2008年和2013年使用改良版“偏倚风险”工具对五家顶尖神经病学期刊中随机对照试验报告质量的研究

Quality of Reporting Randomized Controlled Trials in Five Leading Neurology Journals in 2008 and 2013 Using the Modified "Risk of Bias" Tool.

作者信息

Zhai Xiao, Cui Jin, Wang Yiran, Qu Zhiquan, Mu Qingchun, Li Peiwen, Zhang Chaochao, Yang Mingyuan, Chen Xiao, Chen Ziqiang, Li Ming

机构信息

Department of Orthopedics, Changhai Hospital affiliated to the Second Military Medical University, Shanghai, PR China.

Graduate Management Unit, Changhai Hospital affiliated to the Second Military Medical University, Shanghai, PR China.

出版信息

World Neurosurg. 2017 Mar;99:687-694.e7. doi: 10.1016/j.wneu.2016.12.136. Epub 2017 Jan 9.

DOI:10.1016/j.wneu.2016.12.136

PMID:28089838

Abstract

BACKGROUND

To examine the risk of bias of methodological quality of reporting randomized clinical trials (RCTs) in major neurology journals before and after the update (2011) of Cochrane risk of bias tool.

METHODS

RCTs in 5 leading neurology journals in 2008 and 2013 were searched systematically. Characteristics were extracted based on the list of the modified Cochrane Collaboration's tool. Country, number of patients, type of intervention, and funding source also were examined for further analysis.

RESULTS

A total of 138 RCTs were enrolled in this study. The rates of following a trial plan were 61.6% for the allocation generation, 52.9% for the allocation concealment, 84.8% for the blinding of the participants or the personnel, 34.8% for the blinding of outcome assessment, 78.3% for the incomplete outcome data, and 67.4% for the selective reporting. A significant setback was found in "the selective reporting" in 2013 than that in 2008. Trials performed by multi-centers and on a large scale had significantly more "low risk of bias" trials. Not only the number of surgical trials (5.8%) was much less than that of trials using drugs (73.9%), but also the reporting quality of surgical trials were worse (P = 0.008). Finally, only 17.4% trials met the criterion of "low risk of bias."

CONCLUSIONS

The modified "risk of bias" tool is an improved version for assessment. Methodological quality of reporting RCTs in the 5neurology journals is unsatisfactory, especially that for surgical RCTs, and it could be further improved.

摘要

背景

评估Cochrane偏倚风险工具更新（2011年）前后主要神经学期刊中随机临床试验（RCT）报告的方法学质量的偏倚风险。

方法

系统检索2008年和2013年5种主要神经学期刊中的RCT。根据修改后的Cochrane协作组工具列表提取特征。还对国家、患者数量、干预类型和资金来源进行了检查以进行进一步分析。

结果

本研究共纳入138项RCT。分配方案产生的遵循试验方案率为61.6%，分配隐藏为52.9%，参与者或人员的盲法为84.8%，结果评估的盲法为34.8%，不完整结果数据为78.3%，选择性报告为67.4%。2013年“选择性报告”方面相较于2008年有显著退步。多中心和大规模开展的试验中“低偏倚风险”试验显著更多。外科试验的数量（5.8%）不仅远少于药物试验（73.9%），而且外科试验的报告质量更差（P = 0.008）。最后，只有17.4%的试验符合“低偏倚风险”标准。