Longitudinal Course and Predictors of Aggression in Children with ASD

Background:  Although the nature, course, and predictors of aggression have been well researched in typically developing samples, there have been few large-scale studies of this topic in children with autism spectrum disorder (ASD).  This is surprising given that aggression can have significant negative effects for individuals and families.  Recent research on the prevalence and cross-sectional correlates of aggression in children with ASD suggests that several co-occurring problems are associated with aggression in this population, including sleep problems, gastrointestinal problems, sensory problems, and self-injurious behavior.  However, longitudinal studies examining the course of aggression and the variables that may predict change in aggression over time in children with ASD have not been conducted.

Objectives:  1) To examine changes in aggression over time among children with ASD, and 2) to investigate whether specific demographic, functional, or symptomatic factors predict changes in aggression over time among children with ASD. 

Methods:  Participants included children and adolescents (ages 2-18) with ASD enrolled in the Autism Speaks Autism Treatment Network (AS-ATN).  The AS-ATN maintains a longitudinal clinical registry database, and data for the current study were collected at yearly intervals (enrollment, and first, second, and third annual follow-up visits).  Data were available for 5,216 children at enrollment, with substantially fewer participants with complete data across all time points.  Aggression was measured by the Aggressive Behavior Syndrome Scale from the Child Behavior Checklist (CBCL).  Additional measures included demographic and parental concerns questionnaires, Autism Diagnostic Observation Schedule (ADOS), IQ, Children’s Sleep Habits Questionnaire, Short Sensory Profile, and Vineland Adaptive Behavior Scales – Second Edition.  

Results:  Data were initially analyzed using an unconditional means model and unconditional growth model.  These models suggested that: 1) children had different baseline aggression scores and rates of change, and 2) the average child in the sample had less aggression over time.  Theoretically relevant predictor variables were then entered into hierarchical linear models (HLMs) in order to determine which variables accounted for variability in initial aggression scores and in changes in aggression over time.  HLMs showed that higher initial aggression scores were associated with greater sleep problems, sensory problems, and self-injurious behavior, while GI problems were not associated with initial aggression scores.  Results also indicated that specific sets of co-occurring symptoms predicted different rates of improvement in aggression over time.

Conclusions:  Children with ASD show marked variability in initial levels of aggression and in how these behaviors change across time.  Consistent with previous findings, the results indicate that co-occurring problems, including sleep problems, sensory problems and self-injury, are highly associated with aggression in children with ASD.  Further, the current results indicate that these co-occurring problems may affect long-term improvement in aggression.  This was the first large scale study to investigate longitudinal patterns of change in aggression in a large, well-characterized sample of children with ASD.  The results provide important information about patterns of aggression and potential targets for intervention.   Future longitudinal work should include more comprehensive measures of aggression, and should examine the extent to which treatment for aggression and co-occurring problems have reciprocal effects.