Mahmoud Karim D, Lennon Ryan J, Holmes David R
Division of Cardiovascular Diseases, Mayo Clinic, Rochester, Minnesota; Department of Cardiology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands.
Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, Minnesota.
Am J Cardiol. 2015 Aug 1;116(3):355-63. doi: 10.1016/j.amjcard.2015.04.045. Epub 2015 May 9.
Randomized clinical trials (RCTs) are considered the gold standard for evidence-based medicine. However, an accurate estimation of the event rate is crucial for their ability to test clinical hypotheses. Overestimation of event rates reduces the required sample size but can compromise the statistical power of the RCT. Little is known about the prevalence, extent, and impact of overestimation of event rates. The latest RCTs on 10 preselected topics in the field of cardiovascular interventions and devices were selected, and actual primary event rates in the control group were compared with their respective event rate estimations. We also assessed what proportion of the nonsignificant RCTs was truly able to exclude a relevant treatment effect. A total of 27 RCTs randomizing 19,436 patients were included. The primary event rate in the control group was overestimated in 20 of the 27 RCTs (74.1%) resulting in a substantial relative difference between observed and estimated event rates (mean -22.9%, 95% confidence interval -33.5% to -12.2%; median -16.3%, 95% confidence interval -30.3% to -6.5%). Event rates were particularly overestimated in RCTs on biodegradable polymer drug-eluting coronary stents and renal artery stenting. Of the 14 single end point superiority trials with nonsignificant results, only 3 (21.4%) actually resulted in truly negative conclusions. In conclusion, event rates in RCTs evaluating cardiovascular interventions and devices are frequently overestimated. This under-reported phenomenon has fundamental impact on the design of RCTs and can have an adverse impact on the statistical power of these trials to answer important questions about therapeutic strategies.
随机临床试验（RCT）被认为是循证医学的金标准。然而，准确估计事件发生率对于其检验临床假设的能力至关重要。事件发生率的高估会减少所需样本量，但可能会损害RCT的统计效力。关于事件发生率高估的普遍性、程度和影响，人们知之甚少。我们选取了心血管介入和器械领域10个预选主题的最新RCT，将对照组的实际主要事件发生率与其各自的事件发生率估计值进行比较。我们还评估了非显著性RCT中真正能够排除相关治疗效果的比例。共纳入了27项随机分配19436例患者的RCT。27项RCT中有20项（74.1%）高估了对照组的主要事件发生率，导致观察到的和估计的事件发生率之间存在显著的相对差异（平均-22.9%，95%置信区间-33.5%至-12.2%；中位数-16.3%，95%置信区间-30.3%至-6.5%）。在关于可生物降解聚合物药物洗脱冠状动脉支架和肾动脉支架的RCT中，事件发生率尤其被高估。在14项结果无显著性的单终点优效性试验中，只有3项（21.4%）实际上得出了真正的阴性结论。总之，评估心血管介入和器械的RCT中的事件发生率经常被高估。这种报告不足的现象对RCT的设计有根本性影响，并可能对这些试验回答有关治疗策略的重要问题的统计效力产生不利影响。
