Abstract
BACKGROUND: Autistic children experience significantly higher rates of anxiety compared to nonautistic children. The precise relations between autism characteristics and anxiety symptoms remain unclear in this population. Previous work has explored associations at the domain level, which involve examining broad categories or clusters of symptoms, rather than the relationships between specific symptoms and/or individual characteristics. We addressed this gap by taking a network approach to understand the shared structure of autism characteristics and anxiety symptoms. METHOD: Data were pooled from five studies from Canada, Singapore, the UK, and the USA, totaling 623 autistic children (17% female sex; aged 6-18 years), for whom the parent-report Spence Children's Anxiety Scale (SCAS-P) was available. We derived two undirected regularized networks, first from the SCAS-P items only, and then by adding autism characteristics pertaining to social communication, highly focused and repetitive behavior, and sensory hypersensitivity. From these models' metrics, we extracted nodes' predictability, key bridging nodes, and community detection. RESULTS: The anxiety-only network was highly connected and consisted of four key clusters: General Anxiety, Social Anxiety, Separation Anxiety, and Panic/Agoraphobia. These broadly aligned with the existing SCAS-P structure based on DSM-IV-TR criteria. In the autism-anxiety network, the structure of anxiety remained mostly stable, with autism features forming their own community. Preference for predictability (i.e., sameness) and sensory hypersensitivity were key nodes that linked autistic features and anxiety symptoms, primarily through generalized anxiety. CONCLUSION: This study identified some of the key characteristics that bridge the broadly independent structures of autism characteristics and anxiety symptoms. The findings are discussed in the context of guiding the assessment, prevention, and treatment of anxiety in autism.