Which Motor Stereotypy Subtypes Are Most Indicative of an Autism Spectrum Disorder in Early Childhood?

Poster Presentation
Friday, May 11, 2018: 5:30 PM-7:00 PM
Hall Grote Zaal (de Doelen ICC Rotterdam)
A. K. McCullough1,2, S. Lopez-Pintado3 and S. Goldman4, (1)Neurology, Columbia University Medical Center, New York, NY, (2)Teachers College, Columbia University, New York, NY, (3)Biostatistics, Columbia Universty - Mailman school of Public Health, New York, NY, (4)Department of Neurology, Division of Child Neurology, Columbia University Irving Medical Center, New York, NY

Motor stereotypies, defined as repetitive, patterned, purposeless behaviors, are one of the hallmark behaviors in children with a diagnosis of autism spectrum disorders (ASD). However, these hyperkinetic movements are also reported in selected populations of children with non-autistic neurodevelopmental disorders, and during infancy in typically developing children. To date, the majority of studies focusing on stereotypies have used parent-reported questionnaires, which offer insufficient details for targeted discriminative analyses of restrictive and repetitive motor behaviors.


Therefore, the purpose of this study was to use a reliable motor stereotypy video coding system (Goldman, 2008) to identify and score each motor stereotypy in children with ASD and children with non-ASD developmental disorders (NASDD) in order to 1) examine which stereotypy subtypes reveal the most salient characteristics of restrictive and repetitive motor behavior in early childhood, and 2) determine which stereotypy subtypes best distinguish children with ASD from NASDD children.


Motor stereotypies were coded individually from the first 15 minutes of 30-minute standardized video-recorded play sessions. A cohort of N=277 preschool children with a neurodevelopmental disorder [F: 24.5%(68); age: 4.6(1.2) years], with and without a comorbid diagnosis of ASD [ASD: 47%(129)], was recruited through a multi-site study focusing on neuropsychological and psychiatric assessments. For each video, the following topographical descriptors were used to generate eight mutually exclusive stereotypy subtypes: face, head/trunk, arm/leg, hand/finger, hand/finger with object, gait, self-directed, and visual.

Data were analyzed in MATLAB R2016b, and descriptive statistics were calculated as Mean (SD), and Frequencies [%(n)]. Among children with stereotypies, frequencies of stereotypy subtypes were entered into a principal components analysis (PCA) with a varimax rotation. PCA stereotypy subtypes with loadings ≥0.40 were used to interpret components. Then, using the entire cohort, associations between an ASD diagnosis and the presence of select motor stereotypy subtypes, as determined by the PCA, were tested using logistic regression (LR) and the odds ratios (OR) were calculated. Finally, the Receiver Operating Characteristic Area Under the Curve (AUC) was evaluated. All models were run with an a priori significance level of p<0.05.


All children [n=122, ASD: 72%(88)] with stereotypy frequencies >1 were included in the PCA. Results showed that three components explained 76.6% of the variance, and could best be described by the following motor stereotypy subtypes: (1) self-directed, (2) limb (arm, leg, foot, and hand stereotypies combined), and (3) object-oriented. LR showed that, in the full cohort, the presence of limb (OR: 6.9) and object-oriented (OR: 3.01) stereotypies were significantly associated with an ASD diagnosis, while self-directed stereotypies were not (OR: 1.4). The LR model AUC was 0.75, with a sensitivity and specificity of 86% and 58%, respectively.


A subset of motor stereotypies (self-directed, limb, and object-oriented) appear to be the most salient motor stereotypy subtypes at preschool in both children with and without ASD. Only the limb and object-oriented subtypes, however, appeared to sufficiently identify children with an ASD diagnosis. These results are particularly relevant to clinical assessments of autism in young children, as they may help to strengthen a diagnosis.