Controlling confounding in the study analyses
- 3 methods
- stratification
- multi variable analysis
- standaisation
Each can be used to control for confounding and evaluate whether it has occurred
Must have measured potential confounder!
Basic recipe of stratification
- Calculate the measure of association between exposure and outcome (crude, univariate, unadjusted)
- Divide potential confounder into strata (levels)
- For each stratum, calculate the measure of association between the exposure and outcome (i.e. calculate the stratum-specific measures of association)
- Compare stratum-specific measures of association
Pros of stratification
- Easy for small number of potential confounders with limited
strata - Can evaluate impact of confounding
- Can identify effect modification (soon…)
Cons of stratification
Can leave residual confounding
Not feasible when dealing with lots of potential confounders with many strata
What is multi variable analysis
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’
When do we age standardise
Problems with standardisation
Similar issues as stratification with multiple potential confounders/number of strata
Multivariable analysis is often more efficient in analytic studies
Potential issues: controlling in study analyses
- residual confounding
- can only control what you’ve measured
As long as u measure the confounders it can still be controlled in the analysis
Yes
How much change indicates confounding?
General guideline is if controlling for confounding changes the measure of association by 10% or more
Confounding vs effect modification
