Eighteen RCTs (n=1,855) were included.
High-dose intravenous PPIs (4 RCTs, n=682).
Compared with placebo, high-dose PPI was associated with a significant decrease in re-bleeding (-14.6%, 95% CI: -16.2, -12.9) and surgery (-5.4%, 95% CI: -8.4, -2.4), but a non statistically significant reduction in mortality (-2.7%, 95% CI: -9.2, 3.8). Endoscopic treatment was used in 52% of the patients.
Compared with H2RA, high-dose PPI was associated with a significant decrease in re-bleeding (-20.6%, 95% CI: -24.7, -16.6), but there was no significant difference between treatments for surgery (-1.0%, 95% CI: -8.0, +6.1) or mortality (-2.4%, 95% CI: -17.7, +12.9). After excluding one RCT with a low mortality rate in the placebo group, the reduction in mortality with high-dose PPI was statistically significant (-5%, 95% CI: -7.7, -2.3).
High-dose oral PPIs (4 RCTs, n=448).
Compared with placebo, high-dose oral PPI was associated with a significant decrease in re-bleeding (-15.3%, 95% CI: -16.5, -14.0), but there was no significant difference between treatments for surgery (-3.3%, 95% CI: -6.3, +0.3) or mortality (-1.4%, 95% CI: -2.7, +0.2).
There was no statistically significant difference for re-bleeding, surgery or mortality between high-dose oral PPI and either H2RA or SST (the results were reported).
Non high-dose PPI treatment (10 RCTs, n=725).
Compared with placebo, non high-dose PPI was associated with a significant decrease in re-bleeding (-25.0%, 95% CI: -29.3, -20.7), surgery (-16.2%, 95% CI: -18.1, -14.2) and mortality (-3.5%, 95% CI: -4.6, -2.4).
Compared with H2RA or SST, there was a statistically significant decrease in re-bleeding (-14.4%, 95% CI: -21.2, -7.7), but there was no significant difference between treatments for surgery or mortality (the results were reported).