嬰幼兒綜合發展測驗之判定準確度及切截點分析

Assessment Accuracy and Cut-off Points of Comprehensive Developmental Inventory for Infants and Toddlers (CDIIT)

本研究主要目的在探究運用「嬰幼兒綜合發展測驗」（CDIIT）所得結果來判定發展遲緩的準確度及最佳切截點。依醫學診斷或鑑定老師研判為「發展遲緩」和「正常發展」之結果為效標，以760 名3 至71 個月之幼兒（發展遲緩者152 人，正常發展者608 人）在CDIIT 診斷和篩選測驗總分所得之發展商數和Z 分數，進行「接受者操作特徵」（Receiver-Operating-Characteristic, ROC）分析，以探究該測驗之診斷與篩選準確率，並從敏銳度、特異度、陽性概率比、陰性概率比、擊中率、約登指數等指標，綜合研判測驗最佳切截點。結果顯示，CDIIT 篩檢測驗的ROC 曲線下面積為.92、診斷測驗為.96，篩檢或診斷之準確率極高。根據上述指標綜合研判的結果，CDIIT 篩選測驗總分所得Z 分數用以判斷疑似遲緩的最佳切截點為-1，而診斷測驗總分所得DQ 用以判斷發展遲緩的最佳切截點為77.5（即-1.5SD）。

This study had two aims: first, to use Receiver Operating Characteristic (ROC) analysis to evaluate the ability of a Comprehensive Developmental Inventory for Infants and Toddlers (CDIIT) to discriminate developmentally delayed cases from normal cases; secondly, to determine the cut-off points which could assist assessors to identify cases with developmental delays. Total of 760 children aged from three to seventy-one months old, including 152 children with developmental delays and 608 normally developing children were involved. As the results showing from the area under the ROC curve (AUC), both CDIIT Screening Test (AUC=.92) and Diagnostic Test (AUC=.96) showed high accuracy for screening and diagnosis purposes. Judging from true positive fraction (TPF, also known as sensitivity), false positive fraction (FPF), positive likelihood ratio (LR+), negative likelihood ratio (LR-), hit rate, Youden Index, and Kappa, Z scores less than -1 and developmental quotients (DQ) less than 77.5 (-1.5SD) were identified as the best cut-off points for the CDIIT Screening and Diagnostic Test. That is, children whose total scores fall below the cut-off points can be considered as having “suspected developmental delay” in the Screening Test or “developmental delay” in the Diagnostic Test.