Causal association in pharmacovigilance and pharmacoepidemiology: thoughts on the application of the Austin Bradford-Hill criteria.

The methods used for the evaluation of drug safety signals (including major signals leading to withdrawal of products from the market) are inconsistent and sometimes of poor quality. While the assessment of the safety of medicines needs to consider specific issues such as drug interactions and variation in compliance, the general principles, which are used to study environmental hazards, can be applied for this purpose. The criteria proposed by Sir Austin Bradford-Hill more than 35 years ago for attributing disease causation to environmental factors have been used widely in epidemiology, are applicable to pharmacovigilance and pharmacoepidemiology. The Austin Bradford-Hill criteria include strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experimental evidence and analogy. The paper reviews each of these criteria with emphasis on pharmacovigilance and pharmacoepidemiology and with some examples. The application of the Austin Bradford-Hill criteria to the evaluation of causal association in pharmacovigilance and pharmacoepidemiology is very useful. However, it requires understanding of the limitations of the data, such as, under-reporting, poor quality of information from third parties and misclassification. Further work is required to develop strategies to handle these limitations.