2016 Volume 57 Issue 4 Pages 603-617
The short-form of the Autism Spectrum Screening Questionnaire (ASSQ) was administered to 1,919 5-year-old children (1,002 boys/917 girls) and 4,374 school-aged children (ages 7-14, 2,221 boys/2,153 girls) from the same city to verify applicability of the ASSQ short-form in 5-year-old children. Fifty-nine of the 5-year-olds were diagnosed with autism spectrum disorder (ASD, 37 boys/22 girls). The results of confirmatory factor analysis confirmed factor structure to be the same for 5-year-olds as it was for school-aged children; however, multiple-group analysis demonstrated that estimates of factor loadings differed among groups. Furthermore, two-way (age (5-year-olds: school-age) ×gender (male: female)) analysis of variance found main effects of age and gender, but no significant effects of interaction. Meanwhile, multiple linear regression analysis with ASD diagnosis as the dependent variable and the three factors of the ASSQ short-form as independent variables demonstrated that each factor could significantly predict a diagnosis. Further, Receiver Operating Characteristic analysis by gender demonstrated favorable accuracy of identification (AUC=boys .92, girls .91), demonstrating the possibility that the ASSQ short-form can identify ASD in 5-year-old children with a certain degree of accuracy despite differences in scores and modes from school-age children. The findings suggested validity of cut-off scores of 3 for boys (sensitivity .94, specificity .81) and 4 for girls (sensitivity .73, specificity. 94) for screening purposes, and a cut-off score of 8 for both boys (sensitivity .49, specificity .99) and girls (sensitivity .54, specificity .98) for identifying cases strongly suspected of ASD. Positive precision value at the latter cutoff score was 61% for boys and 53% for girls, suggesting functionality of the cut-off score indicating a relatively strong suspicion of ASD.
