Each year drugs are withdrawn due to adverse drug reactions with a mean time of approximately 6 years from approval to withdrawal. Therefore, if we could detect signals earlier, multiple patients could potentially avoid being exposed contributing to increased patient safety. My PhD project aims to develop and test an active surveillance system allowing both signal detection and rapid evaluation as an answer to the limitations of our current pharmacovigilance. The project focuses on antidepressant safety using both traditional epidemiological methods and data mining in healthcare register data.