Methods to Control for Confounding
- 1.Methods used to control for confounding include:
Restriction can be a powerful method to address a limited number of confounders.
Restriction by the indication for a drug can be used to address confounding-by-indication in observational studies of medication use.
Stratification involves dividing the study population into strata, and then weighing and combining the stratum-specific results.
Restriction, stratification, and matching may not be appropriate for dealing with multiple confounding variables.
Randomization balances both measured and unmeasured characteristics.
A factor is likely to be confounding a given association if adjustment for that factor substantially changes the strength of the association.