Sequential Analysis

During the past decade, various sequential design methods in clinical trial designs have been developed to allow for interim decisions and trial modifications based on cumulated information. As part of well-known adaptive designs in group sequential analysis, Yu Shen and Lloyd Fisher proposed a sequential strategy for monitoring clinical trials, namely self-designing trials. The final test statistic is a weighted average of the sufficient statistic from each stage, where the weight for each stage is determined by the observed data prior to that stage. One important feature of this self-designing trial is that the maximum sample size or the maximum number [Page 1342]of interim analyses, m, is not specified in advance but at a random stopping time. The flexibility of modifying the sample size allows ...