Between a ROC and a hard place: decision making and making decisions about using the SCQ.
Academic Article
Overview
abstract
BACKGROUND: The Social Communication Questionnaire (SCQ), formerly the Autism Screening Questionnaire (ASQ), is based on a well-validated parent interview, the Autism Diagnostic Interview (ADI). It has shown promise as a screening measure for autism spectrum disorders (ASDs) in a research-referred older sample, though recent studies with younger children reported lower sensitivities when using the suggested cutoff of > or = 15 to differentiate ASDs from children with nonspectrum disorders (NS). METHODS: Diagnostic discrimination of the SCQ was evaluated alone and in combination with the ADOS (Autism Diagnostic Observation Schedule) in a clinical and research-referred sample of 590 children and adolescents (2 to 16 years), with best estimate consensus diagnoses of autism, pervasive developmental disorder, not otherwise specified (PDD-NOS) and non-ASD disorders. The SCQ was completed before the evaluation in most cases. Performance of the SCQ was also compared with the Autism Diagnostic Interview - Revised (ADI-R). RESULTS: Absolute scores and sensitivity in the younger children and specificity for all groups were lower than reported in the original study. Using receiver operating curves (ROC) to examine the area under the curve (AUC), the SCQ was more similar to the ADI-R total score in differentiating ASD from NS disorders in the older (8-10, >11) than younger age groups (<5, 5-7). Lowering the cutoff score in the 2 younger groups improved sensitivity, with specificity remaining relatively low in all groups. Using the SCQ in combination with the ADOS resulted in improved specificity. Diagnostic discrimination was best using the ADI-R and ADOS in combination. CONCLUSIONS: Those interested in using the SCQ should consider adjusting cutoff scores according to age and purpose, and using it in combination with another measure. Sensitivity or specificity may be prioritized for research or screening depending on goals.
