Chen Xin-lin, Liu Feng-bin, Guo Li, Liu Xiao-bin
Department of Preventive Medicine and Health Statistics, College of Basic Medical Sciences, Guangzhou University of Chinese Medicine, Guangzhou 510407, Guangdong Province, China.
Zhong Xi Yi Jie He Xue Bao. 2010 Feb;8(2):121-5.
To investigate the scientificity of patient-reported outcome (PRO) scale for myasthenia gravis (MG), which was used to evaluate the clinical effects of traditional Chinese and Western medicine treatment on MG patients. Psychometric performance of the MG-PRO scale was also expected to be evaluated in this study.
A total of 100 MG patients and 100 healthy people were face-to-face interviewed by well-trained investigators, and the data of MG-PRO scale were collected. The classical theory test (CTT) and item response theory (IRT) methods were used to analyze the psychometric performance such as validity, reliability, person separation index (PSI) and differential item functioning (DIF) in the MG-PRO scale.
The results of CTT analysis showed that the split-half reliabilities of the MG-PRO scale and each dimension were greater than 0.7. In the analysis of internal consistency of each dimension, the Cronbach's alpha was greater than 0.8. Each facet had greater correlation with its dimension than the other dimensions. Four principal components were extracted by exploratory factor analysis, which represented all dimensions of the scale, and the cumulative variance was 55.54%. The scores of each of the 8 facets between MG patients and healthy people were different (P<0.01). The results of IRT showed that the PSI of each model was greater than 0.8, and all items did not have uniform DIF and non-uniform DIF.
The MG-PRO scale reflects the definition and connotation of quality of life and contains special issues of MG patients as well, and shows good reliability (split-half reliability, Cronbach's alpha), validity (content validity, construct validity, discriminate validity) from the results of CTT, and good psychometric performance from the results of IRT.
探讨用于评估中西医治疗重症肌无力（MG）患者临床疗效的患者报告结局（PRO）量表的科学性。本研究还期望评估MG-PRO量表的心理测量学性能。
由训练有素的调查员对100例MG患者和100名健康人进行面对面访谈，收集MG-PRO量表的数据。采用经典理论测试（CTT）和项目反应理论（IRT）方法分析MG-PRO量表的效度、信度、人区分指数（PSI）和项目功能差异（DIF）等心理测量学性能。
CTT分析结果显示，MG-PRO量表及其各维度的分半信度均大于0.7。在各维度的内部一致性分析中，Cronbach's alpha大于0.8。每个子维度与其所属维度的相关性高于其他维度。通过探索性因子分析提取了4个主成分，代表了量表的所有维度，累积方差为55.54%。MG患者与健康人8个子维度的得分均有差异（P<0.01）。IRT结果显示，各模型的PSI均大于0.8，所有项目均无一致DIF和非一致DIF。
MG-PRO量表反映了生活质量的定义和内涵，也包含了MG患者的特殊问题，从CTT结果看具有良好的信度（分半信度、Cronbach's alpha）、效度（内容效度、结构效度、区分效度），从IRT结果看具有良好的心理测量学性能。
