Criterion validity of the Autism Spectrum Rating Scales 2-5 parent report

Background: Questionnaires are frequently utilized in autism spectrum disorder (ASD) assessment and research, but there is limited evidence regarding their psychometric properties. Only two studies have investigated the diagnostic utility of the Autism Spectrum Rating Scale-2-5 parent report (ASRSp 2-5), and in both of these studies, the use of a gold-standard ASD measure in the diagnostic process is unclear. Method: The current study investigated the criterion validity of the ASRSp-2-5 in a clinical sample of 238 children (X¯ _age = 3.66; autism[AUT] n = 84; non-autism[NOT] n = 154). Children were diagnosed via a comprehensive assessment that included the Autism Diagnostic Observation Schedule-2. Results: Analyses included t-tests/ANCOVAs, Logistic Regression (with and without covariates), and ROC analyses (with t-scores and predicted probabilities to account for covariates). The ASRSp 2-5 scores intended to be used for diagnosis (Total and DSM-5) were non-significant in all analyses. Higher scores in the NOT group were found for Unusual Behavior and Behavioral Rigidity. However, mean differences were non-significant with inclusion of control variables (IQ and age). Social, Peer Interaction, and Social-Emotional Reciprocity demonstrated the strongest findings. The ASD group had significantly higher scores on these scales, and findings held with covariates included (IQ for Social; age for Peer Interaction). However, odds ratios were small whether or not covariates were included, and AUCs were poor-fair (.69–.70). Sensitivity and specificity could not both be optimized at any cutpoint. Conclusions: The findings were not as strong or as consistent as expected. Results suggest the ASRSp 2-5 is most useful in ruling out ASD in cases of low scores (<60). When the ASRSp 2-5 is utilized, behavioral, social, and emotional problems associated with other disorders must be considered in interpreting scores. Additional research on this measure is necessary.