Treatment of Helicobacter pylori infection.

Abstract We read with great interest the article of Liou et al published in Gut,1 and we would like to highlight some points. First, the authors stated that patients were considered positive only if two out of three tests performed (ie, rapid urease test, histology and culture) were positive. However, due to its high speci!city (almost 100%),2 a positive culture alone can be considered a marker of Helicobacter pylori infection, as stated by guidelines for clinical trials on H. pylori infection.3 Therefore, it would be clinically useful to know how many patients with positive culture alone were excluded from the study. Second, the successful rate of culture in naïve patients is not clear. This information is methodologically signi!cant as culture was part of the gold standard to consider a patient infected. Third, the authors stated that susceptibility test results for primary resistance were available in only 64.6% of patients. This produces a bias in the interpretation of the results of eradication rates, as it could be said that for only two thirds of the patients there was no difference for prevalence of primary resistance to the antibiotics tested. Fourth, “secondary resistance before second-line treatment” is also reported in table 4, as though a second endoscopy with biopsies was performed in patients who failed the !rst treatment. However, neither the "ow chart nor the material and methods clarify this point. Fifth, it was reported that, for power calculation, a sample size of at least 200 in each groupwas necessary to detect a 10%difference in the eradication rate between the clarithromycin triple therapy (assumed to have an eradication rate of 80%) and the levo"oxacin triple therapy. However, it is not clear if the difference to detect was positive or negative (ie, if the performance of the levo"oxacin triple therapy hypothesised was 70% or 90%) as the number to enrol could be different. Furthermore, it seems that the authors did not account for patients who might drop the study, lowering the power of the study. Finally, the power to detect a difference between the two therapies after the !rst-line