"A confounder is a factor that is prognostically linked to the outcome of interest and is unevenly distributed between the study groups"
Confounding leads to an error in the interpretation of a measurement.
Your measurement might be accurate, but if "confounded" you may attribute it to the wrong cause.
One can correct for known confounders using statistical methods or by organising results according to known confounders (thus separating the sample into groups). The unknown confounders can be corrected for with the use of "propensity score analysis ", which (according to LITFL) may not be useful. All sorts of confounders can be corrected for with the use of randomisation - if the confounders are equally and randomly distributed between the study groups, then their influence on the outcome is negated.