Past, present, and future of Phase 3 vaccine trial design: rethinking statistics for the 21st century

Vaccines are crucial for protecting health globally; however, their widespread use relies on rigorous clinical development programmes. This includes Phase 3 randomized controlled trials (RCTs) to confirm their safety, immunogenicity, and efficacy. Traditionally, such trials used fixed designs with predetermined assumptions, lacking the flexibility to change during the trial or stop early due to overwhelming evidence of either efficacy or futility. Modern vaccine trials benefit from innovative approaches like adaptive designs, allowing for planned trial adaptations based on accumulating data. Here, we provide an overview of the evolution of Phase 3 vaccine trial design and statistical analysis methods from traditional to more innovative contemporary methods. This includes adaptive trial designs, which offer ethical advantages and enable early termination if indicated; Bayesian methods, which combine prior knowledge and observed trial data to increase efficiency and enhance result interpretation; modern statistical analysis methods, which enable more accurate and precise inferences; the estimand framework, which ensures the primary question of interest is addressed in a trial; novel approaches using machine learning methods to assess heterogeneity of treatment effects; and statistical advances in safety analysis to evaluate reactogenicity and clinical adverse events. We conclude with insights into the future direction of vaccine trials, aiming to inform clinicians and researchers about conventional and novel RCT design and analysis approaches to facilitate the conduct of efficient, timely trials.