ABSTRACT
This chapter discusses the functionality found in Bayesian analysis to incorporate historical information, estimate many parameters in a multi-level model, and update information as it accrues to face these stated challenges that are applicable to safety statistics. Bayesian statistics represents a philosophy of statistical analysis which differs from the more traditional frequentist statistical philosophy in the fundamental notion of what probability. However, there is a risk that incorrect decisions may still occur if they are made based on interim information. Bayesian analysis does provide natural protection against this kind of multiplicity, for example, through the use of a skeptical prior. Estimating many parameters within the same statistical model is a challenge. A key assumption that is often made within a level of hierarchical models is exchangeability. Ongoing safety monitoring of clinical trials is an essential aspect of sponsors’ responsibility to patient safety.
