Studying Restricted and Repetitive Behaviors : An Individual Items Versus a Clustered Approach of Analysis

Background: Restricted and repetitive behaviors (RRBs) are one of the two core domains required for an autism diagnosis (APA, 2013). Different classifications of RRBs have been used in young autistic children. Some observational studies have used a small number of clusters (Turner, 1999; Barber et al. 2012; MacDonald et al. 2007; Stonach & Wetherby; 2012), while others have used a more detailed approach (Goldman et al. 2008; Loh et al. 2007; Mottron, et al. 2007; Ozonoff et al. 2007). Using many items gives a more precise picture of RRBs, but could mean a loss of statistical power when multiple tests are conducted. Therefore, comparing different methods of clustering RRBs will help understand how these methods influence the results.

Objectives: This study aims to compare the RRBs profiles of autism spectrum disorder (ASD) and typically developing (TD) preschooler groups using the Montreal Stimulation Play Situation (MSPS) items analysed individually or in 3 clusters inspired by the ADOS-2 classification.

Methods: 53 autistic children (43.4 months ± 15.2; MSEL standard score 65.5 ± 20.0) and 46 TD children (48.1 months ± 11.5, p<0.05; MSEL standard score 110.8 ± 17.3, p<0.001) were exposed to the MSPS, which includes free and semi-free play sessions with objects of interest to ASD children. The frequency and duration of the 49 RRBs were coded twice by coders blind to the diagnosis using Noldus Observer. The mean duration and frequency of the 49 items were compared between the ASD and TD groups using non-parametric analysis. The 49 RRBs items of the MSPS scale were then pooled according to the clusters of the ADOS-2 algorithm (except those related to vocalization): 1) unusual sensory interests in play material/person; 2) hand, finger and other complex mannerisms; 3) unusually repetitive interests/stereotyped behaviors.  Differences between the groups in the mean duration and frequency of these three clusters were determined using a Student’s t-test.

Results: When analysed separately, three of the 49 items were significantly higher on both mean duration and frequency in the ASD group (p<0.01):  hand flapping, arm movements, close look to an object. Hand flapping and arm movements are included in the mannerisms cluster, while close look to an object is included in the unusual sensory interests cluster. When pooled using 3 clusters from the ADOS-2 classification, only mannerisms was significantly higher on both frequency and duration for the ASD group.  

Conclusions: When the RRBs were analysed separately, differences between the ASD and TD groups were found in one of the unusual sensory interests. However, this information was lost in the pooled analysis, where only the mannerisms cluster showed significant differences. This suggests that by using a small number of RRB clusters, the differences between TD and ASD children may be blurred by RRBs which are less specific to ASD children. Future studies should aim to determine the optimal number and the contents of clusters.