Choi Seong Hye, Park Moon Ho
Department of Neurology, Inha University College of Medicine, Incheon, Korea.
Department of Neurology, Korea University College of Medicine, Seoul, Korea.
Geriatr Gerontol Int. 2016 Feb;16(2):252-8. doi: 10.1111/ggi.12464. Epub 2015 Feb 6.
To screen for and determine cognitive dysfunction, cognitive tests and/or informant reports are commonly used. However, these cognitive tests and informant reports are not always available. The present study investigated three screening methods using the Mini-Mental State Examination (MMSE) as the cognitive test, and the Korean dementia screening questionnaire (KDSQ) as the informant report.
Participants were recruited from the Korea Clinical Research Center for Dementia of South Korea, and included 2861 patients with Alzheimer's disease (dementia), 3519 patients with mild cognitive impairment and 1375 controls with no cognitive dysfunction. Three screening methods were tested: (i) MMSE alone (MMSE(cut-off) ); (ii) a conventional combination of MMSE and KDSQ (MMSE+KDSQ(cut-off) ); and (iii) a decision tree with MMSE and KDSQ (MMSE+KDSQ(decision tree) ).
For discriminating any cognitive dysfunction from controls, MMSE+KDSQ(cut-off) had the highest area under the receiver operating characteristic curve (0.784). For discriminating dementia from controls, MMSE+KDSQ(cut-off) had the highest area under the receiver operating characteristic curve (0.899). For discriminating mild cognitive impairment from controls, MMSE(cut-off) had the highest area under the receiver operating characteristic curve (0.683). MMSE+KDSQ(decision tree) showed the highest sensitivity for all discriminations. For overall classification accuracy, MMSE+KDSQ(decision tree) had the highest value (70.0%).
These three methods had different advantageous properties for screening and staging cognitive dysfunction. As there might be different availability across clinical settings, these three methods can be selected and used according to situational needs.
为筛查和确定认知功能障碍，通常会使用认知测试和/或知情者报告。然而，这些认知测试和知情者报告并非总是可用。本研究调查了三种筛查方法，使用简易精神状态检查表（MMSE）作为认知测试，韩国痴呆筛查问卷（KDSQ）作为知情者报告。
参与者从韩国临床痴呆研究中心招募，包括2861例阿尔茨海默病（痴呆）患者、3519例轻度认知障碍患者和1375例无认知功能障碍的对照者。测试了三种筛查方法：（i）仅使用MMSE（MMSE（临界值））；（ii）MMSE和KDSQ的传统组合（MMSE + KDSQ（临界值））；以及（iii）结合MMSE和KDSQ的决策树（MMSE + KDSQ（决策树））。
为区分对照者与任何认知功能障碍，MMSE + KDSQ（临界值）在受试者工作特征曲线下的面积最高（0.784）。为区分对照者与痴呆，MMSE + KDSQ（临界值）在受试者工作特征曲线下的面积最高（0.899）。为区分对照者与轻度认知障碍，MMSE（临界值）在受试者工作特征曲线下的面积最高（0.683）。MMSE + KDSQ（决策树）在所有区分中显示出最高的敏感性。对于总体分类准确性，MMSE + KDSQ（决策树）的值最高（70.0%）。
这三种方法在筛查和分期认知功能障碍方面具有不同的优势特性。由于不同临床环境下的可用性可能不同，可根据实际情况选择并使用这三种方法。
