Learning Objectives
- Detail the purpose of the Bradford Hill criteria as a framework for evaluating evidence of a causal relationship between an exposure and an outcome.
- Distinguish between all nine Hill criteria, including strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experiment, and analogy.
- Understand why these principles provide necessary but not strictly sufficient support for establishing absolute medical causality.
1. The Bradford Hill Criteria for Causality
The Bradford Hill criteria comprise a group of principles used in epidemiology to evaluate whether observed statistical associations reflect true causal relationships.
| Criterion | Definition & Core Concept | Clinical & Epidemiological Application |
|---|---|---|
| Strength | The magnitude of association between the presumed cause and effect. | A larger relative risk or odds ratio provides stronger evidence for causation. |
| Consistency | Repeated observations of the findings by different investigators in multiple distinct samples. | Results are reproducible across diverse populations, settings, and study designs. |
| Specificity | The degree to which the presumed cause leads to a single, specific effect. | A cause linked to a single specific disease provides stronger causal backing (though many causes have multiple effects). |
| Temporality | The requirement that the presumed cause must precede the effect by an expected amount of time. | The only absolute, mandatory criterion for establishing causality in epidemiology. |
| Biological Gradient | A dose-response relationship where greater exposure leads to a greater effect. | Increased pack-years of smoking leading to a higher incidence of lung cancer. |
| Plausibility | The existence of a conceivable biological mechanism by which the cause leads to the effect. | Supported by baseline physiological or biochemical knowledge of disease pathways. |
| Coherence | Compatibility with existing scientific and biological knowledge and consensus. | The causal hypothesis does not conflict with established natural history or laboratory data. |
| Experiment | Empirical evidence supporting the cause-and-effect relationship from controlled studies. | Evidence derived from animal models, in vitro experiments, or clinical trials. |
| Analogy | Similarities between the observed association and other established cause-and-effect relationships. | If one teratogenic drug causes birth defects via a specific mechanism, a similar drug acting similarly is suspected. |
Clinical Integration & Methodological Rules:
- The Temporal Rule: While most criteria provide supporting evidence rather than absolute rules, temporality is strictly mandatory. An exposure cannot cause a disease if it occurs after the disease has already developed.
- Association vs. Causation: Observational studies routinely establish statistical associations (via relative risk or odds ratio), but applying the Bradford Hill framework helps researchers bridge the gap toward establishing true biological causation.
- Not All Criteria Required: Except for temporality, satisfying all other criteria is not strictly required to infer causality; rather, they serve as a weighted checklist of supporting evidence.
Key Points & Memory Hooks:
- The Golden Rule: Temporality is non-negotiable—the cause must always happen before the effect.
- Dose-Response: Biological gradient means “more dose equals more disease” (dose-response relationship).
- Mechanism Check: Plausibility and coherence ask whether the association makes sense based on current medical science.