Confounding refers to differences (in prognostic factors, or concomitant treatments) between treatment comparison groups, other than the treatments being compared.
Confounding results in biased effect estimates.
For a factor to lead to confounding, it must differ between the treatment comparison groups, and affect the outcome of interest.
An example could be age for treatments that aim to improve physical mobility. Older people will tend to have lower mobility irrespective of the treatment.
If you feel that this definition hasn't helped you to understand the term, click on our monkey to let us know.