Associations Between Aggression and Restricted, Repetitive, and Stereotyped Behaviors and Interests in Children with Autism Spectrum Disorder: A Multi-Informant, Multi-Method Study

Background:  Aggression in children with autism spectrum disorder (ASD) is highly prevalent and seriously impairing to the child, family, and community. Despite this burden, correlates of aggression have not been identified in this population. Although preliminary data indicate a positive association between restricted, repetitive, and stereotyped behaviors and interests (RRSBI) and aggression in individuals with ASD (Kanne and Mazurek, 2011), these studies have been hampered by methodological limitations.  Therefore, detailed analysis of this relationship is critical prior to the development of targeted prevention and intervention strategies.

Objectives:  To examine: a) if RRSBI is positively associated with aggression across multiple assessment methods, informants, and definitions of aggression; b) what methodological factors have a greater influence on this relationship (i.e., informant, such as teacher- or parent-report, vs. method, such as semi-structured interview or questionnaire); and c) if the definition of aggression influences this association (i.e., either narrowly defined as physical aggression, such as hitting and kicking, or aggression broadly defined across multiple behaviors, including verbal aggression and affective instability).

Methods:  The sample consisted of 2648 children from the Simons Simplex Collection, a 12-site North American, university-based research study that gathers phenotypic data from families with a single child with ASD (Mean Age = 9y, SD=3.7y; 87% male; 74% Caucasian). Diagnosis was confirmed using the ADI-R (Rutter, Le Couteur, & Lord, 2003) and the ADOS (Lord, DiValorne, & Risi, 2002). Aggression was measured using two items from the ADI-R (#81, #82) and the CBCL - Aggressive Behavior Syndrome Scale (parent and teacher versions) (Achenbach and Rescorla, 2001). RRSBI was measured utilizing the Restricted, Repetitive, and Stereotyped Patterns of Behavior subscale of the ADI-R, the total score from the RBS-R (parent report) (Lam and Aman, 2007), and the Autistic Mannerisms subscale from the SRS (parent and teacher report) (Constantino et al., 2003). Multiple linear regression models were employed to calculate mean differences in RRSBI across levels of aggression while adjusting for IQ, race, household income, child gender, and maternal education. Estimates from those models were used to calculate effect sizes (Cohen’s d) which were averaged, adjusted (for informant vs. method confounding/overlap), and compared across informant, method, and type of aggression.

Results:  A significant, positive association was found for 31 of the 40 RRSBI x Aggression analyses (p < .05). Although the overall average effect size was small (Δ = .36), there was tremendous heterogeneity in these estimates (Δ ranged from 0 to 1.19). Differences in effect sizes for informant (Δ =.51) was twice as great as method (Δ = .25). A similar finding was evident when aggression was broadly (Δ = .56) vs. narrowly defined (Δ = .31).

Conclusions:  As theory suggests RRSBI to be an antecedent of aggression (Reese et al., 2003), results from this analysis suggest that interventions targeting the reduction of RRSBI may result in a concomitant reduction in aggression. Methodologically, these findings indicate differences in informant, method, and definition of aggression introduce unexplained variance into this association and should be addressed in prospective intervention studies.