What is known and ObjectiveDigestive disorders represent the most common metformin side effects for type 2 diabetes. The mechanism of these metformin side effects is unclear. The aim of this study was to assess whether asymptomatic chronic gastritis could influence metformin tolerance in patients with type 2 diabetes. MethodsDemographic, anthropometric, ultrasound and laboratory data were obtained from 144 metformin naive patients with diabetes. The diagnosis of chronic gastritis was based on endoscopic and histopathological examination, and H.pylori infection was assessed based on C-13 urea breath test (UBT). All subjects started metformin at 500mg/day and increasing progressively to 1500mg/day over 4weeks. A score of gastrointestinal side effects (abdominal pain, diarrhoea, nausea, vomiting, bloating and anorexia) was assessed each week, and metformin dose was adjusted as appropriate. Results and discussionBased on endoscopy, 64 patients were categorized as non-gastritis subjects and 80 as chronic gastritis subjects. At baseline, there is no statistical difference in gastrointestinal symptoms between two groups. With metformin, the mean scores for gastrointestinal symptoms in the non-gastritis and gastritis subjects were 102171 vs. 218 +/- 205 (P=0001), 020 +/- 065 vs. 050 +/- 089 (P=0022), 0 vs. 006 +/- 024 (P=0024) and 108 +/- 103 vs. 171 +/- 166 (P=0028). The mean final metformin dose used by gastritis subjects was 70624 +/- 56890mg, significantly less than the mean dose used by non-gastritis subjects (110156 +/- 57858mg, P=0001). After adjustment for age and sex, the odds ratio (OR) for a final metformin dose of less than 1500mg/day was found to be 276 (95% CI 138-553, P=0004) for chronic gastritis subjects. The OR for a final metformin dose of less than 1000mg/day was found to be 398 (95% CI 191-827, P=0001) for chronic gastritis subjects. What is new and conclusionsOur data suggest that pre-existing non-symptomatic gastritis was associated with metformin-related gastrointestinal side effects.