5th Jan

Question 1

What variables do you need in an analytical study?

Answer 1

Exposure
Outcome
Covariates - variables that affect exposure and/ or outcome

Question 2

How do you find covariates?

Answer 2

What are the likely causes of exposure?
What are the likely causes of the outcome?
What is likely to be caused by the exposure?

Question 3

What are confounders?

Answer 3

Variables causing exposure and outcome

Question 4

What are mediators?

Answer 4

Variables caused by exposure that cause outcome

Question 5

What are competing exposures?

Answer 5

Variables causing the outcome but no relationship with exposure

Question 6

What are DAG graphs?

Answer 6

Directed Acyclic Graphs
Summarise relationships between variables
Identify covariates acting as confounders, competing exposures and mediators

Question 7

Why do we need to adjust for confounders?

Answer 7

They create a pseudo-casual path between outcome and exposure which will generate a statistical relationship even when none exists

Question 8

Why do we not adjust for mediators?

Answer 8

They are part of the causal path between outcome and exposure

Question 9

Why might we adjust for competing exposures?

Answer 9

If they cause a substantial amount of variation in the exposure, adjusting it can make any association between remainder and exposure easier to detect

Question 10

What is a causal relationship?

Answer 10

Can be:
functional (no contraception = teenage mother)
empirical (based on previous statistical analysis)
theoretical (teenage grandmother = teenage mother)
speculative (teenage grandfather = teenage mother)

Causal relationship is only possible if cause variable precedes effect