L26-27: Confounding

Question 1

What are the 3 properties of potential confounders?

Answer 1

• independently associated with exposure
• independently associated with outcome
• not on the causal pathway

Question 2

This is a mixing or muddling of effects when the relationship we’re interested in is confused by the effect of something else.

Answer 2

Confounding

Question 3

What are the common 4 effects of confounding?

Answer 3

• over-estimation of true association
• under-estimation of true association
• change direction of true association (risk factor becomes protective factor and vice versa)
• give appearance of an association when there isn’t one

Question 4

How do we identify potential confounders?

Answer 4

• use literature to identify known and suspected risk factors (property 1: look for risk factors with exposure)
• collect info on factors strongly associated with outcome (property 2: look for strongly associated factors, protective or risk, with outcome)

Question 5

What are the 3 ways to minimise confounding during the design phase?

Answer 5

• restriction
• randomisation
• matching

Question 6

What are the 3 limitations of randomisation?

Answer 6

• does not work well with a small population
• need clinical equipoise
• need intention-to-treat analysis

Question 7

What are the 4 limitations of restriction?

Answer 7

• reduce generalisability
• reduce number of potential participants
• residual confounding
• only one potential confounder can be restricted

Question 8

This controls for confounding by choosing people to make the control/comparison group have the same composition as the case/exposed group regarding potential confounder.

Answer 8

Matching

Question 9

What are the 3 ways to control for confounding in the analysis phase?

Answer 9

• multivariable analysis
• stratification
• standardisation

Question 10

What are the benefits of stratification? (3)

Answer 10

• good for small number of potential confounders
• can evaluate impact of confounding
• identify effect modification

Question 11

What are the limitations of stratification?

Answer 11

• can leave residual confounding
• not good for multiple potential confounders

Question 12

This is a statistical method for estimating measure of association whilst controlling for multiple potential confounders.

Answer 12

Multivariable analysis

Question 13

Which method of controlling for confounding is most suitable for analytic studies?

Answer 13

Multivariable analysis

Question 14

This happens when the association between exposure and outcome differs across strata of the effect modifier. (so different MoA across strata shows this)

Answer 14

Effect modification