Six studies (522 patients) were included in the review. No studies met all the quality criteria; the number of criteria met ranged from three to seven (out of eight). Two studies had adequate sequence generation or adequate allocation concealment, three had adequate blinding or adequate addressing of incomplete data, four had sample size calculations, five were free from baseline imbalances and all six were free from selective reporting. Follow up ranged from 28 to 90 days.
Compared to the H2RA group, the PPI group had a significantly lower bleeding rate (OR 0.5, 95% CI 0.3 to 1.0; Ι²=39%; five studies). Subgroup analysis according to type of endoscopic treatment indicated that PPI was significantly more effective than H2RA in preventing bleeding for endoscopic submucosal dissection induced ulcer (OR 0.4, 95% CI 0.2 to 0.9; Ι²=31%; five trials) but there was no evidence of a difference between groups with endoscopic mucosal resection induced ulcer. Subgroup analysis according to duration of medication indicated that PPI was significantly more effective than H2RA in preventing bleeding with eight weeks of medication (OR 0.4, 95% CI 0.2 to 0.8; Ι²=0%; three trials), but not with four weeks of medication. Sensitivity analysis did not change the results markedly.
There was no evidence of significant differences in epigastric pain between comparison groups. No studies found evidence of a difference in healing rates between comparison groups (data not reported).
There was no evidence of publication bias.