中文版PROMIS癌症特异性抑郁项目库计算机自适应测试版本的构建及测试  

Development and testing of a computer adaptive testing of the Chinese version of PROMIS cancer depression item bank

作　　者：朱瑞 杨瑒 蔡婷婷 吴傅蕾 周婷婷 袁长蓉 ZHU Rui;YANG Yang;CAI Ting-ting;WU Fu-lei;ZHOU Ting-ting;YUAN Chang-rong(School of Nursing,Fudan University,Shanghai 200032,China;Department of Nursing,Fudan University Shanghai Cancer Center/Department of Oncology,Shanghai Medical College,Fudan University,Shanghai 200032,China)

机构地区：[1]复旦大学护理学院,上海200032 [2]复旦大学附属肿瘤医院护理部/复旦大学上海医学院肿瘤学系,上海200032

出　　处：《复旦学报（医学版）》2023年第4期567-574,共8页Fudan University Journal of Medical Sciences

基　　金：国家自然科学基金面上项目(72074054)。

摘　　要：目的评估中文版患者报告结局测量信息系统(Patient Reported Outcome Measurement Information System,PROMIS)癌症特异性抑郁项目库的心理测量学属性,构建计算机自适应测试版本(PROMIS Cancer Depression CAT),并评价其效果。方法采用便利抽样法选取2020年11月—2021年7月在上海2所三级甲等医院的648例癌症患者作为研究对象。在心理测量学评估中,首先基于Rasch模型采用残差的主成分分析、条目残差相关系数、条目特征曲线检验PROMIS癌症特异性抑郁项目库的单维性、局部独立性、单调性假设;其次,通过Rasch模型中的选项特征曲线对选项进行拟合,采用Rasch模型中的Infit MNSQ、Outfit MNSQ、难度参数对每个条目进行拟合;最后,将经过校准后的项目库内容作为PROMIS Cancer Depression CAT的题库,采用Post-hoc模拟方法评价PROMIS Cancer Depression CAT的测量效率和准确性。结果PROMIS癌症特异性抑郁项目库由测试解释的经验方差为68.6%,条目之间的残差相关值都小于0.70,条目特征曲线是单调递增曲线。PROMIS癌症特异性抑郁项目库中的5个选项的峰值均未被相邻选项的特征曲线覆盖,5个选项之间阈值的间隔均大于1且小于5,条目的Infit MNSQ值为[0.54,1.49],Outfit MNSQ值为[0.55,1.49],选项和条目的拟合均在理想范围内,可作为构建PROMIS Cancer Depression CAT的项目库。Post-hoc模拟执行CAT的θ值与整个项目库的θ值相关系数为0.964,选择的平均使用条目数为6.348个。结论中文版PROMIS癌症特异性抑郁项目库满足单维性、局部独立性以及单调性假设,可以用来构建PROMIS Cancer Depression CAT。基于PROMIS癌症特异性抑郁项目库参数构建的计算机自适应测试版本能够高效、准确地测量癌症患者的抑郁水平。Objective To evaluate the psychometric properties of the Chinese version of the Patient Reported Outcome Measurement Information System(PROMIS)Cancer Depression Item Bank,to develop a PROMIS Cancer Depression CAT and evaluate its effectiveness.Methods From Nov 2020 to Jul 2021,a total of 648 cancer patients in 2 GradeⅢ-A hospitals in Shanghai were selected by convenience sampling.In psychometric assessment,principal component analysis of residuals,item residual correlation coefficient,and item characteristic curve were used to test the unidimensional,local independence,and monotonicity hypothesis of the PROMIS Cancer Depression Item Bank based on Rasch model.Then,the options were fitted with the Category Response Curves in the Rasch model,and each item was fitted with Infit MNSQ,Outfit MNSQ,and difficulty parameters in the Rasch model.Finally,Post-hoc simulation was used to evaluate the measurement efficiency and accuracy of PROMIS Cancer Depression CAT by taking the calibrated item bank as the test bank of PROMIS Cancer Depression CAT.Results The empirical variance of test interpretation in the PROMIS Cancer Depression Item Bank was 68.6%,the residual correlation values between the items were all less than 0.70,and Item Characteristic Curve was monotonically increasing.The peak values of the five options in the PROMIS Cancer Depression Item Bank were not covered by the characteristic curves of adjacent options.The threshold intervals between the five options were greater than 1 and less than 5,and the Infit MNSQ value of the item was[0.54,1.49]and the Outfit MNSQ value was[0.55,1.49].The fitting of the options and items was within the ideal range and could be used as the test bank of PROMIS Cancer Depression CAT.The correlation coefficient between theθvalue of CAT in Post-hoc simulation and theθvalue of the full item bank was 0.964,and the average number of selected items was 6.348.Conclusion The Chinese version of PROMIS Cancer Depression Item Bank meets the assumptions of unidimensionality,local independen

关 键 词：患者报告结局测量信息系统(PROMIS) 癌症 抑郁 计算机自适应测试(CAT) RASCH模型 

分 类 号：R473.73[医药卫生—护理学]