Implications of using different types of priors in a three-level hierarchical Bayesian Model for the analysis of Drug Adverse Events.