Adaptive Designs for Clinical Trials

Sunday morning, July, 8

Werner Brannath, Medical University of Vienna
Frank Bretz, Novartis Pharma AG

The aim of this short course is to give an introduction to the principles and methodologies of adaptive designs for clinical trials. Adaptive designs allow for mid-course design adaptations that are based on un-blinded interim data as well as on all information from in- or outside the trial without compromising the overall type I error rate. Design adaptations could be adjustments of sample sizes and/or the number of interim analyses that are based on interim estimates of the variance, the treatment effect and safety parameters. Adaptive designs are particularly useful in trials with multiple endpoints or with several experimental treatments/doses. In this case one can also change or select endpoints and drop or add treatments/doses at interim analyses. This is of particular interest for drug development programs to combine different phases (e.g. II and III) into a single trial. Strategies for combining clinical phases may lead to substantial savings in cost and time. Utilizing the adaptive design methodology and the closed testing principle one can build adaptive designs that control the multiple type I error rate. In order to control the multiple level one need not prefix or fully anticipate the adaptations in advance. This allows to react to unforeseen events and to implement complex decisions that are based on convoluted internal and external information.

The course will provide an overview of methods from the published literature of the last two decades including the most recent developments. The course will discuss sample size adjustments and to a large extend multiple hypotheses testing with adaptive designs. Estimation following adaptive designs will also be a topic. Software implementations and regulatory issues will be discussed as well. All methods will be illustrated by examples or case studies.

  1. Introduction to adaptive designs: Motivation and methods
  2. Multiple hypotheses testing in adaptive designs: Methods and case studies
  3. Estimation in adaptive designs
  4. Software implementations
Bio sketch:
Werner Brannath, is Associate Professor in Biostatistics at the Medical University of Vienna. He has received his PhD in mathematics in 1997 and is working at the Department of Medical Statistics since August 1998. He has written several papers on multiple testing and adaptive designs. One of his research focuses is estimation in adaptive designs.

Frank Bretz, Ph.D., is Expert Statistical Methodologist, Novartis Pharma AG. He received his PhD in biostatistics from the University of Hannover in 1999 and continued working there from 2000 until 2004, when he joined Novartis Pharma AG. He has been working on multiplicity issues arising in clinical trials, dose finding trials and adaptive designs since 1997 and written several papers/chapters on these topics.