The MRC superficial bladder cancer trial of intravesical mytomicin-c after complete surgical resection. Sequential statistical methods applied to survival data from a randomised clinical trial

Reduction in the duration of a study and in the number of patients required can be obtained when we adopt a sequential design. In this paper we re-analyse a trial completed by the British Medical Research Council on the effects of chemotherapy to prevent the recurrence of surgically removed superficial bladder cancer as if it had been monitored sequentially. The aim is to illustrate the use and benefits of sequential designs (stopping rules) and to highlight how to handle some potential problems when the assumptions of the statistical model are not satisfied. These problems are not exclusive to the sequential design, but are also present when a conventional design is used.