研究目的：本研究目的探究「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)」與「克氏行為量表(Clancy Behavior Scale, CBS)」篩檢2歲以下嬰幼兒；經過18個月後的短期預測正確率，及兩者與語言能力、問題行為，及自閉症症狀的關係。研究方法：本研究首次收案時間為2012年6月至2014年6月，追蹤時間為18個月後，即2013年12月至2015年12月。參與者為18至24個月的自閉症類群障礙症(Autism Spectrum Disorder, ASD)或發展遲緩嬰幼兒各25名，接受T-STAT評估，並由家長填寫CBS。18個月進行追蹤評估確認診斷，評估語言能力、問題行為與自閉症症狀。研究結果：T-STAT的短期預測正確率相當理想(76至88%)，但是CBS短期預測正確率不佳(40至76%)。ASD組，T-STAT和18個月後的語言能力、自閉症症狀顯著相關；CBS和18個月後的外化問題顯著相關。發展遲緩組，CBS和18個月後的外化問題顯著相關。研究結論：本研究結果顯示T-STAT短期預測穩定度佳，再次支持T-STAT是具備良好效度的層次二ASD早期篩檢工具。CBS和追蹤時的問題行為有關，顯示家長對小孩的自閉症症狀解釋為行為特質而非障礙。因此，需要教導家長ASD早期症狀知識，協助其辨識小孩早期症狀。未來的研究可以募集更多參與者，持續驗證本研究結果。

PURPOSE: The purpose of this study is to investigate the short-term predictive accuracy of the Screening Tool for Autism in Two-Year-Olds Taiwan Version (T-STAT) and the Clancy Behavior Scale (CBS) in detecting autism spectrum disorder (ASD) in toddlers before age 2. The T-STAT is an interactive tool that allows an examiner to assign a classification to ASD in toddlers, and the CBS is a parent-report questionnaire for caregivers to fill out based on the daily behaviors of their children. We examined the relationship between a child’s T-STAT and CBS scores at the initial assessment and the child’s language ability, behavioral problems, and autistic symptoms at 18 months follow-up. METHODS: Participants included 50 toddlers aged 18 to 24 months, 25 with ASD and 25 with developmental delays. All toddlers were first recruited for participation between June 2012 and June 2014, and the follow up were executed between December 2013 and December 2015. Both the T-STAT and CBS were used to evaluate each toddler at an initial assessment. Eighteen months later, the measures of children’s language ability, behavioral problems, and autistic symptoms were administered to the same participants whose diagnoses were also confirmed at the time. RESULTS: The T-STAT classification had high short-term predictive accuracy (76~88%). However, the CBS classification had poor short-term predictive accuracy (40~76%). For the toddlers with ASD, the T-STAT score significantly correlated with language ability and autistic symptoms, and the CBS score significantly correlated with externalizing behavioral problems. For toddlers with developmental delays, only the CBS score significantly correlated with externalizing behavioral problems. CONCLUSIONS: The results indicated high short-term predictive accuracy for both tests for toddlers with ASD before age 2, supporting previous assertions that the T-STAT is a promising level-two screening tool for differentiating toddlers with ASD from toddlers with developmental delays. The finding that the T-STAT score correlated with language ability and autistic symptoms, and the CBS score correlated with externalizing behavioral problems indicates that parents think of autistic symptoms as behavioral characters instead of impairments. Parents need education regarding early signs of ASD to enhance their ability to recognize early indicators. The next step for future research is to replicate these findings with a larger cohort of participants.