Cross-replicating findings on unique motor impairments of children with ASD using confirmatory factor analysis and a novel SPARK study sample.

Abstract

A series of recent reports have shed light on the pervasive nature of motor impairments in children with ASD (Bhat, 2020, 2021, Bhat et al., 2022), underscoring the importance of providing ASD clinicians efficient and accurate tools for motor screening. The Developmental Coordination Disorder-Questionnaire (DCD-Q) is a widely used motor screening tool, yet scant evidence exists regarding its psychometric properties in children with ASD. In a recent Exploratory Factor Analysis (EFA) of the 15-item DCD-Q in a large sample of children with ASD (SPARK study), we found a 5-factor latent structure that identified unique motor impairments in a large sample of children with ASD (Bhat et al., 2022). In the current study, we extend this work by cross-replicating the EFA findings of unique ASD-related motor issues using Confirmatory Factor Analysis (CFA) in a new, more recent wave of children with ASD from the SPARK study (N=9721). The fits and interpretability of 11 hypothesis-driven CFA models, including 8 correlated-factors, 1second-order, and 2 bifactor models were compared. Our findings supported the previous 5-factor model with 2 gross motor subdomains, 1 fine motor domain (similar to the original DCD-Q) and 2 general coordination subdomains. This model demonstrated acceptable fit in the new sample as well as superior fit compared to several other parsimonious correlated-factors models. However, the second-order and bifactor models fit slightly better and supported the presence of a general motor skills factor, although 38% of the common variance in the DCD-Q items remained attributable to the 5 subdomains. Using one of the bifactor models, measurement invariance was also supported for DCD-Q across sex, race, and co-occurring conditions of language disorder and intellectual disability. Only partial invariance was supported when testing DCD-Q scores across different age groups. These findings reveal a more complex dimensional picture of the DCD-Q in children with ASD. Results suggest that the DCD-Q can be used in two ways, total scores adequately assess general motor skills for brief screening and subdomain scores offer unique information on the multidimensional motor problems of children with ASD. If subdomain data are of interest, our findings call into question the practice of relying on 3 original subscales of the DCD-Q when screening for ASD-related motor difficulties, whereas 4 out of 5 subscale scores may better highlight domain-specific motor problems. Future studies should continue to further validate DCD-Q's ability to screen for multidimensional motor problems.