Abstract：
Inappropriately reported or conducted studies may decrease the quality of care due to under- or overestimation of the benefits or harms of interventions. Our aim was to evaluate how often hand surgical randomized controlled trials (RCTs) use and report adequate methods to ensure internal validity, and whether inadequate reporting or methods are associated with the magnitude of treatment effect estimates.
Data Sources were the Cochrane Central Register of Controlled Trials, MEDLINE, and Embase databases until November 2020. We included published RCTs investigating the effects of any surgical intervention in the hand and wrist region. We assessed internal validity using the Cochrane Risk of Bias (RoB) tool for 6 domains: selection, performance, detection, attrition, selective reporting, and \"other\" bias. We extracted the primary outcome and calculated the effect size for each study. We used mixed-effect meta-regression to assess whether the RoB modified the magnitude of the effects.
For 207 assessed trials, the RoB was unclear or high for 72% in selection, 93% in performance, 88% in detection, 25% in attrition, 22% in selective reporting, and 34% in the \"other\" bias domain. Trials with a high or unclear risk of selection bias yielded 0.28 standardized mean difference (95% confidence interval, 0.02-0.55) larger effect sizes compared to studies with a low risk. Risks of bias for other domains did not modify the intervention effects. The risk for selection bias declined over time: the odds ratio for a high or unclear RoB was 0.90 (95% confidence interval, 0.85-0.95) per additional year of publication CONCLUSIONS: The internal validity and credibility of hand surgical RCTs can be improved by using established methods to achieve true randomization, blinding of the participants and study personnel, publishing the trial protocol and avoiding selective reporting of the outcomes, and reporting the trial as recommended in the Consolidated Standards of Reporting Trials statement.
Clinicians should be aware that RCTs that do not use or report proper randomization and allocation concealment may overestimate the treatment effects.
Data Sources were the Cochrane Central Register of Controlled Trials, MEDLINE, and Embase databases until November 2020. We included published RCTs investigating the effects of any surgical intervention in the hand and wrist region. We assessed internal validity using the Cochrane Risk of Bias (RoB) tool for 6 domains: selection, performance, detection, attrition, selective reporting, and \"other\" bias. We extracted the primary outcome and calculated the effect size for each study. We used mixed-effect meta-regression to assess whether the RoB modified the magnitude of the effects.
For 207 assessed trials, the RoB was unclear or high for 72% in selection, 93% in performance, 88% in detection, 25% in attrition, 22% in selective reporting, and 34% in the \"other\" bias domain. Trials with a high or unclear risk of selection bias yielded 0.28 standardized mean difference (95% confidence interval, 0.02-0.55) larger effect sizes compared to studies with a low risk. Risks of bias for other domains did not modify the intervention effects. The risk for selection bias declined over time: the odds ratio for a high or unclear RoB was 0.90 (95% confidence interval, 0.85-0.95) per additional year of publication CONCLUSIONS: The internal validity and credibility of hand surgical RCTs can be improved by using established methods to achieve true randomization, blinding of the participants and study personnel, publishing the trial protocol and avoiding selective reporting of the outcomes, and reporting the trial as recommended in the Consolidated Standards of Reporting Trials statement.
Clinicians should be aware that RCTs that do not use or report proper randomization and allocation concealment may overestimate the treatment effects.
摘要：
不适当的报告或进行的研究可能会由于低估或高估干预措施的益处或危害而降低护理质量。我们的目的是评估手外科随机对照试验（RCT）使用的频率，并报告适当的方法以确保内部有效性。以及不充分的报告或方法是否与治疗效果估计的大小相关。
数据来源是Cochrane中央对照试验登记册，MEDLINE,和Embase数据库，直到2020年11月。我们纳入了已发表的RCT，以调查在手和腕部区域进行任何手术干预的效果。我们使用Cochrane偏差风险(RoB)工具评估了6个领域的内部有效性：选择，性能，检测,自然减员,选择性报告，和“其他”偏见。我们提取了主要结果，并计算了每项研究的效应大小。我们使用混合效应元回归来评估RoB是否改变了效应的大小。
对于207项评估试验，RoB在选择中不清楚或高72%，93%的性能，88%的检测，25%的减员，22%的选择性报告，34%属于“其他”偏见领域。选择偏倚风险较高或不清楚的试验产生了0.28的标准化平均差（95％置信区间，0.02-0.55）与低风险研究相比，效应大小更大。其他领域的偏倚风险并未改变干预效果。选择偏差的风险随着时间的推移而下降：高或不清楚的RoB的比值比为0.90（95％置信区间，0.85-0.95）每发表一年结论：通过使用既定的方法实现真正的随机化，可以提高手外科RCT的内部有效性和可信度，致盲参与者和研究人员，公布试验方案，避免选择性报告结果，并按照合并报告标准试验声明中的建议报告试验。
临床医生应该意识到，不使用或报告正确的随机化和分配隐藏的RCT可能会高估治疗效果。
数据来源是Cochrane中央对照试验登记册，MEDLINE,和Embase数据库，直到2020年11月。我们纳入了已发表的RCT，以调查在手和腕部区域进行任何手术干预的效果。我们使用Cochrane偏差风险(RoB)工具评估了6个领域的内部有效性：选择，性能，检测,自然减员,选择性报告，和“其他”偏见。我们提取了主要结果，并计算了每项研究的效应大小。我们使用混合效应元回归来评估RoB是否改变了效应的大小。
对于207项评估试验，RoB在选择中不清楚或高72%，93%的性能，88%的检测，25%的减员，22%的选择性报告，34%属于“其他”偏见领域。选择偏倚风险较高或不清楚的试验产生了0.28的标准化平均差（95％置信区间，0.02-0.55）与低风险研究相比，效应大小更大。其他领域的偏倚风险并未改变干预效果。选择偏差的风险随着时间的推移而下降：高或不清楚的RoB的比值比为0.90（95％置信区间，0.85-0.95）每发表一年结论：通过使用既定的方法实现真正的随机化，可以提高手外科RCT的内部有效性和可信度，致盲参与者和研究人员，公布试验方案，避免选择性报告结果，并按照合并报告标准试验声明中的建议报告试验。
临床医生应该意识到，不使用或报告正确的随机化和分配隐藏的RCT可能会高估治疗效果。
