Framework for Epidemiology

Bradford Hill's Criteria for inferring causation 1965.

Mnemonic: 'ACCESS PTB'

1. Strength of association (eg RR) - does removal of factor decrease/ prevent presumed outcome? (most powerful criteria for showing causality).
2. Consistency of association - repeated demonstration in different populations
3. Specificity of association
4. Temporal sequence of association (effect after exposure)
5. Biological gradient - dose effect?
6. Plausibility of association - biomechanical
7. Coherence - absence of conflict with other knowledge
8. Experimental evidence
9. Analogy

Plus recent additions to criteria:

- generating and testing hypotheses
- actively searching for evidence to disprove theory