Accuracy of Three Screening Instruments in Identifying Preschool Children at Risk for Autism Spectrum Disorder
Pages 156-163
Khaled Alkherainej and Jane Squires

DOI: http://dx.doi.org/10.6000/2292-2598.2015.03.04.1

Published: 07 January 2016

Abstract: An efficient approach for screening and identifying children at risk for autism spectrum disorder (ASD) remains a pressing need. The aim of this exploratory study was to examine the ability of two general developmental screening tests to identify children at risk for ASD. We compared the accuracy of one general developmental screening instrument, Ages and Stages Questionnaires (ASQ), and one general social emotional screening instrument, the Ages and Stages Questionnaire: Social Emotional (ASQ:SE), with the Social Communication Questionnaire (SCQ), an ASD-specific screening instrument.

Two hundred eight children between 36 and 66 months were recruited through 19 community ASD programs, websites, and magazines. The three screening instruments were given to 285 parent/child dyads with and without a diagnosis of ASD, online via a screening website linked to a university. Sixty-four children had been diagnosed with ASD and were receiving special education services (e.g., behavioral interventions) prior to their participation. The classification agreement of the ASQ (i.e., sensitivity = 84.38%, specificity = 81.45%) outperformed the other two screening instruments; classification agreement of the SCQ was sensitivity = 70.31% and specificity = 87.33%; and of the ASQ: SE, sensitivity = 82.81% and specificity = 72.40%. Agreement among the questionnaires ranged from moderate to strong as measured by Pearson Product Moment Correlation Coefficients. Children diagnosed with ASD had scores below the screening cutoff points, indicating risk, most often on three ASQ domains: (a) communication, (b) gross motor, and (c) personal social. This exploratory study indicated the feasibility of using the ASQ in screening clinics for finding children at risk for ASD, if the ASQ is followed by specific ASD assessments. Design limitations, including a sample of children with ASD already receiving intervention services may explain the somewhat lower sensitivity of the SCQ.