Evaluation and improvement of the Gamma Poisson Shrinkage for detecting adverse events following immunisation

Spontaneous reporting databases are important data sources for monitoring vaccine safety using passive surveillance. Various methods for signal detection in drug spontaneous reporting databases are well-established, but their performance in detecting adverse events following immunisation is not well-understood. This thesis discussed some vaccine-related features and evaluated their effects on the performance of a Bayesian safety signal detection method for detecting adverse events following immunisation from spontaneous reporting databases. We quantified how accurate and timely detection of adverse events following immunisation depends on the reference data set, seasonal effects, the signal generation criteria, and the method of data accumulation.