Robinson N Bryce, Rahouma Mohamed, Audisio Katia, Cancelli Gianmarco, Demetres Michelle, Soletti Giovanni, Hameed Irbaz, Girardi Leonard N, Ruel Marc, Fremes Stephen E, Gaudino Mario
Department of Cardiothoracic Surgery, Weill Cornell Medicine, New York, New York.
Division of Cardiothoracic Surgery, Yale University School of Medicine, New Haven, Connecticut.
Ann Thorac Surg Short Rep. 2023 Jun 12;1(3):537-541. doi: 10.1016/j.atssr.2023.05.017. eCollection 2023 Sep.
This analysis was conducted to characterize contemporary randomized controlled trials (RCTs) in cardiothoracic surgery.
We selected randomized controlled trials published in the journals with the highest impact factor in medicine, general surgery, and cardiothoracic surgery and published between 2008 and 2020. Trial characteristics as well as measures of reporting and quality were summarized and compared.
Ninety-three trials were included; 44 (47.3%) were prospectively registered and 14 (31.8%) had a discrepancy between the registered and published primary outcome. Most trials (n = 83 [89.1%]) used a superiority design, a composite primary outcome (n = 82 [88.2%]), and a major clinical event as the primary end point (n = 67 [72.0%]). Blinding was used infrequently, and most trials did not control for surgeon experience (n = 74 [79.5%]) or monitor the intervention (n = 90 [96.7%]). Twenty-four (25.8%) trials had high risk of bias. Twenty-one (27.3%) trials were funded by industry. A median 1.62% of patients (interquartile range, 0.00-3.70) crossed over between trial arms. Most trials reported a favorable outcome (n = 53 [58.9%]). For eligible trials, the median fragility index was 2.0 (interquartile range, 0.0-4.0), meaning the change of 2 patient outcomes would render the significant result insignificant. Spin, or distortion in reporting, was identified in 9 of 53 trials (17.0%). The median number of citations was 25 (10-56).
Contemporary trials in cardiothoracic surgery are pragmatic with low rates of loss to follow-up and crossover. Few trials implemented measures to ensure quality of the intervention, and the presence of spin was infrequent.
本分析旨在描述当代心胸外科随机对照试验（RCT）的特征。
我们选择了2008年至2020年间发表在医学、普通外科和心胸外科领域影响因子最高的期刊上的随机对照试验。总结并比较了试验特征以及报告和质量指标。
纳入93项试验；44项（47.3%）进行了前瞻性注册，14项（31.8%）注册的和发表的主要结局存在差异。大多数试验（n = 83 [89.1%]）采用优效性设计，采用复合主要结局（n = 82 [88.2%]），并将主要临床事件作为主要终点（n = 67 [72.0%]）。很少使用盲法，大多数试验未控制外科医生经验（n = 74 [79.5%]）或监测干预措施（n = 90 [96.7%]）。24项（25.8%）试验存在高偏倚风险。21项（27.3%）试验由行业资助。试验组间患者交叉比例中位数为1.62%（四分位间距，0.00 - 3.70）。大多数试验报告了有利结局（n = 53 [58.9%]）。对于符合条件的试验，脆弱性指数中位数为2.0（四分位间距，0.0 - 4.0），这意味着2例患者结局的改变会使显著结果变得不显著。在53项试验中的9项（17.0%）中发现了报告偏倚或歪曲。引用次数中位数为25（10 - 56）。
当代心胸外科试验注重实用性，随访失访率和交叉率较低。很少有试验采取措施确保干预质量，报告偏倚或歪曲情况不常见。
