A simulation study used to design the sequential monitoring plan for a clinical trial

In designing a sequential monitoring plan for the AVID (Antiarrhythmics Versus Implantable Defibrillators) study, a multi-centred clinical trial, we completed a simulation study to help ourselves and the Data Safety Monitoring Board understand the implications of various monitoring plans. In particular, we compared three test statistics and three use functions. Simulation was based on approximate trial parameters and two plausible but distinct alternative hypotheses. The power to detect a difference between the two treatment arms was computed as a function of time. The simulated results illuminate several of the important statistical and ethical issues involved in selecting a sequential monitoring plan and illustrate how one can use simulations to design a clinical trial.