Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Is there a role for Emergency Department proton pump inhibitors in acute upper gastrointestinal bleeding?

Three Part Question

In [patients with acute upper gastrointestinal bleed] does giving [intravenous proton pump inhibitor prior to endoscopy] [improve morbidity and mortality]?

Clinical Scenario

After seeing a 50 year old man with coffee ground vomiting secondary to NSAID use you refer him to the RMO. You know the SIGN guidelines don't advocate pre-endoscopic proton pump inhibitors in non-variceal upper gastrointestinal haemorrhage, but ,yet again, the RMO requests an IV PPI. You wonder whether the medical SHO cohort know something that you and SIGN do not, so decide to evaluate the evidence yourself.

Search Strategy

Medline via Ovid Interface 1950 to November week3 2010 and PubMed
{Exp Proton Pump Inhibitor OR Exp Omeprazole OR Proton Pump OR OR OR OR OR OR OR OR OR} AND {Exp Infusion, Intravenous OR Exp Injections, Intravenous OR OR} AND {Exp Endoscopy OR Exp Endoscopy, Gastrointestinal OR} AND {Exp Peptic Ulcer Haemorrhage OR Exp Gastrointestinal Haemorrhage OR Exp Malaena OR Exp Haematemesis OR Gastrointestinal OR Gastrointestinal OR GI OR GI OR OR} AND {Exp Mortality OR OR Exp Survival OR}limit to Humans.
CINAHL, EMBASE Cochrane Library and Cochrane Central Register of Controlled Trials were searched using the search strategy:
(proton pump inhibitors OR ppi OR omeprazole OR lanzoprazole OR esomeprazole OR pantoprazole OR losec OR zoton OR nexium OR protium) AND intravenous AND (Haemorrhage OR bleeding OR haematemesis OR malaena)

Search Outcome

This search identified 237 potential studies. Several were repeated hits from different databases. 6 studies were related to the topic. 1 was a Cochrane systematic review and meta-analysis from 2006 directly answering the 3 part question. Another was a 2010 update of the same Cochrane review.1 was a study used in the 2010 Cochrane review,another was a review article on the same paper and 2 were retrospective case reviews.
The updated 2010 Cochrane review was the only paper used to answer the 3 part question

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Sreedharan A et al
6 RCTs involvolving 2223 participants with signs of upper gastrointestinal bleeding were randomised to receive pre-endoscopic proton pump inhibitors (intravenous or oral)(1114)or control (placebo or H2 antagonists)(1109)Systematic Review and Meta-analysis30 day Mortality6 trials. No significant heterogeneity (p=0.60, I2=0%) PPI vs control 4.9% vs 4.3%. (OR 1.12, 95% CI 0.75 - 1.68). No significant differenceOn sensitivity analysis, the significant reduction in endoscopic findings were heavily influenced by a single study and when that study was removed the difference became non-significant. 3 studies had no a priori sample size estimate. 4 studies had inadequate description of the randomisation process.5 studies had inadequate description of allocation concealment.1 study was described as unblinded and 2 had inadequate description of the blinding process.1 study did not describe dropouts/withdrawals from the study. 3 studies included participants with variceal bleeding (accounting for 2.5-4.4% of the total population of the affected studies) Only 1 study reported co-morbidities and 2 reported haemodynamic status at presentation. 2 studies did not describe time to endoscopy. The other 4 described endoscopy within the first 24 hours.
Rebleeding within 30 daysFive trials. No significant heterogeneity (p=0.45, I2=0%) PPI vs control 11% vs 13.1% (OR 0.81, 95% CI 0.62 - 1.06). No significant difference.
Surgery within 30 days5 trials. No significant heterogeneity (p=0.53, I2=0%). PPI vs control 7.2% vs 7.9% (OR 0.90, 95% CI 0.65 - 1.25). No significant difference
Proprtion needing blood transfusion4 trials. No significant heterogeneity ( P=0.36, I2= 6.1%) PPI vs Control 53.2% vs 54.5% (OR 0.95, 95% CI 0.78 - 1.16). No significant difference
Units of blood transfused2 trials. No meta-analysis performed. Lau 2007 PPI vs control 1.54 vs 1.88 (p=0.12), No significant difference. Naumovski 2005 PPI vs control 2.5 vs 4.2 (p=0.001) Significant reduction.
Endoscopic Stigmata of recent haemorrhage4 trials. No significant heterogeneity (P=0.20, I2= 35%) PPI vs Control 37.2% vs 46.5% (OR 0.67, 95% CI 0.54 -0.84). Significant reduction
Blood in stomach on Endoscopy3 trials. Statistical heterogeneity (P=0.07,I2=62.9%), PPI vs Control 20.6% vs 27% (OR 0.64, 95% CI 0.32 - 1.30). No significant difference.
Active bleeding at Endoscopy4 trials. No significant heterogeneity (P= 0.52, I2= 0%) PPI vs Control 11.3% vs 14.7% (OR 0.74, 95% CI 0.54 - 1.02). No significant difference.
Haemostatic treatment at endoscopy3 trials. No significant heterogeneity (P=0.41, I2=0%) PPI vs Control 8.6% vs 11.7% (OR 0.68, 95% CI 0.50 - 0.93). Significant reduction.
Length of Stay3 trials. Meta-analysis not possible. Daneshmend 1992 PPI vs Control 5 vs 6 days (not significant), Wallner 1996 PPI vs Control 8 vs 7.6 (Not significant), Lau 2007 PPI vs Control 3 vs 3 days (Statistically different


The 2010 update of the good quality sytematic review and meta-analysis by the Cochrane group on this topic demonstrates that pre-endoscopic administration of a proton pump inhibitor may reduce the stigmata of recent haemorrhage and need for haemostatic intervention at endoscopy. However this did not translate into a reduction in 30 day mortality, rebleeding within 30 days, surgery within 30 days or need for blood transfusion. Subgroup analyses had also been planned a priori. They found that the control therapy (H2 antagonist or placebo), route of administration (oral or intravenous), which PPI was used and diagnosis at endoscopy made no difference to these findings. In the 4 trials that reported on time to index endoscopy they were generally performed within the first 24 hours, so whether the results of this review are generalisable to your practice depends on how quickly an endoscopy can be performed in your institution.

Clinical Bottom Line

Proton pump inhibitors should not routinely be given to patients in the Emergency Department presenting with non-variceal upper gastrointestinal haemorrhage. However, PPI therapy may have a role if endoscopy is not readily available.


  1. Sreedharan A, Martin J, Leontiadis GI, Dorward S, Howden CW, Forman D, Moayyedi P Proton pump inhibitor treatment initiated prior to endoscopic diagnosis in upper gastrointestinal bleeding (Review) Cochrane Database of Systematic Reviews 2010,Issue 7. Art. No.: CD005415. DOI: 10.1002/14651858.CD005415.pub3