T-STAT評估18～24個月自閉症類嬰幼兒效度探究

Validation of “The Screening Tool for Autism in Two-Year-Olds Taiwan Version (T-STAT)” in Children Aged between 18 and 24 Months

研究目的：適用於2歲以下自閉症類疾患嬰幼兒的層次二篩檢工具很少。因此，本研究探究層次二篩檢工具「台灣版兩歲期自閉症篩檢工具（Screening Tools for Autism in Two-Year-Olds Taiwan Version, T-STAT）」篩檢18~24個月自閉症類疾患嬰幼兒的效度。研究方法：本文研究期間為2011年6月至2013年6月。兩組18~24個月大的嬰幼兒，包括20位自閉症類疾患嬰幼兒（平均心理年齡15.34個月，全距12~20.5個月），及20位發展遲緩嬰幼兒（平均心理年齡15.11個月，全距12.5~19.5個月），使用T-STAT做為篩檢的評估工具。T-STAT是一項12題、互動式的自閉症篩檢工具，包含四種領域，分別是：遊戲（兩題）、要求（兩題）、共享式注意力（四題）與模仿（四題）。每一領域，依據失敗題目轉換成0~1分，總分為0~4分；分數越高代表社會溝通障礙越嚴重，自閉症類疾患風險越高。研究結果：利用訊號偵測理論決定T-STAT總分的切截分數，發現2.25與2.5都是不錯的選擇。以2.25作為切截分數，敏感度與特異度都是.95，陽性預測值（positive predictive value）與陰性預測值（negative predictive value）也都是.95。以2.5作為切截分數，敏感度與特異度分別是.90與.95；陽性預測值與陰性預測值分別是.95與.91。研究結論：研究結果支持T-STAT是有效的層次二自閉症類疾患嬰幼兒篩檢工具。基於早期篩檢與早期介入，宜使用2.25作為切截分數。後續研究需擴大樣本驗證效度，縱貫研究探究早期診斷穩定度。

Purpose: Few level 2 screening tools exist for detecting autism spectrum disorders (ASDs) in children under 24 months old. The Screening Tool for Autism in Two-Year-Olds Taiwan Version (T-STAT) is a level 2 screening tool and initially developed for children with ASDs between 24 and 35 months of age. The current study examines the validity of the T-STAT for children with ASDs between 18 and 24 months of age. Methods: All children were recruited for participation between June 2011 and June 2013. Two groups of 18 to 24 month olds were assessed, including 20 toddlers with ASDs (mean mental age=15.34 months, range=12~20.5 months) and 20 toddlers with developmental delays (mean mental age=15.11 months, range=12.5~19.5 months). The T-STAT was used to distinguish the toddlers with ASDs from toddlers with developmental delays. The T-STAT is a 12-item interactive autism-specific screening measure, including four behavioral domains: play (2 items), requesting (2 items), joint attention (4 items) and imitation (4 items). The number of failed items in each domain was averaged to obtain a domain score, which ranged from 0 to 1. The scores of the four domains were summed, obtaining a total T-STAT score from 0 to 4. A higher score indicated greater social-communicative impairment and high risk ASDs. Results: The optimal cutoff of the total T-STAT score was decided using signal detection procedures. The results indicated that either a 2.25 or 2.5 total T-STAT score were the best cutoffs. Using the 2.25 total score as a cutoff yielded both sensitivity and specificity of.95. In addition, both the positive predicative value (PPV) and negative predicative value (NPV) were.95. Using the 2.5 total score as a cutoff yielded sensitivity and specificity of .90 and .95, respectively. In addition, both the PPV and NPV were .95 and .91, respectively. Conclusions: The preliminary data from this study indicate that the T-STAT is a promising level 2 screening tool for differentiating toddlers with ASDs from toddlers with developmental delays. We suggest that a 2.25 total T-STAT score be used as a cutoff for early screening and intervention with toddlers with ASDs. Further studies are needed to replicate the results to establish validity and to provide longitudinal data to examine the stability of an early diagnosis.