Background: A vast number of Helicobacter pylori treatment trials have been conducted. Regimens may vary in efficacy in different patient populations.
Methods: We identified sources of treatment effect variation from 618 treatment groups using weighted cross-classified multi-level meta-regression models. Summary effect estimates were calculated within groups that lacked identified heterogeneity.
Results: Overall, treatment was less successful with shorter treatment duration and dual drug (versus triple or quadruple drug) therapies. For nitroimidazole-based regimens, treatment was less successful in populations with frequent childhood H. pylori infection or metronidazole resistance and more successful in northeastern Asia. Non-nitroimidazole treatments of longer duration and those from less recent reports were most successful. Some one-week regimens--(nitroimidazole/ tetracycline/bismuth, ranitidine bismuth citrate/amoxicillin/clarithromycin, and clarithromycin/amoxicillin/proton pump inhibitor) were highly successful in northeastern Asia regardless of metronidazole resistance. The most successful regimen in populations with both a high prevalence of metrondiazole resistance and frequent infection in children (metronidazole/furazolidone/amoxicillin) eliminated fewer than 70% of infections.
Conclusions: More effective treatments are needed for most populations of the world where H. pylori infection in children and drug resistance are common. Current treatment guidelines do not coincide with the best treatment regimens identified in this meta-analysis.