Bayesian methods in non-clinical pharmaceutical statistics.

This dissertation is composed of three research papers investigating the application of Bayesian methods to pharmaceutical non-clinical statistics. In the first paper, we present an application of Bayesian assurance and sample size determination to the manufacturing process validation life-cycle. In particular, we show how the two-prior approach for sample size determination, which has been considered extensively in the clinical biostatistics area, can be applied to determine the number of commercial scale batches that are needed for Stage II of the process validation life-cycle. In the second paper, we propose a Bayesian methodology for estimation of resistance parameters used in the validation of steam sterilization processes for aseptic manufacturing of drug products. Since the assessment of the sterility of manufacturing equipment and product is fundamentally an assessment of probability, we show that a Bayesian approach permits validation conclusions to be made based on the quantification of the risk of non-sterility. Finally, in the third paper, we show how accelerated drug stability results can be used to inform long-term stability studies in the area of drug shelf-life determination using Bayesian methods.