Department of Occupational Therapy, Taipei City Hospital Songde Branch, Taipei, Taiwan.
Department of Recreation and Sports Management, University of Taipei, Taipei, Taiwan.
Medicine (Baltimore). 2023 Aug 11;102(32):e34093. doi: 10.1097/MD.0000000000034093.
This study aimed to examine unidimensionality of the Cognitive Abilities Screening Instrument (CASI) using Rasch analysis and estimate Rasch person reliability in people with dementia. CASI data of people with dementia was collected from medical records of one general hospital in northern Taiwan. A total of 506 people with dementia were recruited from the Department of Neurology. The confirmatory factor analysis was first conducted to verify the fitness of one-factor model of the CASI. Unidimensionality was confirmed through 2 assumptions: the infit and outfit mean square were 0.5 to 1.5, and residual variance of the first principal component in principal component analysis was ≤20%. Rasch person reliability was estimated after undimensionality was supported. The results of one-factor model had shown that the Comparative Fit Index = 0.99, Tucker Lewis Index = 0.99, Root Means Square Error of Approximation = 0.015, and Standardized Root Mean Square Residual = 0.067, representing a good fit to the model. Both the infit and outfit mean square were ranged 0.87 to 1.37 and 0.86 to 1.42, respectively, and low residual variance of the first principal component (12.3%). Rasch person reliability result of 0.58 was satisfactory. The person-item map indicated the difference between item difficulty and person ability was within the acceptable limits (0.22 logits). Differential Item Function was found between -0.48 to 0.44 logits for gender, indicating the CASI functioned similarly for both genders. The 46 items of the CASI showed a unidimensional construct. The CASI had been demonstrated as a valid and reliable screening tool for assessing overall cognitive function in people with dementia, capturing their cognitive functions effectively.
本研究旨在使用 Rasch 分析检验认知能力筛选工具（CASI）的单维性，并估计痴呆患者的 Rasch 个体可靠性。从台湾北部一家综合医院的神经内科病历中收集了痴呆患者的 CASI 数据。共招募了 506 名来自神经内科的痴呆患者。首先进行验证性因素分析，以验证 CASI 的单因素模型的拟合度。通过两个假设来确认单维性：拟合和 outfit 均方在 0.5 到 1.5 之间，以及主成分分析中第一主成分的残差方差≤20%。在支持单维性后，估计 Rasch 个体可靠性。单因素模型的结果表明，比较拟合指数=0.99，塔克-刘易斯指数=0.99，近似均方根误差=0.015，标准化均方根残差=0.067，与模型拟合良好。拟合和 outfit 均方分别在 0.87 到 1.37 和 0.86 到 1.42 之间，第一主成分的残差方差较低（12.3%）。个体-项目图表明项目难度和个体能力之间的差异在可接受范围内（0.22 对数单位）。性别之间发现了差异项目功能，在-0.48 到 0.44 对数单位之间，表明 CASI 对两性的功能相似。CASI 的 46 个项目显示出单维结构。CASI 已被证明是一种有效的、可靠的筛选工具，用于评估痴呆患者的整体认知功能，能有效地捕捉他们的认知功能。
