I have a question about confounding. It is about controlling for confounders in an observational study that investigated the effect of several genetic polymorphisms on the occurrence of cardiovascular events. We had to answer questions about this article, but I became confused by the answer provided by my teacher.
This was the asnwer: "Confounders are common causes for the exposure of interest (the presence of a polymorphism) and the outcome (a composite death from any cause, non-fatal AMI, or stroke). However, none of the prognostic factors (Table 3, (this was for example age but also the type of drugs used by the patient)) in the statistical model can be a cause for a genetic polymorphism" Therefore, the multivariate model makes little sense, and is not valid. (They controlled for these confounders).
I thought that an confouder should cause the outcome, but only has to be associated (so no exactly causing) the exposure. For example, since it is an observational study, only as result of chance, people with a certain polymorphism could also be older. Then age can contribute to the risk of the outcome. So I thought, that in such a situation you should control for age if you want to study the true effect of the polymorphism.
But I'm just confused now, so I hope that someone can help me with this.
