Sixteen RCTs with 3,154 participants were included in the review.
Numerous design flaws were found in the majority of the studies, such as a relatively low dosage of omeprazole used, a small number of patients, and inhomogenous patient characteristics with multiple bleeding etiologies. Only 5 of the studies were double-blinded.
In the 11 studies in which endoscopic therapy was not administered, the rebleeding rate in the control groups had a very wide range, from 1.5 to 85%. Omeprazole, however, was generally associated with lower rates of rebleeding, ranging from 0 to 24%; this attained statistical significance in 3 studies.
In the 5 studies that used concomitant initial endoscopic therapy in both arms, rebleeding rates in the control groups ranged from 17 to 24%. The active therapy groups again showed a generally lower rebleeding rate, ranging from 4 to 26%. Only one of these studies reached statistical significance.
Three studies described the incidence of recurrent bleeding based on endoscopic criteria, e.g. stigmata of recent haemorrhage, which can be important, independent predictors of recurrent ulcer bleeding. Two of the studies reported statistically-significant results in terms of a decreased bleeding rate (p<0.001 and p<0.05), particularly with type IIa and IIb ulcers from the Forrest classification (non-bleeding visible vessels or clots).
The incidence of continued bleeding requiring surgical intervention was similar in the studies with and without endoscopic therapy. In the 11 trials that did not use endoscopic therapy, 2 demonstrated statistically-significant reductions compared with placebo, while 3 studies described a trend toward lower surgical rates, although these did not attain statistical significance. The other 6 trials did not reveal any statistically-significant results pertaining to this outcome. Among the 5 trials that used adjunct endoscopic treatment, 2 reported statistically-significant lower surgical rates compared with placebo.
None of the studies demonstrated a significant mortality reduction.