REPORTING BAYESIAN ANALYSES OF CLINICAL-TRIALS

Many clinicians wrongly interpret p-values as probabilities. that trea tment has an adverse effect and confidence intervals as probability in tervals. Such inferences can be validly drawn from Bayesian analyses o f trial results. These analyses use the data to update the prior (or p re-trial) beliefs to give posterior (or post-trial) beliefs about the magnitude of a treatment effect. However, for these methods to pin acc eptance in the medical literature, understanding between statisticians and clinicians of the issues involved in choosing appropriate prior d istributions for trial reporting needs to be reached. I focus on two t ypes of prior that deserve consideration. The first is the non-informa tive prior giving standardized likelihood distributions as post-trial probability distributions. Their use is unlikely to be controversial a mong statisticians whilst being intuitively appealing to clinicians. T he second type of prior has a spike of probability mass at the point o f no treatment effect. Varying the magnitude of the spike illustrates the sensitivity of the conclusions drawn to the degree of prior scepti cism in a treatment effect. With both, graphical displays provide clin ical readers with the opportunity to explore the results more fully. A n example of how a clinical trial might be reported in the medical lit erature using these methods is given.