Who cares about causation? (5)
- Patients
- Carers and relatives
- Health professionals
- Researchers
- Readers of medical journals
Why is cause important? (3)
- To know why things happen
- To prevent disease by removing the cause
- To improve treatment through greater understanding of natural history of disease
How can we prove causation? (3)
- Find an association
- Consider what might cause the association
- Use criteria that help inform a decision
Who argued falsifiability?
Karl Popper (1934)
What is falsifiability?
Can’t prove a theory is true, but can prove a theory is false
What factors would you consider when finding an association? (4)
- Is it any more common than you would expect?
- Degree of association
- Statistical significance
- Clinical significance
What is chance?
- Random error - associations purely happen by chance
- e.g. lottery winners
What is bias?
- Systematic error in the collection or analysis of information
- E.g. coffee drinkers are more likely to approve of the brand they already drink
What does WEIRD stand for and what is its significance?
Represents up to 80% of study participants, but only 12% of the world’s population
- Western
- Educated
- Industrialised
- Rich
- Democratic
What are cofounding variables?
Both factors are not directly associated but are linked by a third factor
What are the components of the Bradford Hill Criteria? (9)
- Strength of association
- Temporal association
- Consistency
- Theoretical plausibility
- Coherence
- Specificity
- Dose response relationship
- Experimental evidence
- Analogy
What is strength of association?
How strong is the association
What is temporal association?
The effect must occur after the cause
What is consistency?
Consistent findings observed strengthen the likelihood of an effect
What is theoretical plausibility?
- Understanding of the mechanisms helps ascertain causality
- E.g. does high fat diet cause coronary heart disease?
What is coherence?
- Coherence between epidemiological and laboratory findings increases the likelihood of an effect
- E.g. does bacteria cause sexually transmitted infections?
What is specificity?
- Causation is likely if there is a specific population at a specific site with no other likely explanation
- E.g. does vinyl chloride monomer cause hemangiosarcoma?
What is the dose response relationship?
- Greater exposure should generally lead to greater effect
- E.g. does exercise improve health?
What is experimental evidence?
Evidence from laboratory experiments
What is analogy?
- The use of analogies or similarities between the observed association and any other associations
- E.g. does passive smoking cause cancer
What is research design?
How would we design research to test cause and effect
Cross-sectional studies can help generate what…?
Generate hypotheses but not prove causality
How can experimental evidence increase confidence in finding causation? (2)
- Randomisation
- Importance of time (before and after)
What are some difficulties with causation and association? (3)
- A cause may be sufficient or not
- A cause may be necessary or not
- One cause may be preceded by another and by another
- E.g. what causes people to smoke in the first place?
