Predictors of failure of Helicobacter pylori eradication and predictors ofulcer recurrence: a randomized controlled trial

Objective: ln light of evidence that Helicobacter pylori treatment fails 5% to 20% of the time, the objective of this study was to determine predictor s of unsuccessful H. pylori eradication and of duodenal ulcer recurrence. Design: Randomized, double-blind, placebo-controlled trial. Setting: Gastroenterology services of 2 general hospitals in Montreal, Que. Patients: All patients (aged 16 to 90) with an endoscopically proven duoden al ulcer within the previous year and H. pylori infection detected on antra l biopsy were asked to participate; 85 were included. Interventions: Patients were randomized in double-blind fashion to 1 of 2 e radication therapies, consisting of metronidazole, bismuth subcitrate and e ither amoxicillin or placebo. Endoscopy was performed at follow-up every 3 months for 12 months. Outcome measures: Demographic data, characteristics of patients and disease , previous history and family history of ulcer disease, compliance at day 1 0 and day 28 of therapy; in vitro metronidazole resistance of H. pylori; er adication of H. pylori (determined by endoscopic biopsy 3 months after ther apy); and ulcer recurrence within 12 months after therapy. Results: Metronidazole resistance (odds ratio [OR] 0.11, 95% confidence int erval [CI] 0.017 to 0.69) was the only independent predictor of eradication . Compliance las defined in the study), density of organisms on culture, as well as several other factors examined, were not significant predictors. T reatment group, although a significant factor on univariate analysis, was n ot an independent predictor on multivariate analysis, as there were relativ ely good eradication rates (82% and 97% among compliant patients) in both g roups. With regard to ulcer recurrence, 3 independent predictors were ident ified: failed H. pylori eradication (OR 86.5, 95% CI 4.2 to 1769), unemploy ment (OR 13.2, 95% CI 1.8 to 95) and a family history of ulcer disease (OR 12.2, 95% CI 1.2 to 128). Conclusions: The best predictor of ulcer recurrence is failure of H. pylori eradication, which, in turn, depends on metronidazole resistance. Hence, t reatments containing metronidazole should be avoided in populations with hi gh rates of metronidazole resistance. A family history of ulcer disease and unemployment were also predictors of ulcer recurrence, which suggests a po tential role for treatment of contacts.