Ex-drinking may be a surrogate for unmeasured risk factors for upper gastrointestinal bleeding: reappraisal and an additional survey of subjects from a case–control study in Japan

In our recent case-control study on the association between non-steroidal anti-inflammatory drugs (NSAIDs) and upper gastrointestinal bleeding (UGIB), the risk of UGIB was higher in non-drinkers than in heavy-drinkers. To explore this unexpected finding, we re-analyzed the data to test whether non-drinking represented risk factors for UGIB and also conducted an additional survey to find any disagreement with the original study that might suggest information bias. In the re-analysis, we estimated odds ratios (ORs) of UGIB by classifying 183 non-drinkers in the original study into 39 ex- and 144 never-drinkers. We also measured the prevalence of potential risk factors of UGIB in controls. In the new survey, we sent a self-administered questionnaire on previous alcohol consumption. In the re-analysis, the OR for ex-drinking (5.4) was higher and the OR for never-drinking (2.6) was lower than the OR for non- (ex- plus never-) drinking (3.2) in the original study. Ex-drinkers had higher prevalence of H. pylori infection, history of previous ulcer and use of GI drugs than other types of drinkers. The answers on their previous alcohol consumption were consistent between the original and new surveys both in cases (kappa = 0.74) and controls (kappa = 0.73). Ex-drinking may represent unmeasured risk factors for UGIB such as undiagnosed ulcer. In an observational study of UGIB, ex-drinking may be distinguished from never-drinking as a potential surrogate for UGIB risk. When prescribing an NSAID to an ex-drinker in clinical practice, one may make sure to explore whether an ex-drinker has a history of ulcer.