Bradford Hill criteria for causation
- Concept
Nine viewpoints, proposed in 1965, for judging whether an association is causal. Only one is essential: the cause must come before the effect.
What it is
In 1965 Austin Bradford Hill proposed nine aspects for judging whether an observed association is probably causal: strength, consistency, specificity, temporality, biological gradient (dose-response), plausibility, coherence, experimental evidence and analogy. They are not a checklist to add up: they are questions that guide judgement after chance, bias and confounding have been ruled out. Temporality is the only one considered essential.
Worked example
- Made-up case: an industrial solvent and liver disease in workers. The evidence is gathered criterion by criterion.
- Strength: pooled RR 2.2 across five studies.
- Consistency: the five studies, in four countries, give RRs of 1.8 to 2.6.
- Temporality: exposure precedes diagnosis by a median of 8 years.
- Gradient: RR 1.5 with under 5 years of exposure, 2.1 with 5 to 10 years and 3.0 with over 10.
- Plausibility and coherence: the solvent damages liver cells in animals and fits what is known of other liver toxins.
- Experiment: after extractor fans are installed, cases at the factory fall from 12 to 4 per year.
- Reading: with temporality met and nearly every criterion in favour, a causal relationship is very likely.
The nine criteria
| Criterion | Question it asks |
|---|---|
| Strength | Is the association large (high RR or OR)? |
| Consistency | Is it repeated across studies, places and populations? |
| Specificity | Is the exposure linked to one particular outcome? |
| Temporality | Does the exposure come before the outcome? (essential) |
| Biological gradient | More exposure, more risk (dose-response)? |
| Plausibility | Is there a reasonable biological mechanism? |
| Coherence | Does it fit what is known of the disease's natural history? |
| Experimental evidence | Does removing or reducing the exposure lower the risk? |
| Analogy | Do similar exposures cause similar effects? |
In practice
- These criteria are used in public health, occupational health and pharmacovigilance to judge whether a drug causes an adverse event.
- A strong association with a dose-response gradient is harder to explain by confounding alone.
Common pitfalls
- Treating them as a checklist where adding up points is enough.
- Demanding specificity: one cause can produce many diseases.
- Dismissing a cause because no mechanism is known: plausibility depends on the knowledge of the day.
Test yourself
Think of the answer, then tap the question to see it.
Which Bradford Hill criterion is the only one considered essential?
Temporality: the exposure must come before the outcome.
What is the biological gradient?
A dose-response relationship: the greater the exposure (dose, duration, intensity), the greater the risk of the outcome.
Why is specificity a weak criterion?
Because many causes produce several diseases and many diseases have several causes; tobacco is the classic example.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Bradford Hill criteria | UMC (Uppsala Monitoring Centre, 2026); Chapter 14: Completing ‘Summary of findings’ tables and grading the certainty of the evidence (Cochrane, 2024).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.