16 Sequential and Group Sequential Designs in Clinical Trials: Guidelines for Practitioners

In a classical fixed sample design, the sample size is set in advance of collecting any data. The main design focus is choosing the sample size that allows the clinical trial to discriminate between the null hypothesis of no difference and the alternative hypothesis of a specified difference of scientific interest. A disadvantage of fixed sample design is that the same number of subjects will always be used regardless of whether the true treatment effect is extremely beneficial, marginal, or truly harmful relative to the control arm. Often, it is difficult to justify because of ethical concerns and/or economic reasons. Thus, specific early termination procedures have been developed to allow repeated statistical analyses to be performed on accumulating data and to stop the trial as soon as the information is sufficient to conclude. However, repeated analyses inflate the false positive error to an unacceptable level. To avoid this problem, many approaches of group sequential methods have been developed. Although there is an increase in the planned sample size under these designs, due to the sequential nature, substantial sample size reductions compared with the single-stage design is also possible not only in the case of clear efficacy but also in the case of complete lack of efficacy of the new treatment. This feature provides an advantage in utilization of patient resource. These approaches are methodologically complex but advancement in software packages had made the planning, monitoring, and analysis of comparative clinical trials according to these approaches quite simple. Despite this simplicity, the carrying on of a trial under group sequential design requires efficient logistics with dedicated team of data manager, study coordinator, biostatistician, and clinician. Good collaboration, rigorous monitoring, and guidance offered by an independent data safety monitoring committee are all indispensable pieces for its successful implementation. In this chapter, we provide a review of sequential designs and discuss the underlying premise of all current methods. We present a recent example and an historical example to illustrate the methods discussed and to provide a flavor of the variety and complexity in decision making. A comprehensive list of softwares is provided for easy implementation along with practical guidelines. Few areas with potential for future research are also identified.