On the efficiency of adaptive designs for flexible interim decisions in clinical trials
On the efficiency of adaptive designs for flexible interim decisions in clinical trials
复制标题
DOI：
10.1016/j.jspi.2005.08.014
复制
发表时间：
2006-06-01
影响因子：
0.9
通讯作者：
Posch, M
中科院分区：
文献类型：
--
作者：
Brannath, W;Bauer, P;Posch, M
结果表明，Tsiatis和Mehta [2003.关于临床试验监测的适应性设计的无效性。Biometrika 90，367-378]是更一般的灵活设计的特殊情况，其允许在调整预定方式以花费拒绝和接受概率之后进行有效的推断。例如，临床试验中出现不可预见的安全性问题，可能会改变预先计划的中期分析数量及其样本量。我们推导出灵活的设计，具有等效的拒绝和接受区域，如果没有进行适应，但在同一时间允许适应的支出函数，并有条件的最优性。(c)2005 Elsevier B. V.保留所有权利。
It is shown that the optimal group sequential designs considered in Tsiatis and Mehta [2003. On the inefficiency of the adaptive design for monitoring clinical trials. Biometrika 90, 367-378] are special cases of the more general flexible designs which allow for a valid inference after adapting a predetermined way to spend the rejection and acceptance probabilities. An unforeseen safety issue in a clinical trial, for example, could make a change of the preplanned number of interim analyses and their sample sizes appropriate. We derive flexible designs which have equivalent rejection and acceptance regions if no adaptation is performed, but at the same time allow for an adaptation of the spending functions, and have a conditional optimality property. (c) 2005 Elsevier B.V. All rights reserved.