Laursen David Ruben Teindl, Paludan-Müller Asger Sand, Hróbjartsson Asbjørn
Centre for Evidence-Based Medicine Odense (CEBMO), Odense University Hospital, Odense, Denmark,
Nordic Cochrane Centre, Rigshospitalet, Copenhagen, Denmark,
Clin Epidemiol. 2019 Feb 11;11:169-184. doi: 10.2147/CLEP.S188752. eCollection 2019.
Run-in periods are occasionally used in randomized clinical trials to exclude patients after inclusion, but before randomization. In theory, run-in periods increase the probability of detecting a potential treatment effect, at the cost of possibly affecting external and internal validity. Adequate reporting of exclusions during the run-in period is a prerequisite for judging the risk of compromised validity. Our study aims were to assess the proportion of randomized clinical trials with run-in periods, to characterize such trials and the types of run-in periods and to assess their reporting.
This was an observational study of 470 PubMed-indexed randomized controlled trial publications from 2014. We compared trials with and without run-in periods, described the types of run-in periods and evaluated the completeness of their reporting by noting whether publications stated the number of excluded patients, reasons for exclusion and baseline characteristics of the excluded patients.
Twenty-five trials reported a run-in period (5%). These were larger than other trials (median number of randomized patients 217 vs 90, =0.01) and more commonly industry trials (11% vs 3%, <0.01). The run-in procedures varied in design and purpose. In 23 out of 25 trials (88%), the run-in period was incompletely reported, mostly due to missing baseline characteristics.
Approximately 1 in 20 trials used run-in periods, though much more frequently in industry trials. Reporting of the run-in period was often incomplete, precluding a meaningful assessment of the impact of the run-in period on the validity of trial results. We suggest that current trials with run-in periods are interpreted with caution and that updates of reporting guidelines for randomized trials address the issue.
导入期偶尔用于随机临床试验，以在纳入患者后、随机分组前排除患者。理论上，导入期增加了检测潜在治疗效果的可能性，但可能会影响外部和内部效度。充分报告导入期的排除情况是判断效度受损风险的前提条件。我们的研究目的是评估有导入期的随机临床试验的比例，描述此类试验及其导入期类型，并评估其报告情况。
这是一项对2014年470篇被PubMed收录的随机对照试验出版物的观察性研究。我们比较了有和没有导入期的试验，描述了导入期的类型，并通过记录出版物是否说明了排除患者的数量、排除原因以及被排除患者的基线特征来评估其报告的完整性。
25项试验报告了导入期（5%）。这些试验比其他试验规模更大（随机分组患者的中位数为217例对90例，P=0.01），且更常见于行业试验（11%对3%，P<0.01）。导入程序在设计和目的上各不相同。25项试验中有23项（88%）对导入期的报告不完整，主要是由于缺少基线特征。
约每20项试验中有1项使用导入期，不过在行业试验中更为常见。导入期的报告往往不完整，无法对导入期对试验结果效度的影响进行有意义的评估。我们建议谨慎解读当前有导入期的试验，并建议随机试验报告指南的更新应解决这一问题。
