DESIGNS FOR GROUP SEQUENTIAL-TESTS
DESIGNS FOR GROUP SEQUENTIAL-TESTS
复制标题
DOI：
10.1016/s0197-2456(84)80014-8
复制
发表时间：
1984-01-01
期刊：
影响因子：
--
通讯作者：
OBRIEN, PC
中科院分区：
文献类型：
--
作者：
FLEMING, TR;HARRINGTON, DP;OBRIEN, PC
此前，提出分组序贯程序是为了满足临床试验中中期分析的伦理需求。提出了一种替代程序，该程序为在初始结果极端时提前终止提供了良好的机会，同时基本上保持了仅在最终分析的预定时间计算相应检验统计量时所应用的程序所提供的能力。新的设计也足够灵活，可以轻松地允许由于某些原因（例如研究的累积速度意外缓慢）而增加最大分析数量，尽管不是出于考虑中期分析中观察到的结果而产生的原因。 [研究人员经常进行前瞻性临床试验，以比较两种治疗方法的疗效]。
Previously, group sequential procedures were proposed to satisfy the ethical need in clinical trials for interim analyses. An alternative procedure is proposed that offers a good opportunity for early termination when initial results are extreme, while essentially maintaining the power provided by the procedure that applied when the corresponding test statistic is computed only at the predetermined time of final analysis. The new design is also suffficiently flexible to readily allow the increase of the maximum number of analyses for certain reasons such as unexpectedly slow accrual into the study, although not for reasons arising from consideration of outcomes observed at interim analysis. [Prospective clinical trials frequently are carried out by investigators seeking to compare the therapeutic efficacy of 2 treatments.].