Examining the Factor Structure of the Spence Children's Anxiety Scale – Parent Version in Children with Autism Spectrum Disorder

Background: Anxiety disorders are estimated to occur in approximately 40% of youth with ASD, and represent one of ASD’s most highly comorbid psychopathologies (van Steensel et al. 2011). The assessment of anxiety in ASD is problematic because ASD-specific characteristics, such as social skill impairments and alexithymia, may alter the experience of anxiety in this population (Kerns & Kendall, 2012). Consequently, the appropriateness of anxiety assessment tools originally developed for typically developing children cannot be assumed in ASD. Research examining the psychometric properties of existing anxiety measures in ASD samples is emerging. Initial evidence for the reliability of the Spence Children’s Anxiety Scale – Parent Report (SCAS-P; Spence, 1999) - a widely used anxiety rating scale - has been demonstrated in two samples of children with ASD (Magiati et al., 2015; Zainal et al., 2014). Support for a six correlated factor structure in the SCAS-P (separation anxiety, social phobia, generalized anxiety, panic attacks and agoraphobia, obsessive-compulsive disorder, physical injury fears) has been found in community and clinical samples (Arendt et al., 2014; Nauta et al., 2004). However, the SCAS-P’s factor structure when used with children with ASD has not yet been examined. 

Objectives: The purpose of this study was to examine the factor structure of the Spence Children’s Anxiety Scale – Parent Report (SCAS-P) in children with ASD. 

Methods: Data were drawn from the Canadian Pathways in ASD study (N=421) and included children whose parents who had completed all 38 items of the SCAS-P (n=167; 83.1% boys; mean age 7.74 years old [SD=.22]). All children had a diagnosis of ASD based on DSM-IV criteria and confirmed with ADI-R/ADOS. Confirmatory factor analysis (CFA) using robust weighted least squares estimation was used to examine the fit of the previously supported six-correlated factor model, and four other theoretically guided models (Arendt et al., 2014; Nauta et al., 2004). If the CFA did not result in model fit, principal axis factoring (PAF) using direct oblimin rotation was used to explore an alternative plausible structure. Parallel analysis guided factor selection.

Results: CFA suggested the model only partially met criteria (CFI=.888, TLI=.878, RMSEA=.040, [CI 90% .031-.048; Test of close fit p=.983], SRMR =.117). PAF suggested a six-factor solution best described the data, explaining a 48.3% of the variance. Four factors emerged: physical symptoms of autonomic arousal; panic, agoraphobia, and compulsive behaviors; social phobia; and separation anxiety. Two factors had mixed item groupings: the first was comprised of items relating to separation anxiety and physical symptoms; and the second contained items related to obsessions, compulsions, and generalized worries. Four items (13, 21, 23, 30) had small loadings <|.3|  on all factors. 

Conclusions: CFA on five SCAS-P factor structures previously suggested in community and clinical samples did not show a good fit in a small sample of children with ASD. As a follow up, exploratory analyses resulted in an alternative six-factor solution, with four items not showing strong loadings onto any factors. The findings are discussed with regards to the nature and assessment of anxiety in ASD.