Missing data
Missing data
复制标题
DOI：
10.1080/17434470410019753
复制
发表时间：
2004-09
期刊：
影响因子：
--
通讯作者：
John L.P. Thompson;Gilberto Levy
中科院分区：
文献类型：
--
作者：
John L.P. Thompson;Gilberto Levy
本文提出了一个实际问题：我们如何避免随机对照试验（RCT）表明治疗对肌萎缩性侧向硬化症（ALS）患者具有重要意义和临床重要的好处，但结果的可信度是由于缺少数据而被质疑？这是一个真正的可能性，对于ALS调查人员来说，这是一个非常严重的问题。统计学家和其他人的责任推荐将其最小化的步骤。在本文中，我们强调了RCT中缺少数据的重要性，并讨论了如何通过插补程序以公正的方式处理问题。然后，我们总结了这些插补程序的局限性，并以针对ALS的RCT量身定制的一些试验设计和行为建议得出结论。
This paper raises a practical concern: how can we avoid a situation in which a randomized controlled trial (RCT) shows that a therapy has a statistically significant and clinically important benefit for amyotrophic lateral sclerosis (ALS) patients, but the credibility of the result is called into question because of missing data? This is a real possibility, and potentially a very serious problem for ALS investigators. It is the responsibility of statisticians and others to recommend steps to minimize it. In this paper we emphasize the importance of missing data in RCTs, and discuss how the problem can be handled in an unbiased way by imputation procedures. We then summarize the limitations of these imputation procedures, and conclude with some recommendations for trial design and conduct that are tailored to RCTs for ALS.