ABSTRACT
Analyses in drug safety studies involve simple to rather complex statistical methods. Often they involve categorical outcomes, such as whether a certain adverse event occurred or not. The risk factors of interest are also often categorical, such as whether a patient was on a certain medication or not. Therefore, knowledge regarding analysis of 2 × 2 contingency tables is fundamental. Analyses involving multiple risk factors require stratification or covariate adjustment. These studies sometimes involve millions of data points and hundreds of covariates. Bayesian methods are particularly useful in modeling such high dimensional data. In this chapter, we discuss some measures of effect, basic modeling techniques and parameter estimation methods in both frequentist and Bayesian domains. It is assumed that readers are familiar with basic statistical concepts including measures of central tendency, measures of dispersion and sampling variability.
