Background: Several authors have reported low prevalence of Helicobacter pylori infection in patients with upper gastrointestinal bleeding (UGIB). Our aim was to study the prevalence of H. pylori in bleeding duodenal ulcer (DU), with both invasive and non-invasive methods, and to assess the role of non-steroidal anti-inflammatory drugs (NSAIDs).
Methods: Ninety-two patients with bleeding DU were prospectively studied. The use of NSAIDs was evaluated by specific questionnaire. As a control group, 428 patients undergoing outpatient evaluation for the investigation of dyspepsia and found to have a DU at endoscopy were included. At endoscopy, two antral biopsies were obtained (H&E stain). A 13C-urea breath test was carried out in all patients. Breath test was repeated in patients treated with omeprazole during the hospitalization if H. pylori was not detected with the first test.
Results: Gastric biopsies could be obtained in 39 patients with UGIB. Three patients with UGIB treated with omeprazole and being H. pylori-negative with the first breath test were finally considered infected with the second test. Overall, 92.4% (95% CI, 85%-96%) of the patients with UGIB were infected (89.7% with histology and 92.4% with breath test (P = 0.15)). Concordance kappa value for both diagnostic tests was 0.64. NSAID intake was more frequent in patients with UGIB (34%) than in those without UGIB (5.6%) (P < 0.001), while H. pylori infection was less frequent in patients with UGIB (92.4% (85%-96%)) than in those without UGIB (99.1% (98%-100%); P < 0.001). Even in patients with UGIB, NSAID intake was the only risk factor in 5% of cases. The proportion of cases without H. pylori infection and without NSAID intake was very low in both bleeding and non-bleeding ulcers (2% and 0.5%, respectively; P = 0.146). H. pylori prevalence in bleeding ulcers was of 84% (67%-93%) in patients with NSAID intake, and 96.7% (89%-99%) when patients taking NSAIDs were excluded. In the multivariate analysis, NSAID intake (odds ratio, 9.8 (5.2-18.4)) correlated with UGIB; however, neither H. pylori status nor the interaction between H. pylori infection and NSAID intake correlated with UGIB. In the multivariate analysis in the subgroup of patients with UGIB, NSAID use was the only variable which correlated with H. pylori prevalence (odds ratio, 0.18 (0.03-0.97)).
Conclusions: The most important factor associated with H. pylori-negative bleeding DU is NSAID use, and if this factor is excluded prevalence of infection is almost 100% (97%), similar to that found in patients with non-bleeding DU (and without NSAID intake). Bleeding DU patients with neither H. pylori infection nor NSAID use are extremely rare (only 2%), which suggests that the pathogenesis of bleeding DU is similar to that of non-complicated DU disease.