Statistical inference for extended or shortened phase II studies based on Simon's two-stage designs.
Statistical inference for extended or shortened phase II studies based on Simon's two-stage designs.
复制标题
DOI：
10.1186/s12874-015-0039-5
复制
发表时间：
2015-06-07
影响因子：
4
通讯作者：
Feng XP
中科院分区：
文献类型：
--
作者：
Zhao J;Yu M;Feng XP
Simon的两阶段设计是进行II期临床试验的热门选择，特别是在肿瘤学试验中，以减少无效实验治疗的患者数量。最近，Koyama和Chen（2008）讨论了如何为此类研究进行适当的推断，因为他们发现Simon设计中使用的推断程序几乎总是忽略实际使用的抽样计划。特别是，他们提出了一种推理方法，用于实际第二阶段样本量与计划样本量不同时的研究。我们考虑了一种基于似然比的替代推理方法。特别是，我们为了允许的样本路径下西蒙的两阶段设计使用相应的条件似然。通过这种方式，我们可以使用常见的定义来计算p值：获得至少与零假设下观察到的一样极端的检验统计量值的概率。除了为Koyama和Chen的方法难以应用的几种情况提供推断之外，基于我们的方法的估计在许多数值模拟中的推断属性方面似乎具有一定的优势。它通常导致较小的偏差和较窄的置信区间，同时保持相似的覆盖率。我们还说明了这两种方法在一个真实的数据设置。西蒙设计中使用的推理程序几乎总是忽略实际的抽样计划。报告的响应率的P值、点估计值和置信区间通常不针对设计的适应性进行调整。应使用适当的统计推断程序。
Simon’s two-stage designs are popular choices for conducting phase II clinical trials, especially in the oncology trials to reduce the number of patients placed on ineffective experimental therapies. Recently Koyama and Chen (2008) discussed how to conduct proper inference for such studies because they found that inference procedures used with Simon’s designs almost always ignore the actual sampling plan used. In particular, they proposed an inference method for studies when the actual second stage sample sizes differ from planned ones. We consider an alternative inference method based on likelihood ratio. In particular, we order permissible sample paths under Simon’s two-stage designs using their corresponding conditional likelihood. In this way, we can calculate p-values using the common definition: the probability of obtaining a test statistic value at least as extreme as that observed under the null hypothesis. In addition to providing inference for a couple of scenarios where Koyama and Chen’s method can be difficult to apply, the resulting estimate based on our method appears to have certain advantage in terms of inference properties in many numerical simulations. It generally led to smaller biases and narrower confidence intervals while maintaining similar coverages. We also illustrated the two methods in a real data setting. Inference procedures used with Simon’s designs almost always ignore the actual sampling plan. Reported P-values, point estimates and confidence intervals for the response rate are not usually adjusted for the design’s adaptiveness. Proper statistical inference procedures should be used.