this, little is known about prognostic markers of RRB change, particularly outside toddlerhood. Here we test whether anxiety, sensory processing and/or executive function play a role.

Objectives:

To examine stability/change in RRB, in total and across different subdomains, and investigate factors that may predict change.

Methods:

164 children, adolescents and adults with ASD aged 6-30 years (M=15.55, SD=5.54) from the EU-AIMS Longitudinal European Project were included. RRB were assessed at two time points (T1, T2; time between: M=19.67 months, SD=3.37) using the Repetitive Behaviour Scale-Revised (RBS-R). Both total and subscale scores were examined. Anxiety was measured using the SDQ ‘Emotional symptoms’ score, sensory processing with the Short Sensory Profile. EF was indexed by cognitive flexibility (reversal learning: perseverative errors) and response inhibition (go/no-go: commission errors). Change scores were computed such that higher values indicate improvement. Change groups (‘improve’, ‘no change’, ‘worsen’) were calculated using +/-0.5 interquartile range of T1 median, due to skewness (1.49) – common on the RBS-R. We report effect sizes for Wilcoxon signed-rank test (r) and Spearman’s correlations (r_s). Predictors of change were examined using linear regression.

Results:

At the group level, only the total and Stereotyped Behaviour subscale decreased significantly from T1 to T2 (total: effect size r=-0.15, p=0.006; stereotyped: effect size r=-0.16, p=0.004). Whilst total RRB severity was significantly negatively related to age cross-sectionally (T1:r_s=-0.31, T2:r_s=-0.26, p’s<0.001), degree of change did not vary with age (r_s=-0.08, p=0.29). Change groups: 18.3% of individuals showed improvement in total RRB, 69.5% showed no change and 12.2% worsened (Fig1). RRB subdomains showed similar profiles; 15.2-21.3% improved while 7.9-13.4% worsened. Change groups did not differ on age or IQ, however, there was a higher proportion of females in the ‘worsen’ group and a higher proportion of males in the ‘improve’ group (Fig2; X₂=6.49, p=0.03, φ=0.20). No differences were found between change groups in anxiety or sensory processing at T1. At T2, the ‘worsen’ group had significantly greater anxiety (X²=11.19, p=0.004) and sensory symptoms (X²=7.50, p=0.02) than both other groups. Change in sensory symptoms significantly predicted change in RRB (F(3,103)=5.358, p=0.002, R²=0.24; sensory: p=0.004). Comparatively, change in anxiety significantly predicted RRB total at T2 (F(3,103)=4.285, p=0.007, R²=0.22; anxiety: p=0.001). Groups did not differ on EF measures at T1 or T2, nor did performance changes between time points predict RRB change or T2 RRB total.

Conclusions:

We found RRB stability was considerably more frequent than change, though there was some evidence for overall RRB reduction over time. Males were over-represented in those who ‘improved’, whilst females were over-represented in those who ‘worsened’. Findings support the close links between RRB, anxiety and sensory symptoms, but do not provide evidence for an EF role. Future research should further consider the mechanisms by which RRB symptoms ameliorate or deteriorate.