健康与遗传心理学研究室知识库

    目的：轻度认知障碍(mild cognitive impairment，MCI)属于正常老化与老年期痴呆之间的过渡阶段，也是转归为老年期痴呆，尤其是阿尔茨海默病的高危人群。蒙特利尔测验(montreal cognitive assessment, MoCA)是专门用于筛查MCI患者的认知测验。MoCA于2006年由王炜等人进行适当本土化之后引入我国，由于我国地域差异较大，目前国内有五个版本的MoCA，其中以北京版(MoCA-BJ)使用最多。MoCA-BJ在国内临床和研究中部分研究者认为部分项目较难或存在文化差异并不适合我国人群，需要进行修改。研究显示MoCA-BJ得分可能受教育、年龄、性别等因数影响，建议建立常模，而少有研究关注测验是否存在公平性问题。一般认为MoCA-BJ是一个用时大概10分钟的快速测验工具，但在实际应用中，测验时间远不止10分钟，且在临床筛查和实验研究中MoCA-BJ通常结合了其他筛查工具和方法使用，有必要对MoCA-BJ进行简化缩短期耗时从而扩大使用范围。基于以上，本研究目的一是对MoCA-BJ部分项目的进行修改形成修改版MoCA(MoCA-Re)并验证前人研究质疑；二是探讨MoCA-BJ是否在教育、性别、地区因素上存在功能性差异。三是对MoCA-BJ进行简化，形成简版MoCA并验证其适用性。
    方法：研究一对MoCA-BJ部分项目的进行修改形成修改版(MoCA-Re)，依据项目反应原理(item response theory, IRT)，选择二参数等级反应模型(Graded ResponseModel, GRM)比较俩个版本的项目特征曲线、项目信息曲线及测验信息曲线，进一步判断MoCA-BJ是否需要进行修订。研究二基于IRT，选用有序逻辑斯蒂克回归(ordinal logistic regression, LOR)模型分析MoCA-BJ基于教育、性别、地区分组的DIF，进而探讨测验公平性问题。研究三通过计算机自适应技术提取出MoCA-BJ中使用频率在4％以上的项目结合实际经验形成简版MoCA(s-MoCA)并验证其检测力。
    结果：研究一MoCA-Re与MoCA-BJ相比，虽然在项目难度上有所降低，但每个项目内部难度等级之间变得混乱，且区分度受到影响。项目的信息量和测验信息量也表明，MoCA-BJ不管是在项目信息量还是在测验信量都要大于MoCA-Re的信息量。
    研究二结果显示，以教育分组，同等认识能力水平的被试有7个项目存在测量不公平性，即七个项目存在DIF。以性别分组，有十个项目存在DIF。以地区分组，共有八个项目存在DIF。测验功能差异结果显示，教育分组下MoCA-BJ版存在轻微测验功能差异，表现为高教育水平老人较低教育水平老人更容易做出正确反应。但由于两组教育水平的老人潜在特质分布存在一定差异，可能对DIF检测力产生影响，故测验在教育水平上存在DIF的结果需谨慎对待。基于性别分类和社区分类几乎不存在测验功能差异。
    研究三基于IRT结合计算机自适应（computerised adaptive testing, CAT）对MoCA-BJ进行简化后生成简版MoCA，共包含15道题，满分17分，测验耗时远小于北京版。由于简版MoCA几乎包含了MoCA-BJ版所有维度的项目，较好的保留了MoCA-BJ测量整体认知的功能，可适用于不同类型的认知损伤。ROC分析结果显示不管是用于区分健康人和总体认知障碍患者，还是健康人和MCI患者亦或是区分MCI患者和Dem患者，简版MoCA的检测力和MoCA-BJ检测力均无差异。
    结论： MoC-BJ 各项参数优于MoCA-Re，一定意义上回应了前人的质疑。MoCA-BJ 存在教育不公平性，同等认知水平的的老人中高教育老人较低教育老人更容易做出正确的反应。简版MoCA 测验将原本的28 个项目降低到17 个项目，但对MCI 老人和痴呆老人的检测力与MoCA-BJ 的检测力几乎无差别，且测验耗时大大降低，可推荐用于大型流调和临床门诊快查筛查。

    Objective: Mild cognitive impairment (MCI) is a transitional phase between normal aging and dementia, and it is a high-risk group that is converted to dementia, especially the group who has Alzheimer's disease. The Montreal Cognitive Assessment (MoCA) is a test to screen patients with MCI. The MoCA for screening Chinese MCI elderly can be divided into five editions based on geographical differences. The Beijing edition is widely used in clinical and research more than others. Some researchers believe that partail items of MoCA-BJ are not suitable for Chinese patiences because of cultural differences, and hard to conduct. Therefore, the MoCA-BJ need to be modified properly. Many test results will be affected by factors such as education, age and gender. The MoCA is no exception. In general, most researchers use these factors to classify elders as a mode to reslove the problem. Nevertheless, only few studies focus on the test itself, or argue whether there is a fairness problem during the test. MoCA is generally considered to be a rapid test tool that takes about 10 minutes. However, in our actual application, the time is much longer than 10 minutes. In clinical screening and experimental researches, MoCA is not only apliied singly, instead of frequently used combined with other screening tools.As a result, MoCA test time need to be shorttened whereby expand the range of application. Based on the above, the purpose of the study is to modify the partial items of MoCA-BJ to form the revised version MoCA (MoCA-Re) to verify the previous research conclusions. Secondly, we will explore whether there have functional differences in education, gender and regional factors of MOCA-BJ. Thirdly, we want to simplify the MOCA-BJ, then form a short form of MoCA (s-MoCA) and verify its applicability.
    Method: There are three studies in thesis. The first one based on the questions raised by previous studies, MoCA-BJ was revised to form a modified version (MoCA-Re). According to the item response theory, the comparison of the items characteristics curve and test information curve of the two versions determines whether MoCA-BJ needs to be revised. The second one analyzes the DIF of each item in the MoCA test to determine whether the test is unfair. The last one is focus on the issue of simplification of the MoCA-BJ. Both of the item response theory and computerised adaptive testing simulation are devotaed to establish an abbreviated MoCA.
    Results: Compared with the MoCA-BJ, although the difficulty level of the items are reduced, but there is confusion between the difficulty levels inside each item. At the same time, the degree of discrimination is reduced as well. The item information and the test information indicates that the MoCA-BJ is better than the edition provided by the revised version.
    The second research shows that in the group of education, the participants with the same level of cognitive ability have unfair measurement in 7 items which associated with DIF. Grouped by sex category, there are 10 items belong to DIF items.
Grouped by region, there are 8 items belong to DIF items. Overall, for the elderly with different levels of education but the same level of cognitive ability, the MoCA-BJ has a slight difference test function, showing that the overall score of the higher education level is higher than the low education level of the elderly. There is almost no difference about gender and community classification.
    The results of the third study is based on the IRT combined with computerized adaptive testing simulation to simplify the MoCA (s-MoCA). It contains 15 items, out of 17 points, and the test takes less minutes than MoCA-BJ. Since the items s-MoCA contained almost includes all dimensions of the MoCA-BJ, the simplified version can be considered as a whole cognitive test whereby applied to different types of cognitive impairments. The ROC data analysis results shows that no matter whether it is used to distinguish between healthy people and patients with general cognitive impairment, healthy people and MCI patients, or MCI patients and Dem patients, Specificity and other indicators are not significantly different.
   Conclusion: We modified partail items of MoCA-BJ to form the revised version MoCA (MoCA-Re). The comparison results of two versions show that MOC-BJ parameters better than MoCA-Re. The MOCA-BJ has education’ unfairness, the same cognitive level of the elderly with higher education level is easier to make the right response. The s-MoCA, which was formed using computer adaptive simulation technology, reduced the original 28 items to 17 items, however, the detection ability of the elderly with MCI and dementia is almost no different from that of the MOCA-BJ, and the test time is reduced which can be recommended for the rapid screening of flow and harmonic clinical outpatient clinics.