8: Causality in Research- Associations Vs Causation ✅

Question 1

Scientific knowledge requires

Answer 1

Explanations to state the causes of outcomes

Question 2

What criteria must be met to be valid?

Answer 2

1. Chance:
   -has to be excluded as a possible explanation (p-value and 95% Confidence interval)
2. Bias:
   -has to be excluded as a possible explanation (check of errors)
3. Confounding:
   -has to be excluded as a possible explanation (adjust and identify for all potential confounders)

Question 3

Associations vs causations

Answer 3

Valid association cannot automatically prove causation

Question 4

Casual association

Answer 4

The exposure directly or indirectly causes the outcome

Question 5

Bradford Hill: criteria for causality

Answer 5

Shouldn’t be used as definitive proof/disproof of a causal association
->instead to help decide on the likelihood

Some are not relevant any more to modern epidemiology

Question 6

Bradford Hill

Answer 6

1) Temporality
- Exposure must precede outcome

2) Strength 
- The stronger the association between exposure and outcome, the more likely it is to be casual 

3) Biological gradient 
- Dose-response association between exposure and outcome 
"as dose increases, response increases"

4) Reversibility
- Removing the exposure reduces the incidence of the outcome 
 
5) Consistency
- An association is observed in multiple occasions

6) Plausibility
- There is a theoretically sound explanation for the observed association 

7) Analogy 
- There are other analogous observed findings in the literature
- Subjective so not usually used
 
8) Coherence 
- The association fits with the known facts of the natrual history and pathophysiology of the disease primarily coming from in vitro experiments 
- Related to plausibility so therefore not used often 

9) Specificity 
- The exposure of interest is associated only with the outcome of interest and no other outcome 
- Doesnt apply to non-infectious diseases

Question 7

Necessary and Sufficient Cause

Answer 7

• Necessary and sufficient: always present and can cause the disease on its own
  • Necessary but not sufficient: always present but cannot cause the disease on its own
  • Sufficient but not necessary: can cause the disease on its own but not always present
  • Neither necessary nor sufficient: cannot cause the disease on its own and not always present

Question 8

Rothman’s causality pies

Answer 8

Any given outcome has a combination of specific causes

Each individual case of the outcome is caused by a different combination of these causes

Chronic diseases often dont have “necessary and sufficient causes”

Infectious diseases and genetic diseases usually only have one “necessary and sufficient” cause