Longitudinal Changes in the Presence and Severity of Self-Injurious Behaviors, and Predictors of Change, in Children with Autism Spectrum Disorders

Background: Self-injurious behaviors (SIB) are common in children with autism spectrum disorders (ASD). Understanding the natural history of SIB in ASD can inform studies examining the efficacy of interventions. Furthermore, identification of characteristics that predict changes in SIB over time may help target interventions.  

Objectives: To assess parent-reported changes in the presence and severity of SIB in children with ASD over one year and explore baseline characteristics that predict changes.  

Methods: Parent-reported current presence (yes/no) and current severity (mild, moderate, severe) of SIB at both baseline and follow-up visit were examined among children with ASD aged 2-17 years enrolled in the Autism Treatment Network registry.  Participants were categorized into: SIB present at both visits, SIB absent at both visits, SIB present at baseline but not follow-up (improvement), and SIB absent at baseline but present at follow-up (new occurrence). Subjects who improved were compared to those with SIB at both visits.  Subjects with new occurrence were compared to those without SIB at both visits.  The four severity groups were: mild at both visits, moderate-severe at both visits, moderate-severe at baseline and mild at follow-up (improvement), and mild at baseline and moderate-severe at follow-up (deterioration). Subjects who improved were compared to those who remained moderate-severe, and subjects who deteriorated were compared to those who remained mild. McNemar’s test was used to evaluate all differences in proportions and logistic regression was used to assess effects of baseline demographic, developmental, behavioral, and somatic characteristics on all changes.  

Results: Among 1575 participants with data on presence of SIB at both visits, SIB was present in 34.2% at baseline and 28.0% at follow-up. SIB improvement was more common at follow-up (14.7%) than was new occurrence (8.5%, p<0.0001). Improvement was associated with higher baseline adaptive behaviors scores (adjusted OR=1.02; 95% CI: 1.00, 1.04).  Improvement was less likely in children with developmental regression (aOR=0.64; 0.43, 0.95) and low maternal education (aOR=0.66; 0.44, 0.98) at baseline. New SIB occurrence was associated with aggression (aOR=2.09; 1.43, 3.06) and low maternal education (aOR= 1.81; 1.18, 2.78).  New occurrence was less likely with higher baseline adaptive behaviors scores (aOR=0.97; 0.95, 0.99). Only 277 participants had severity data at both visits due to changes in data collection, but had similar characteristics to those without follow-up data.  Among these 277 children, 54.3% had mild SIB and 45.7% had moderate-severe SIB at baseline, while 60.1% had mild SIB and 39.9% had moderate-severe SIB at follow-up. More parents reported change from moderate-severe to mild (19.5%), compared to the reverse (11.5%; p=0.02). Higher IQ at baseline predicted improvement (aOR=1.02; 1.00, 1.04), while having public insurance (aOR=3.02; 1.16, 7.90) and low maternal education (aOR=3.46; 1.28, 9.40) at baseline predicted deterioration. 

Conclusions: At baseline, parents reported SIB in 34.2% of children. More parents reported resolution than development of new SIB. Further, SIB severity was more likely to improve over time than to worsen. Maternal education, public insurance, IQ scores, developmental regression, and adaptive behaviors scores at baseline predicted changes in SIB and may be useful for targeting SIB prevention and interventions.