Global Biostatistical Sciences, Amgen Inc., South San Francisco, CA, USA.
Stat Med. 2013 Mar 30;32(7):1112-24. doi: 10.1002/sim.5700. Epub 2012 Dec 14.
We propose a group sequential Holm procedure when there are multiple primary endpoints. This method addresses multiplicities arising from multiple primary endpoints and from multiple analyses in a group sequential design. It has been shown to be a closed testing procedure and preserves the familywise error rate in the strong sense. When multiple endpoints are the only concern without an interim analysis, the method simplifies to the weighted Holm procedure. The proposed method is more powerful than the parallel group sequential method and avoids the need to anticipate the testing order as in the fixed sequence testing scheme.
我们提出了一种当存在多个主要终点时的分组序贯 Holm 程序。这种方法解决了由于多个主要终点和分组序贯设计中的多个分析而产生的多重性问题。已经证明它是一种封闭的检验程序,并且在强意义上保持了总体错误率。当多个终点是唯一关注点且没有中间分析时,该方法简化为加权 Holm 程序。与平行组序贯方法相比,所提出的方法更有效,并且避免了像固定序列检验方案中那样需要预测检验顺序。