Structured approach to the elicitation of expert beliefs for a Bayesian-designed clinical trial: a case study.

To quantify uncertainty in a formal manner, statisticians play a vital role in identifying a prior distribution for a Bayesian-designed clinical trial. However, when expert beliefs are to be used to form the prior, the literature is sparse on how feasible and how reliable it is to elicit beliefs from experts. For late-stage clinical trials, high importance is placed on reliability; however, feasibility may be equally important in early-stage trials. This article describes a case study to assess how feasible it is to conduct an elicitation session in a structured manner and to form a probability distribution that would be used in a hypothetical early-stage trial. The case study revealed that by using a structured approach to planning, training and conduct, it is feasible to elicit expert beliefs and form a probability distribution in a timely manner. We argue that by further increasing the published accounts of elicitation of expert beliefs in drug development, there will be increased confidence in the feasibility of conducting elicitation sessions. Furthermore, this will lead to wider dissemination of the pertinent issues on how to quantify uncertainty to both practicing statisticians and others involved with designing trials in a Bayesian manner.