Helicobacter pylori infection and gastric cancer: systematic review of the epidemiological studies J. DANESH Clinical Trial Service Unit and Epidemiological Studies Unit, Nuf®eld Department of Medicine, University of Oxford, Radcliffe In®rmary, Oxford, UK. Accepted for publication 3 February 1999 INTRODUCTION When Helicobacter pylori was ®rst identi®ed, those that discovered it suggested that it might cause gastric cancer, 1 which kills about a million people a year worldwide. 2 H. pylori is found in the stomachs of about one-third of the adults in developed countries (where it has been decreasing in prevalence) and in about two- thirds of the adults in developing countries. 3 It is strongly correlated with lower socioeconomic status, usually acquired in early life, probably spread from person to person, and generally persists until old age unless eradicated with speci®c antibiotic treatment. 4 A proportion of infected people develop peptic ulcer- ation. 5 There is, however, a great deal of uncertainty as to whether the infection can also predispose to gastric cancer. In 1994 the International Agency on Research in Cancer declared H. pylori a `carcinogen', 6 but a consensus panel of the National Institutes of Health disagreed. 5 A number of additional studies have been reported recently, but expert opinions continue to differ, with claims ranging from a nine-fold associa- tion 7 to no important association at all. 8 To help clarify the actual evidence, this article provides a systematic review (meta-analysis) of the epidemi- ological studies. SUMMARY Background: The published epidemiological studies of chronic Helicobacter pylori infection and gastric cancer yield con¯icting results, so there is uncertainty as to whether any material association exists and, if so, how strong it is. Aim: To review these studies quantitatively. Methods: A systematic review of sero-epidemiological studies published before 1998 of H. pylori and gastric cancer, as identi®ed by computer-assisted literature searches of relevant journals, reference lists and discus- sions with authors. All relevant studies identi®ed were included, subdivided by study design. The following was abstracted from published reports: adjusted odds ratio (or, in prospective studies, the risk ratio) and con®dence interval, study design, type of controls, mean age, mean duration of follow-up, assay methods, location of study, and degree of adjustment for confounders. Results: The 34 retrospective studies included in total 3300 gastric cancers, but their controls were of uncertain validity. The 10 `nested' case±control com- parisons in prospective studies included in total only 800 gastric cancers, and combined analysis of them yielded a risk ratio of 2.5 (95% CI: 1.9±3.4; 2P < 0.00001) for gastric cancer in people seropositive for H. pylori antibodies. Conclusions: The prospective studies suggest that gastric cancer is 2 or 3 times as common in those chronically infected by H. pylori, but to help investigate causality, further observational studies are still needed, as are large-scale randomized trials of whether antibacterial regimens reduce the eventual incidence of gastric cancer. Correspondence to: Dr J. Danesh, CTSU, Radcliffe In®rmary, Oxford OX2 6HE, UK. E-mail: john.danesh@balliol.ox.ac.uk Aliment Pharmacol Ther 1999; 13: 851±856. Ó 1999 Blackwell Science Ltd 851