Speaker: Dr. Liz Ryan, University of Warwick

Title: “Evaluation of Bayesian adaptive designs for Phase III effectiveness trials”

Abstract: Phase III randomised controlled trials (RCTs) are typically long and expensive, and often fail to achieve the trials’ objective of finding a statistically significant difference between treatments. Researchers and funders have recognised the need for trials to become more efficient, yet the overwhelming majority of trials continue to use traditional methods, particularly with fixed designs.

Bayesian adaptive trial methods are increasingly being used, and have the potential to allow trials to answer their questions more efficiently, often meaning that effectiveness can be determined with fewer patients and in a shorter time. These approaches use Bayesian statistical methods, and include interim analyses, to allow learning from the trial data collected so far, and subsequent modification of the trial. Whilst Bayesian adaptive designs have proved to be popular for early phase studies, their use in phase III trials remains limited.

We aim to demonstrate that implementation of adaptive designs for phase III trials in the Bayesian framework can be achieved fairly easily and using a detailed example, will demonstrate the process of how a Bayesian adaptive design for a phase III trial can be constructed, and the decisions that need to be made during the design phase. We will also show how these approaches can improve the efficiency of trials, and also provide results that can easily be interpreted. This will be achieved by re-designing and performing virtual re-executions of a large, multicentre respiratory care RCT that was designed using traditional methods.

We begin by proposing a candidate set of alternative Bayesian adaptive designs for the trial, and simulate and compare the operating characteristics of each design. A single preferred design will then by chosen by a group of clinicians. We will then re-run the trial using the patient-level data and the Bayesian adaptive designs to determine whether the use of a Bayesian adaptive design would have had advantages in terms of duration, size and cost of the trials. Through this work we hope to encourage clinicians to implement Bayesian adaptive trial designs for phase III trials, which have rarely been used in practice, and to illustrate the merits of these designs.