自闭症谱系障碍的早期筛查工具

基于35项实证研究, 遵循诊断准确性研究的质量评估(QUADAS-2)的循证程序对自闭症谱系早期预警特征筛查工具开展质量评估, 包括婴儿阶段(10个筛查工具, 159388人)、幼儿阶段(14项筛查工具, 11712人), 旨在为自闭症谱系风险的早期识别提供全面的理解。结果表明：幼儿阶段的筛查工具的分类准确性要高于婴儿阶段, 在婴儿阶段达到良好水平的筛查工具有M-CHAT-R/F、PDQ-1; 在幼儿阶段达到优秀水平的筛查工具有OERA、TIDOS。其中, “改良的幼儿自闭症检查表-有修订的后续随访” (M-CHAT-R/F)是目前最具潜力的自闭症谱系风险筛查工具之一。最后, 我们探讨了应用QUADAS-2评估研究质量的局限性, 强调需要规范工具的质量评估标准与进一步验证研究的必要性。

Early screening tools for Autism Spectrum Disorder in infancy and toddlers

The purpose of screening research is to provide a comprehensive understanding about the risks of early identification of autism spectrum disorder. This study comprehensively and systematically reviewed the commonly-used screening tools for early detection of autism in infants and preschool children. Under the guidelines of the Preferred Reporting Items for Systematic review and Meta-Analysis (PRISMA), we queried the topic “screening tools for autism spectrum disorder” against the three major literature retrieval databases (WOS, Scopus, PubMed) and obtained 175 articles, of which 35 articles and 18 corresponding screening tools were further evaluated to comply with the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) checklist. These tools were analyzed from the aspects of application level, psychometric and measurement properties, cultural adaptability, diagnostic criteria and classification accuracy. From the aspect of application level, the screening tools were categorized into three levels based on the types of screening samples, wherein Level- I screening tools, totaling up to 4, focus on general-purpose screening of population and the examination of other developmental disorders, and the evaluation is mainly based on the reports from parents; Level-II screening tools, totaling up to 8, were mainly based on the clinical reports from well-trained clinicians or the observations from the interactions between trained professionals and children; the hybrid-level screening tools, totaling up to 6, were based on the Level-I general population and the Level-II clinical samples. From the aspect of psychometric and measurement properties, existing studies only adopted several metrics of reliability and validity, e.g., internal consistency reliability, rater reliability, test-retest reliability and criterion validity, to evaluate the psychometric properties of screening tools,, whereas the other metrics such as split half reliability, measurement error, content validity, structural validity, cross-cultural validity and hypothesis test were seldom adopted except in only a few studies, and one-third studies did not carry out any psychometric evaluation, potentially resulting in the high-risk bias of QUADAS-2. From the aspect of cultural adaptability, the series of M-CHAT tools（M-CHAT, M-CHAT/F,M-CHAT-R/F）, subjected to extensive evaluation in multiple languages, have demonstrated obvious cultural adaptability, and were recognized as the most widely-used and the most famous screening tool based on parent reports. However, there is an increasing trend that screening tools developed to be localized to the culture and languages in communities or countries with scarce resources. From the aspect of diagnostic standards, the DSM-IV or DSM-IV-TR (accounting for 43%) were still used as the main reference standard instead of the DSM-5 manual (accounting for 34%). It was concluded from sensitivity and specificity that the tools for preschool children outperform those for infants in terms of classification accuracy. The tools rated as good level for infants include M-CHAT-R/F and PDQ-1, and the tools rated as excellent level for children include OERA and TIDOS, wherein M-CHAT-R / F is one of the most promising screening tools for estimating the risks of autism spectrum disorder. Lastly, we discussed the limitations of QUADAS-2 and the necessity of choosing stringent quality assessment measures and the importance of further validating these measurements.