ABSTRACT
In 1984, the Cancer and Leukemia Group B (CALGB) opened a phase III clinical trial for patients with stage III non-small cell lung cancer (NSCLC). The original fixed sample size design called for randomly assigning 240 patients to either radiotherapy alone or chemotherapy followed by radiotherapy. Analysis of results was to occur after 190 deaths. Instead, early in the trial it was decided to apply group sequential concepts using a truncated O'Brien-Fleming stopping rule, implemented via aLanDeMets a-spending function. A study monitoring committee stopped the trial at the fifth interim analysis after only 155 eligible patients had been entered. This chapter presents a Bayesian alternative to the standard frequentist approach of assessing stopping times in clinical trials. It demonstrates how the use of posterior and predictive distributions provides a natural way to address many of the issues raised in monitoring clinical trials.
