Evaluation of a 5-Item Subset of the Social Responsiveness Scale for Distinguishing Between Children With and Without Autism Spectrum Disorder.

purposeIdentifying children with autism spectrum disorder (ASD) of all ages using a screening tool is valuable. We evaluated a shortened version of the 65-item Social Responsiveness Scale (SRS).

methodsA total of 1,462 dyads from the Simons Simplex Collection consisted of a child with a clinically confirmed diagnosis of ASD and a sibling without ASD. All siblings were free of suspected intellectual disabilities and severe learning disabilities, had neither developmental nor psychiatric disorders, and did not have an Individual Education Plan for extensive special education services. Areas under Receiver Operating Characteristics (AUROC) curves quantified scale performance. A non-inferiority margin of - 0.005 was pre-specified.

resultsThe AUROC for the 5-item scale (5iSRS) was 0.9943, the predictive value positive was 97.3% and predictive value negative was 96.9%. The AUROC for the 65-item SRS was 0.9926, resulting in an AUROC difference of 0.0017 (95% CI=[- 0.0003, 0.0037]) and meeting the non-inferiority criterion.

conclusionThe 5iSRS can distinguish between children with and without ASD and is non-inferior to the 65-item SRS, and thus may have substantial utility in research, community, and clinical settings. Given that this initial evaluation of the shortened version was performed among children in simplex families with selected unaffected siblings, further evaluations in community and clinical settings among non-related children with varying levels of autistic traits and/or other behavioral challenges are needed prior to considering the shortened version as a viable ASD screening tool.