Epidemiology

A cohort study finds that coffee drinkers have more lung cancer, but coffee drinkers are also much more likely to smoke. What analysis is most directly useful for evaluating confounding by smoking?