临床试验文献质量评价量表的制作和评价

目的：制备中文临床试验报告质量评价量表，并且对量表的信度和效度进行初步评价。方法：采用专家咨询法（Delphi法）和小组讨论相结合的方法制定量表，咨询专家包括临床药理学家、临床医生、流行病学家、卫生统计学家和医学杂志编辑。专家咨询分三个阶段，形成临床试验文献质量评价量表及量表说明。为评价量表，从2000年治疗高血压的英文临床试验文献中抽取30篇，由两个评价人员应用量表独立进行评价，计算组内相关系数（ICC）评价量表的信度；对文献同时应用CONSORT声明评价作为对照，计算Kendall等级相关系数来判断其效度。结果：经过10名专家第一阶段的咨询，协调系数W为0．505，19个评价项目高度一致；第二阶段咨询，对19个项目赋予权重，共计100分，完成量表制作。信度评价中总分的组内相关系数ICC为0．921，其中引言部分0．987，材料与方法0．875，实施与结果部分0．902，讨论及其他部分0．966，P均〈0．01。与CONSORT声明的评价结果比较，Kendall相关系数r为0.723，P〈0．01，具有较好的一致性。结论：本研究形成的量表制订过程规范，条目清楚，具有较好的信度和效度。

Development and Evaluation of the Scale for Assessing the Report Quality of Clinical Trials

Objective:To develop the instrument for the evaluation of the report quality of clinical trials with the modified Delphi method in combination with panel discussions, and assess its validity and reliability. Method: For the development of the instrument, the modified Delphi method in combination with a panel discussion was applied. The consulting panel included pharmacologists, clinical doctors, epidemiologists, statisticians and editors of journal. 30 reports for the therapy of hypertension published in 2000 were selected for the estimation of the instrument of the validity and reliability. The interclass correlation coefficient (ICC) was used to test the reliability. The CONSORT statement was used as a control, and a Kendall correlation was worked out to estimate the validity. Result:A total of 10 specialists took part in the consultation. During the first round of consultation, the expert harmonious coefficient (W) reached to 0. 505 and 19 items were identified. A total of 100 scores were distributed to the 19 items in the second round, and finally the scale was completed. After scoring the 30 reports on hypertension therapy by two readers independently, the interclass correlation coefficients (ICC) of the total score were 0.921,and 0.987 in the introduction part, 0.875 in materials and methods, 0.902 in results, and 0.966 in the discussion and others, P <0.01. In comparison with the result of CONSORT statement, the Kendall coefficient of the result of evaluation was 0.723, P<0.01, which indicated that the validity was good. Conclusion: The scale has been developed in a relatively strict procedure, and the validity and reliability are good.