ABSTRACT
When considering a clearly statistically significant association found between a disease and an exposure, interest will often focus on whether the relationship is causal. While this may be a relatively straightforward issue for a well conducted randomized controlled trial, it is much more challenging when considering results from an observational study. Even after adjusting for all potential confounding factors, several important issues need to be considered. In a now classic address to the Royal Society of Medicine in 1965, Sir Austin Bradford Hill proposed that several aspects of an association should be given special consideration. These are summarized below:1
1. Strength of association: large relative risk or correlation coefficient, with small standard error 2. Consistency: the results are replicable or in accordance with those of other, similar studies 3. Specificity: a single exposure or risk factor has a specific effect 4. Temporality: the risk factor must precede disease 5. Dose-response: increased exposure to the risk factor increases the risk of disease 6. Plausibility: consistent with current understanding of biological or pathological processes 7. Coherence: agrees with existing knowledge and widely accepted theories 8. Experiment: the relationship can be modified by experimental interference 9. Analogy: confounding factors have been considered and excluded.
