Methods and Applications of Statistics in Clinical Trials, Volume 1: Concepts, Principles, Trials, and Designs

Chapter 5

Alpha-Spending Function

David L. DeMets and K. K. Gordon Lan

5.1 Introduction

The randomized control clinical trial (RCT) is the standard method for the definitive evaluation of the benefits and risks of drugs, biologies, devices, procedures, diagnostic tests, and any intervention strategy. Good statistical principles are critical in the design and analysis of these RCTs [1, 2]. RCTs also depend on interim analysis of accumulating data to monitor for early evidence of benefit, harm, or futility. This interim analysis principle was established early in the history of RCTs [3] and was implemented in early trials such as the Coronary Drug Project [4, 5]. Evaluation of the interim analysis may require the advice of an independent data monitoring committee (DMC) [6, 7], including certain trials under regulatory review [8, 9]. However, although ethically and scientifically compelling, interim repeated analysis of accumulating data has the statistical consequence of increased false positive claims unless special steps are taken.

The issue of sequential analysis has a long tradition [10, 11] and has received special attention for clinical trials [12, 13]. In particular, increasing the frequency of interim analysis can substantially increase the Type I error if the same criteria are used for each interim analysis [13]. This increase was demonstrated in the Coronary Drug Project, which used sequential analysis for monitoring several treatment arms compared with a placebo [4]. Most ...