The efficacy of a medicine is established by the time it becomes marketed. However, the possible harms associated with a drug are usually less well understood. Randomised controlled trials to assess side effects are often impractical as they may need to be unfeasibly large or could be unethical. Observational studies therefore offer an ideal practical method for investigating drug safety issues, and in recent years the number of such studies has increased. Much research in this area is done using anonymised patient information from large primary care databases such as the General Practice Research Database (GPRD). Over the last three years I have gained substantial experience in using the GPRD to answer questions about the safety of medicines. Recent advances in data quality and linkage offer even greater opportunities for research, but better methodologies are needed to take full advantage. I propose the following programme of research in order to improve approaches to using primary care data for drug safety studies: AIMS: Further develop strategies for using primary care data. In particular focusing on diagnostic validity, the impact of consultation frequency and distinguishing prevalent from incident diagnoses. Explore the utility and limitations of the self-controlled case series method. Specifically I will investigate the impact of time varying confounders, situations where the occurrence of an outcome of interest affects the length of the observation period and instances where the likelihood of exposure is dependent on the presence or absence of the outcome of interest. OBJECTIVES: To improve on existing methods and establish optimal methodologies for conducting drug safety studies. METHODOLOGY: I plan to achieve these aims by conducting four drug safety studies examining the associations between: SSRI antidepressants and cerebral haemorrhage; ezetimibe and cancer; tiotropium and stroke; and glitazones and myocardial infarction. I will use the self controlled case series method, comparing results with other study designs, such as case control and cohort methods and using external validatory data from randomised clinical trials. Collaborative input will be provided by the Medical Statistics Unit of LSHTM and the Department of Statistics at the Open University SCIENTIFIC/MEDICAL OPPORTUNITIES: The current regulatory environment demands more observational studies of drug effects. Patients and prescribers rightly expect investigators to produce reliable answers to questions about the safety of a drug and robust methodologies are therefore needed to ensure confidence in the outcome of drug safety studies.
