Bradford Hill Criteria for Causality

Introduction

• Lecture focus: Understanding the Bradford Hill Criteria for causality.
• Objectives:
  • Describe the nine Bradford Hill criteria for causality.
  • Provide examples for each criterion.
  • List modern models of causality.
• Developed by: Sir Austin Bradford Hill, 1965.
• Purpose: To assess whether cigarettes cause diseases like lung cancer.
• Key point: Satisfying these criteria supports causality but failing some does not negate it.
• Hill's view: The criteria are guidelines, not hard rules.

The Nine Bradford Hill Criteria

1. Strength of Association

   • Measured via risk ratios, rate ratios, or odds ratios.
   • Strong associations indicate a higher likelihood of causality.
   • Example:
     • Smokers are 15-30 times more likely to develop lung cancer compared to non-smokers.
     • Weak associations exist (e.g., smoking and heart disease).

2. Consistency

   • Reproducibility of results across different populations and studies.
   • Consistency helps rule out alternative explanations.
   • Example:
     • Not all heavy smokers develop lung cancer; conditions may vary.

3. Specificity of Association

   • A single risk factor should relate to a single effect.
   • Example:
     • Tuberculosis and the bacteria that cause it.
   • Note: Few diseases have only one causal agent (e.g., smoking causes multiple disorders).

4. Temporality

   • The cause must precede the effect.
   • Essential for proving causality.
   • Example:
     • Pulmonary embolism and atrial fibrillation relationship.

5. Biological Gradient

   • Dose-response relationship: more exposure leads to more disease risk.
   • Example:
     • Lung cancer risk increases with the amount smoked.
   • Some diseases need a threshold of exposure before symptoms appear.

6. Plausibility

   • Support from biological or laboratory science strengthens the case.
   • Example:
     • Arguments that environmental tobacco smoke doesn’t cause cancer due to lower doses.

7. Coherence

   • New data should not contradict existing evidence.
   • Supports that an exposure causes a cluster of health events.
   • Example:
     • Smoking leads to multiple health issues, not just lung cancer.

8. Experimental Evidence

   • Ideally comes from well-controlled studies (randomized trials).
   • Example:
     • Randomizing sun exposure to study effects on skin cancer.

9. Analogy

   • Similar factors may cause similar effects; consider them as potential causes.
   • Weakest criterion; speculative in nature.

Key Considerations

• Causal criteria viewed as guidelines, not checklists.
• Importance of reviewing literature for alternative explanations.
• Multiple studies help confirm causation; single studies may be misleading.

Modern Models of Causality

• Additional models:
  • Causal pies
  • Counterfactual models
  • Directed acyclic graphs
• Suggested readings available for further exploration.

Conclusion

• Understanding the Bradford Hill Criteria is fundamental in assessing causality in epidemiology and other fields.