The Optimal Cut-Off Point for Thai Diagnostic Autism Scale and Probability Prediction of Autism Spectrum Disorder Diagnosis in Suspected Children

Abstract

The Thai Diagnostic Autism Scale (TDAS) was developed to diagnose autism spectrum disorder (ASD) under the context and characteristics of the Thai population. Although the tool has an excellent agreement, the interpretation of diagnostic results needs to rely on the optimal cut-off point to maximize efficiency and clarity. This study aims to find an optimal cut-off point for TDAS in the diagnosis of ASD and to compare its agreement with the DSM-5 ASD criteria. This study was conducted on 156 children aged 12–48 months old who were suspected of having ASD and had enrolled from hospitals in the four regions of Thailand in 2017–2018. The optimal cut-off point for TDAS was considered by using receiver operating characteristic (ROC) curves according to the DSM-5 ASD criteria. The areas under the curve (AUCs) for TDAS and ADOS-2 were also compared. Multivariable logistic regression was performed to create a predictive model for the probability of ASD. The AUC of TDAS was significantly higher than that of ADOS-2 (0.8748 vs. 0.7993; p = 0.033). The optimal cut-off point for TDAS was ≥20 points (accuracy = 82.05%, sensitivity = 82.86%, and specificity = 80.93%). Our findings show that TDAS with a cut-off point can yield higher diagnostic accuracy than ADOS-2 and TDAS domain. Diagnosis by using this cut-off point could be useful in practical assessments.