Lecture 27 - Confounding I

Question 1

What is confounding

Answer 1

Confounding is the mixing or muddling of effects when the relationship we are interested in is confused by the effect of another variable - the confounder

Question 2

Three properties of a potential confounder

Answer 2

1. Independently associated with the outcome
2. Independently associated with the exposure
3. Not on the casual pathway

Question 3

What effects can confounding have on studies

Answer 3

• Overestimate and under estimate a true association
• Change the direction of a true association (e.g. Simpsons paradox)
• Give the appearance of an association when none exists

Question 4

Identifying potential confounders

Answer 4

• Plan ahead e.g. collect information on all potential confounders
• Look for imbalance in potential confounders between groups
• Application e.g. what effect might the potential confounder have had on the RR

Question 5

How can confounding be controlled in study design

Answer 5

1. Randomisation - distributes confounders equally across groups
2. Restrictions - Limits study to participants with the same level of the confounder
3. Matching - Ensures comparison groups have similar levels of the confounder

Question 6

What are the strengths of randomisation for controlling confounding

Answer 6

It works best in large sample sizes, ensures groups are similar, and is only used in RCT
BUT
- works best with large sample size
- Need equipoise
- Need intention to treat analysis

Question 7

What is the main limitation of restriction of controlling confounding

Answer 7

It reduces generalisability and the number of potential participants and only controls for one confounder at a time. Potential for residual confounding with imprecisely measured confounders

Question 8

What is matching and when is it used

Answer 8

Matching involved selecting control participants so that their confounder levels match those of cases/exposes groups. It is often used in case control

Question 9

Positives of matching

Answer 9

• useful for difficult to measure/complex potential confounders
• Can improve efficiency of case control studies with small numbers

Question 10

Negatives of matching

Answer 10

• Individual matching can be difficult and limit number of potential participants
• Need special matched analysis for individual matching. Otherwise will underestimate the measure of association

Question 11

Individual and frequency matching

Answer 11

Individual - each case matched with one or more controls having the same confounding variable characteristics
Frequency - Matching at aggregated (more people) level

Question 12

How do we identify potential confounders

Answer 12

• By looking for variables that are imbalanced between groups and using literature to identify known and suspected risk factors for the outcome