Schäfer Helmut, Müller Hans-Helge
Institute of Medical Biometry and Epidemiology, Philipps-University Marburg, Bunsenstrabetae 3, D-35037 Marburg, Germany.
Stat Med. 2004 May 15;23(9):1413-24. doi: 10.1002/sim.1751.
The treatment effect sizes that can be detected with sufficient power up to the different interim analyses constitute a clinically meaningful criterion for the selection of a group sequential test for a clinical trial. For any pre-specified sequence of effect sizes, it is possible to construct group sequential boundaries such that the trial has a pre-specified power to reject the null-hypothesis at or before the corresponding interim analysis under the respective treatment effect. The principle of constructing group sequential designs on the basis of detectable treatment effects is presented. The application in common situations such as two-armed trials with continuous or binary outcome or censored survival times is described. We also present an effective algorithm.
在不同的期中分析中,能够以足够的检验效能检测到的治疗效应大小构成了选择临床试验组序贯检验的一个具有临床意义的标准。对于任何预先指定的效应大小序列,都可以构建组序贯界值,使得在相应的治疗效应下,试验在相应的期中分析时或之前有预先指定的检验效能来拒绝原假设。本文介绍了基于可检测的治疗效应构建组序贯设计的原理。描述了其在常见情况下的应用,如具有连续或二元结局或删失生存时间的双臂试验。我们还提出了一种有效的算法。