【作者】 蒋志伟； 于莉莉； 李婵娟； 夏结来；

【Author】 Zhiwei Jiang~1,Lili Yu~2,Chanjuan Li~1,Jielai Xia~1 1Department of Health Statistics,Fourth Military Medical University, 2Department of Information,Jinan General Hospital

【机构】 第四军医大学卫生统计学教研室； 济南军区总医院信息科；

【摘要】 与传统试验设计方法相比,成组序贯设计可以在试验进行过程中进行多次期中分析以提前得出结论结束试验,具有更强的灵活性。生存数据由于删失数据的存在和数据分布特征不明,其在期中分析时间点的选择和样本量估计方面具有自身的特点。本文从某肿瘤药物临床试验为例,以试验中发生终点事件累积病例数占预计发生总终点事件数的比例来计算信息时间以确定试验的期中分析时间点,并采用基于Monte Carlo模拟的方法,综合同时考虑组间样本比例、脱落率、删失病例等生存数据成组序贯设计中多种情况进行样本量估计,基本解决了生存数据成组序贯设计中的主要问题。更多还原

【Abstract】 When compared to the traditional trial design,the group sequential design,which includes multiple interim analysis and considers early stopping for efficacy,is more flexible.Due to the censored data and unclear distribution,the group sequential design of survival data,including the choosing of time point for interim analysis,sample size estimation,etc,has its own features.In this paper,taken a cancer drug clinical trial as example,it is advised that the information time for interim analysis is evaluated by using the cumulative events occurred and the sample size is estimated by Monte Carlo simulation,which includes the dropout,sample size allocation ratio,censored data,etc.And the major problems in group sequential design of survival data can be solved.更多还原