The conventional frequentist approach to statistical analysis of clinical trials evaluates study hypotheses indirectly by estimating the probability that a treatment effect the same or larger than the observed treatment effect would be obtained if the null hypothesis (which generally assumes that there is no treatment effect) was true. The goal of frequentist analysis is to determine whether the evidence leads one to confidently reject the null hypothesis. In Bayesian analysis, information available prior to the trial about the plausible range of values of the treatment effect (represented as a probability distribution) is updated by the data collected in the trial to produce a revised estimate of the plausible range of values of the treatment effect.
1 Bayesian analysis informs clinical decisions by directly estimating the probability of a hypothesized treatment effect given the observed data.
2,3 In addition, because information about treatment effect from preexisting clinical and biological evidence is formally incorporated into statistical evaluation, Bayesian methods explicitly quantify the otherwise implicit influence of clinical judgment and prior beliefs on the interpretation of trial results.
4-6