Evaluating Bayesian hierarchical models and decision criteria for the detection of adverse events in vaccine clinical trials

This thesis examined the accuracy, precision, and decision-making potential of Bayesian hierarchical methods in detecting vaccine safety concerns within clinical trials in comparison to commonly used frequentist methods within a series of simulation studies. Bayesian hierarchical modelling was found to have advantages over frequentist methods in terms of greater precision of the effect measure estimates and greater power in detecting a potential safety signal in early interims. The implementation of Bayesian hierarchical methods to assess safety in adaptive clinical trials may be beneficial in terms of earlier identification and subsequent reduction in participants exposed to potential safety concerns.