Can anyone help me wiht understanding concept of confounders in epidemiology?

Answer 1

Confounding is a problem where a variable has an effect on BOTH the exposure and outcome, so it interferes with establishing causality. The classic example that everyone (well, all the epi textbooks, anyway) uses is the association between alcohol and lung cancer.

An early study showed a significant association between alcohol consumption and an elevated risk of getting lung cancer. However, that study was confounded by smoking. As any bartender can tell you, people are more likely to smoke if they drink (and while they drink). That's how smoking affects drinking. Then of course, smoking is independently linked to lung cancer. If you control for smoking in the study, you actually find that there is no link between alcohol and lung cancer (but your liver is another story).

Here's a real world example from my own work. I investigated an outbreak of foodborne illness at a Thankgiving party. 300 exposed, 250 ill. Two foods were statistically significantly associated with illness: turkey and gravy.
Turkey (OR=7.86, 95% confidence interval: [3.0428-20.3114])
Gravy (OR=5.60, 95% confidence interval: [2.7580-11.3807])
Both had attack rates above 80%.

I re-ran my analysis controlling for both turkey and gravy. The test for the significant of turkey controlled for gravy showed an odds ratio of 7.27, with 95%ci [1.35–39.05]. But the test of gravy controlled for turkey was no longer significant. Gravy was the confounder.

If you have any further questions, please add them to your question details and I will gladly expound on them.