30463
Utility of the Screening Tool for Autism in Two-Year-Olds (STAT) to Detect Autism in Toddlers before 24 Months of Age in Taiwan: A Longitudinal Study
Autism spectrum disorder (ASD) is thought as an innate developmental disorder. Over past decade, some studies reported that prevalence of ASD has increased markedly. However, compared to Western countries, the prevalence of ASD in Taiwan is lower. There are a few reasons that cause the differences of the prevalence, such as stigma in Chinese culture, policy of government etc. Previous studies reported that early intervention improved outcomes for toddlers with ASD. The benefits of early intervention highlighted the importance of earlier identification of toddlers with ASD. Knowing the importance of early identification, the American Academy of Pediatrics (AAP) has recommended that all infants receive universal screening for autism less than 24 months old (Johnson et al., 2007).
Objectives:
The purpose of the longitudinal study is to examine the predictive validity of the Screening Tool for Autism in Two-Year-Olds (STAT; Stone et al., 2008) as a screening tool for detecting ASD in toddlers before 24 months of age in hospital-based clinical sample. The STAT is a 12 items interactive autism-specific screening tool, including four behavioral domains: play (2 items), requesting (2 items), joint attention (4 items) and imitation (4 items).
Methods:
There were 139 toddlers between 16 to 24 months (Time 1) participated the study and all toddlers were recruited from the teaching hospital in Chia-Yi area. 18 months after the assessment at Time 1, participants were invited for re-assessment and finally diagnosis. There were 119 children between 35 to 46 months (Time 2) received follow-up assessment. The average length of time between the initial and follow-up assessment s was 18.64 months (SD = 1.09). Finally, there are 57 children with ASD and 62 children with developmental delay (DD).
Results:
Using receiver operating characteristics (ROC), 2.25-2.75 was the optimal range of the total of the STAT as cut-off at Time 1. Using cutoff score of 2.25, the sensitivity was .88, and the specificity was .73. In addition, the positive predictive value (PPV) was 74.6%, while the negative predictive value (NPV) was 86.5%. Using cutoff score of 2.50, the sensitivity was .86, and the specificity was.81, the PPV was 80.3%, and the NPV was 86.2%. Using cutoff score of 2.75, the sensitivity was .74, and the specificity was.88, the PPV was 85.7%, and the NPV was 78.6%. The AUC was .87 at Time 1. Using ROC, 1.25-1.50 was the optimal range of the total of the STAT as cut-off at Time 2. Using cutoff score of 1.25, the sensitivity was .86, and the specificity was .71, the PPV was 73.1%, and the NPV was 84.6%. Using cutoff score of 1.50, the sensitivity was .70, and the specificity was.79, the PPV was 75.5%, and the NPV was 74.2%. The AUC was .82 at Time 2.
Conclusions:
The results of this study revealed that STAT showed high predicative validity for detecting ASD in toddlers before 24 months of age. This study indicated that the STAT is a promising good tool to differentiate the toddlers with ASD and toddlers with DD.