Eleven studies were included in the review. The number of patients overall was unclear (possibly 1,365). However, 1,352 were included in an analysis of the effect on persistent or recurrent bleeding, 1,239 were included in an analysis of the need for surgery, and 1,239 were included in an analysis of mortality
Overall, the quality of the trials was limited: out of a maximum score of 5, one study scored 4, four studies scored 3, three scored 2, and two scored only 1.
Effect on persistent or recurrent bleeding (10 studies included in the analysis): the pooled OR was 0.4 (95% confidence interval, CI: 0.27, 0.59) favouring proton-pump inhibitors, but there was statistically-significant heterogeneity (chi-squared 18, p=0.09.
Effect on the need for surgery (8 studies included in the analysis): the pooled OR was 0.7 (95% CI: 0.43, 1.13), but was not statistically significant. There was no heterogeneity (chi-squared 3.98, p=0.68).
Effect on mortality (8 studies included in the analysis): the pooled OR was 0.69 (95% CI: 0.31, 1.57), but was not statistically significant. There was no heterogeneity (chi-squared 3.44, p=0.63).
The subanalyses found no significant effect on persistent or recurrent bleeding between treatments administered by bolus injection. However, they found statistically-significant differences for high-risk patients (those with Forrest classification Ia, Ib or IIa ulcers) (pooled OR 0.28, 95% CI: 0.16, 0.48) in favour of proton-pump inhibitors, and for those not having endoscopic therapy (pooled OR 0.24, 95% CI: 0.13, 0.43).