Content area
Abstract
[...]a disturbing number of interviewed patients were omitted from the study, and the omitted group included 2.4 times as many controls as cases -- again, an enormous opportunity for bias to influence the results. First of all, the data on coffee consumption were limited to the number of cups consumed on a typical day before onset of the current illness. [...]controls who had diseases associated with smoking and alcohol consumption were excluded. Since smoking and coffee drinking are correlated, exclusion of heavy smokers, who would be overrepresented in tobacco-related diseases, would tend to remove heavy coffee drinkers from the control group. [...]in a report of such intense public interest, it is useful to look for more evidence of the association between coffee and pancreatic cancer.