Sequential monitoring of clinical trials: The role of information and brownian motion

Sequential monitoring has been a topic of major interest in clinical trials methodology over the past two decades. This paper presents a unified conceptual framework for sequential monitoring that covers a wide variety of monitoring procedures in a wide variety of clinical trial settings. The central elements of this framework consist of a suitable concept of statistical information and a scheme for using this concept as a basis for summarizing the accumulating results of a trial in a standardized form, through a stochastic process that can be shown to approximate classical Brownian motion. The ideas are developed in a simple step-by-step fashion and illustrated by several practical examples.