*Geek Box: Bradford-Hill Criteria
In 1965, Sir Austin Bradford Hill described his criteria for inferring causation, a framework that encapsulates the concept of considering converging lines of evidence to form conclusions. At the time, the tobacco industry was throwing up smoke and mirrors against the relationship between smoking and cancer, and arguing there was no “proof” of causality. Bradford-Hill published his seminal paper on evaluating causal relationships from observational research. However, in the 1970’s, the term “meta-analysis” was coined, and this approach to synthesising research quickly became a new dogma [more on this in the next Geek Box], superseding the scientific process of considering multiple lines of evidence, and thoroughly thinking through evidence from multiple angles.
The Bradford-Hill criteria includes:
- Strength of the association
- Consistency
- Specificity
- Temporality
- Dose-response
- Plausibility
- Coherence
- Reversibility [i.e., does reducing the exposure influence frequency of disease]
- Experimental evidence
It gives us a framework to consider evidence when we don’t have The-Unicorn-RCT-That-Tells-Us-Everything-We-Want-To-Know. And when we consider evidence through this criteria, we are forced to engaged our ‘Big Picture’ thinking about be critical. Interestingly, because of the limitations of tools like meta-analysis for nutrition science, researchers appear to be turning back to using the Bradford-Hill criteria for assessing the evidence for different research questions. The criteria remains as relevant today as it was when Bradford-Hill published his seminal paper on the environment, disease, and causation.