Because our study was a retrospective cohort study, the diagnosis of lower gastrointestinal (GI) bleeding was largely by exclusion rather than by direct confirmation of the source of bleeding. A lower GI investigation was not performed on every patient because some patients were too frail to undergo further endoscopic examination. Accordingly, presumed small bowel bleeding was a diagnosis of exclusion with no source of bleeding identified in the upper GI tract and the colon. Nevertheless, any misclassification of recurrent lower GI bleeding should be equally distributed between aspirin users and nonusers with the use of blinded adjudication.