What is a cause?
An event, condition or characteristic that plays an essential role in producing an occurrence of the disease
Can we prove causality?
Ultimately medical experiments alone can’t prove causality but rather suggest causality
What is Causal Inference?
Judgements linking postulated causes and their outcomes.
What are the three models for observing causal inference?
Epidemiological triads
Causal pies
Causal webs
These don’t allow you to test causality but rather allow you to model possible causal pathways and then go on to test them
What are the four elements of the epidemiological triad?
Agent (carcinogen in smoke of tobacco)
Host (risk factors that influence susceptibility, exposure or response to agent)
Environment (extrinsic factors that affect the agent and the opportunity for exposure)
Vector (transmits agent to host)
What are the 3 classifications of causal inference for exposure?
Sufficient Cause
Component Cause
Necessary Cause
Define: Sufficient Cause
Factor (or often a combination) that will eventually produce a disease
Define: Component Cause
Factor that contributes towards causation but not sufficient to cause disease on its own
Define: Necessary Cause
Any agent that is required to develop a disease. Always precedes disease.
What are the three levels of disease prevention?
Primary, Secondary and Tertiary
Describe Primary Prevention
Prevent disease well before it develops. Reduce the risk factors in the whole population, selected groups and even healthy individuals. Primary care advice as past of routine consultation such as prevention programs.
Describe Secondary Prevention
Early detection of disease (e.g. screening and intervention for pre-diabetes). Selected individuals at high risk (pre-symptomatic). Primary care risk factor reduction programs such as treating asymptomatic HIV patients with retrovirals.
Describe Tertiary prevention
Treat established disease to prevent deterioration and limit disease progression. e.g. Exercise advice for obese patients.
What are the Bradford-Hill criteria for determining causality?
Temporality Biological Plausibility Consistency Strength of Association Biological Gradient / Dose Response Specificity Coherence Experimental Analogy Reversibility (not technically Bradford-Hill but often used on modern epi)
Describe ‘Temporality’ as it relates to determining causality
Does the cause precede the effect or disease (HPV infection precedes cervical cancer)
Describe ‘Biological Plausibility’ as it relates to determining causality
Is the association explained by a credible biologic mechanism (e.g Studies have shown biological plausibility through in-vitro & animal experiments)
Describe ‘Consistency’ as it relates to determining causality
Can the results be replicated in different populations under different conditions (numerous studies have shown the same association between HPV and Cervical Cancer)
Describe ‘Strength of Association’ as it relates to determining causality
Is the magnitude of the association large enough to rule out error due to chance. (RR=17.47. Infection with HPV increases your risk of cervical cancer 17fold.)
Describe ‘Biological Gradient’ as it relates to determining causality
Does the frequency of disease increase with the dose of exposure? (Risk of cervical cancer may increase with viral load)
Describe ‘Specificity’ as it relates to determining causality
Exposure associated specifically with single effect - not multiple effects. Unlikely to get one outcome –> one disease. More specific association –> higher probability of causal relationship. HPV types 16 & 18 specific to cervical cancer. HPV present only in tumour cells and not stromal cells.
Describe ‘Coherence’ as it relates to determining causality
Results of study are not conflicting with current understanding and natural history of disease. Further studies are needed on either side of the debate to prove causality. Coherent with previous knowledge of natural history of HPV infection to cervical cancer described
Describe ‘Experimental’ as it relates to determining causality
Is there positive experimental evidence? Often not available in human data. Animal models studies show evidence of specific papillomaviruses inducing papillomas and cancers in susceptible hosts
Describe ‘Analogy’ as it relates to determining causality
Do other similar exposures cause similar outcomes?
HPV and cervical cancer model is similar to other cancers caused by viruses
Describe ‘Reversibility’ as it relates to determining causality
Does the removal of the possible cause reduce the disease risk? E.g. cessation of cigarette smoking is associated with reduction in lung cancer risk. Furthermore, not smoking is associated with a decreased risk of lung cancer.
