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

Chapter 32

Flexible Designs

Martin Posch, Peter Bauer, and Werner Brannath

32.1 Introduction

Classical frequentist statistical inference is based on the assumption that the inferential procedure is completely specified in advance. Consequently, the consensus guideline International Conference on Harmonization (ICH) E9 on Statistical Principles for Clinical Trials [1] requires for confirmatory trials that hypotheses and the statistical analysis plan is laid down in advance. For the planning of optimal trial designs knowledge on quantities as the expected efficacy of a new treatment, the safety properties, the appropriate doses or application forms of a treatment, the success rate in the control group, and the variability of outcome parameters are essential. Typically, in the planning phase of a clinical trial, many of these quantities are unknown. However, relevant information may aggregate in the course of the trial. Based on this information, changes in the trial design can become desirable or even inevitable. The necessity to allow for cautious adaptations is also recognized in the ICH E9 document:

If it becomes necessary to make changes to the trial, any consequent changes to the statistical procedures should be specified in an amendment to the protocol at the earliest opportunity, especially discussing the impact on any analysis and inferences that such changes may cause. The procedure selected should always ensure that the overall probability of type I error is controlled ...