Proton Pump Inhibitors (PPIs) Given for Acute Peptic Ulcer Bleeding

34 for mortality (in Asia), no benefit in Western studies

Benefits in NNT

None were helped (preventing death, decreasing blood transfusions or having a shorter hospital stay)
1 in 15 were helped (preventing rebleeding: benefits were found in Asian studies with trends toward harm in Western studies)
None were helped (avoiding a surgical intervention)
None were helped (avoiding repeat endoscopy)
0% were helped by preventing death, decreasing blood transfusions or having a shorter hospital stay
6.6% were helped by preventing rebleeding (benefits were found in Asian studies with trends toward harm in Western studies)
3.2% were helped by avoiding a surgical intervention
10% were helped by avoiding repeat endoscopy

Harms in NNT

None were harmed (death or medication side effects)
0% were harmed by death or medication side effects
View As:

Efficacy Endpoints

Mortality, decrease in rebleeding, surgical intervention, repeat endoscopy.

Harm Endpoints

Medication side effects


Peptic ulcer disease (PUD) is a common medical emergency, where bleeding from ulcers in the stomach or duodenum can result in substantial morbidity, mortality and healthcare costs. Proton pump inhibitors (PPI) are gastric acid inhibitors, but the evidence supporting their clinical efficacy for patient important outcomes has been mixed.

This Cochrane review finds, based on a meta-analysis of 24 randomized trials of 4373 subjects, PPI treatment reduced rebleeding (NNT 15, 95% CI 0.4-0.7), surgical intervention (NNT 32, 95% CI 0.5-0.8) and repeat endoscopy (NNT 10, 95% CI 0.2 to 0.5). However, there was no effect on mortality (OR 1.01, 95% CI 0.74 to 1.40).

In regard to blood transfusions, there was a noted difference of marginal statistical significance (weighted mean difference (WMD)-0.6 units of blood; 95% CI 1.1 to 0.0 p=0.05), which was not robust on the exclusion of individual trials. Looking at these trials, no indication for administering blood transfusions was given, so the strength of the conclusion is limited. Similarly, it did not appear that PPI use had an effect on overall length of stay.


Interestingly, PPI use was associated with robust treatment effects for mortality, rebleeding, and surgical intervention in the trials conducted in Asia. There was a benefit in mortality (NNT=34, 95% CI 0.2 to 0.7), rebleeding (NNT=6, 95% CI 0.2 to 0.4), and surgical intervention (NNT=23, 95% CI 0.2 to 0.5). These benefits remained significant even when any one of the individual Asian trials was excluded from the analysis. In contrast, in the European trials, there was a trend towards increased harm in the PPI group, though it did not reach statistical significance. There are multiple theories about why there appears to be such divergent outcomes in these two patient populations, but until further trial data they remain speculative.

Regardless, it appears that the use of PPI for endoscopically confirmed upper GI bleeding is likely beneficial in the Asian populations studied, and either harmful or unhelpful in European populations. Indeed it is notable that two of the largest, most rigorous European studies, Daneshmund (1992) and Hasselgren (1997) both found increases in mortality that approached statistical significance in the PPI group, and one was stopped due to this apparent increase. Based on these importantly divergent findings it seems wisest to therefore consider the two populations separately rather than to pool them statistically (which would likely obscure or dilute the true effect in both populations).


Manpreet Singh, MD


June 5, 2013