Validating the Screening Tool for Autism in Toddlers and Young Children (STAT) in children aged 24–36 months in a low- to middle-income setting
TL;DR
Preliminary evidence is provided supporting the existing STAT cut-off of ≥ 2 in similar high-risk clinical populations, while larger multicenter and community-based studies are needed to confirm the performance and generalizability of this cut-off.
Abstract
Autism spectrum disorder (ASD) prevalence is increasing globally, including in low- and middle-income settings. Early screening is essential; however, screening tools require local validation. The Screening Tool for Autism in Toddlers and Young Children (STAT) has been incorporated into Vietnam’s Ministry of Health guidelines, yet its diagnostic performance has not been formally evaluated. This cross-sectional study validated the STAT among 86 children aged 24–36 months with language delay attending a tertiary pediatric hospital in Ho Chi Minh City between December 2022 and August 2023. STAT results were compared with DSM-5-based clinical diagnoses, the diagnostic reference standard. The STAT was administered by certified speech therapists, and STAT results were not disclosed to the child psychologist conducting the DSM-5-based diagnostic evaluation. Diagnostic accuracy indices and receiver operating characteristic (ROC) analysis were performed to evaluate diagnostic performance and explore alternative cut-off values. The findings showed a predominance of male children (65.1%) and a high prevalence of ASD (74.4%). At the recommended cut-off of ≥ 2, STAT demonstrated high sensitivity (89.1%) but moderate specificity (50.0%), with positive and negative predictive values of 83.8% and 61.1%, respectively. The area under the ROC curve was 0.7656 (95% CI: 0.639–0.883), indicating fair overall discriminative ability. In an exploratory analysis, a data-derived cut-off of 2.5 yielded the highest Youden Index, with sensitivity of 73.4% and specificity of 68.2%. The exploratory 2.5 threshold requires independent validation before any change to the recommended ≥ 2 threshold is considered. Overall, these findings provide preliminary evidence supporting the existing STAT cut-off of ≥ 2 in similar high-risk clinical populations, while larger multicenter and community-based studies are needed to confirm the performance and generalizability of this cut-off. First validation of STAT in Vietnamese toddlers with language delay. Recommended cut-off ≥ 2 showed sensitivity of 89% and specificity of 50%. AUC 0.7656 indicates fair overall discriminative ability. Exploratory cut-off 2.5 showed sensitivity of 73.4% and specificity of 68.2%.