Confounding - II

Question 1

What are three ways we can control confounding in the study analyses

Answer 1

• stratification
• multivariable analysis
• standardisation

To use these we must have measured a potential confounder!
Each can be used to control for confounding and evaluate whether it has occurred

Question 2

Describe what stratification is and how it works/how to do it

Answer 2

Calculating measure of association for each stratum of potential confounder and comparing them

1. Calculate the measure of association between exposure and outcome (crude, unadjusted data)
2. Divide potential confounder into strata (levels)
3. For each stratum, calculate the measure of association between the exposure and outcome (the stratum-specific measures)
4. Compare stratum-specific measures of association

Question 3

What are the pros and cons of stratification?

Answer 3

Pros
- easy for small number of potential confounders with limited strata
- can evaluate impact of confounding
- can identify effect modification

Cons
- can leave residual confounding
- not feasible when dealing with lots of potential confounders with many strata

Question 4

What is multivariable analysis and why is it good

Answer 4

Statistical method for estimating measure of association whilst controlling for multiple potential confounders
- can work in situations where stratification won’t
- variety of different techniques recognisable by the term ‘regression’
- don’t need to know the specifics of these techniques or how to use them!

Question 5

what is standardisation and some cons about it?

Answer 5

Used when age structures differ ADN disease risk varies by age

Cons
- similar issues as stratification with multiple potential confounders/umber of strata
- multivariable analysis is often more efficient in analytic studies

Question 6

what are the potential issues when controlling in study analyses?

Answer 6

• residual confounding
• can only control what you’ve measured
• it is important to measure potential confounders because you can control for them once you have

Question 7

describe assessing confounding (evaluating confounding and how much change inficates confounding?)

Answer 7

Evaluating confounding
- need to be controlling for potential confounder in analyses
- compare crude and adjusted measures of association

How much change indicates confounding?
- general guideline is if controlling for confounding changes the measure of association by 10% or more
- but also just use your judgement. has confounding maternally changes your interpretation of the measure of association?

Question 8

confounding vs. effect modification

Answer 8

confounding:
- a third factor distorting the association
- a nuisance

effect modification:
- the association between exposure and outcome differs across strata of the effect modifier
- an IMPORTANT finding

Question 9

how do you identify effect modification?

Answer 9

after you have stratified and calculated stratum-specific measure of association, have a look at the numbers
- if its confounding then the stratum specific data will be very similar if not the same
- if its effect modification the stratum specific data will be different (and if you reported it as confounding in this case, it would just be an average, and not a very helpful number/value/wouldn’t tell us much)
- so stratification is important to identify this difference between confounding and effect modification!