Effect of Stress Ulcer Prophylaxis With Proton Pump Inhibitors vs Histamine-2 Receptor Blockers on In-Hospital Mortality Among ICU Patients Receiving Invasive Mechanical Ventilation: The PEPTIC Randomized Clinical Trial

JAMA. Published online January 17, 2020. doi:10.1001/jama.2019.22190

Clinical Question

  • In ICU patients requiring invasive mechanical ventilation, does the use of proton pump inhibitors (PPIs) vs. histamine-2 receptor blockers (H2RBs) for stress-ulcer prophylaxis, reduce 90-day mortality?


  • Upper GI bleeding due to stress ulcers is a common complication of critical illness affecting 2.5% of patients admitted to ICU
  • Prophylaxis against stress ulcers is widely accepted as standard ICU practice but there is wide variation in the choice of drugs
  • Although PPIs may be more effective than H2RBs in suppressing gastric acid secretion and preventing upper GI bleeding, there has been concern that they may also increase the risk of ventilator-associated pneumonia, Clostridioides difficile infection, and immunosuppression


  • Pragmatic, open-label, cluster randomised crossover design, meaning whole units used one class of drug as their default for 6 months, then the other
  • Primary outcome changed in March 2017 prior to any data analysis
  • Power calculation: 80% power to detect an absolute reduction of 2.4% from a baseline mortality of 15%, with a false positive rate of 5%


  • 50 ICUs in 5 countries
  • 2016-2019


  • Inclusion:
    • Aged 18 years or older
    • Requiring invasive mechanical ventilation
    • Within 24 hours or ICU admission
  • Exclusion:
    • ICU admission diagnosis of upper gastrointestinal bleeding
  • 26,771 patients included in primary analysis out of 26,982 patients randomised
  • Intervention & control group well matched at baseline
  • Mean age 58 years
  • Mean APACHE II score = 19
  • Study used existing registries to gather data, making it feasible to conduct such a large trial within a reasonable budget


  • Proton pump inhibitor
    • Use of PPI as default stress-ulcer prophylaxis for the unit
    • Crossover: 5% of patients in PPI group were given H2RB
    • Median time available for exposure to stress ulcer prophylaxis was 2.7 days


  • Histamine-2 Receptor Blocker
    • Use of H2RB as default stress-ulcer prophylaxis for the unit
    • Crossover: 20% of patients in H2RB group were given PPI
    • Median time available for exposure to stress ulcer prophylaxis was 2.7 days

Management common to both groups

  • Requirement for stress ulcer prophylaxis in individual patients was as the discretion of the treating clinicians
  • Patients who were taking long-term PPI or H2RB before admission were to switch to the mandated drug when admitted to the participating ICU
  • Clinicians could also use discretion to override the mandated drug (whether H2RB or PPI)
  • PPI were always used for treatment of any upper GI bleeds that did develop
  • All other therapy was at clinician’s discretion


  • Primary outcome:
    • 90 day in-hospital mortality – no significant difference
      • 18.3% for the PPI group vs. 17.5% for the H2RB
        group (95% CI for risk ratio of PPI vs H2RB was 1.00 – 1.10)
  • Secondary outcomes: comparing PPI vs. H2RB groups
    • Clinically important upper GI bleeding – significantly lower in the PPI group
      • 1.3% vs.  1.8%
      • Relative risk 0.73 (95% CI 0.57 – 0.92)
    • Clostridium difficile infection – no significant difference
      • 0.3% vs. 0.43%
      • Relative risk 0.74 (95% C.I. 0.51-1.09)
    • 90 day in-hospital mortality in patients who had cardiac surgery – significantly increased in PPI group
      • Relative Risk 1.27 (95% CI, 1.04-1.57)
  • Tertiary outcomes:
    • Median hours until removed alive from mechanical ventilation – no significant difference
      • 48 vs. 48
    • Ventilator associated conditions – no significant difference
      •  6.5% vs. 5.8%
      • Relative risk 1.18 (95% C.I. 0.87-1.59)
  • Post-Hoc Analysis:
    • Difference in mortality was mostly found in the group of patients with high illness severity, comparing primary outcome  for PPI vs. H2RB according to patients’ APACHE II scores:
      • APACHE II score 0-13: 3.7% vs. 4.0%, RR 0.92 (95% C.I. 0.77-1.11)
      • APACHE II score 14-17: 7.1% vs. 7.5%, RR 0.96 (95% C.I. 0.86-1.08)
      • APACHE II score 18-23: 17.5% vs. 14.7%, RR 1.15 (95% C.I. 1.05-1.25)
      • APACHE II score 24-61: 45.3% vs. 43.5%, RR 1.05 (95% C.I. 1.00-1.11)

Authors’ Conclusions

  • Among ICU patients requiring mechanical ventilation, a strategy of stress ulcer prophylaxis with use of proton pump inhibitors vs histamine-2 receptor blockers resulted in hospital mortality rates of 18.3% vs 17.5%, respectively, a difference that did not reach the significance threshold. However, study interpretation may be limited by crossover in the use of the assigned medication


  • Highly relevant clinical question, applying to the majority of ventilated ICU patients
  • Worldwide recruitment from a large number of ICUs giving a high level of external validity
  • Highly powered study due to large number of clusters and cluster-cross over design (post-trial recalculation suggested even greater power than original power calculation)
  • The pragmatic design is well suited to answering the question, “what would be the impact of making this drug the default for stress ulcer prophylaxis in our ICU?”
  • If real, the point estimate of 0.8% absolute risk reduction with H2RB vs PPI, although small, would be clinically very significant, given the very large number of patients given this prophylaxis worldwide
  • Although the lower bound of the 95% confidence interval was 1.00, this does still give a clear answer that it is much more likely that mortality is increased with PPI, than reduced with PPI
  • Low risk of selection bias, with good baseline balance (e.g. mean APACHE II score of 18.7 in both groups) and minimal loss to follow up. Sensitivity analysis was performed for missing data using worst-best and best-worse case scenarios


  • There was a large proportion of crossover from the intervention arm to the control arm, particularly when units were randomised to using H2RB as default prophylaxis (20% crossover to PPI). This would probably bias towards finding no difference between groups, but it is impossible to know for certain
  • There was a lack of detailed patient-by-patient data, making it difficult to assess the effect of a drug on individuals (as opposed to the unit-wide effect of implementing the strategy). For instance, data on ventilator associated conditions was only available from the 8 Canadian ICUs
  • Data obtained from registries may be subject to random error
  • Existing databases and the data extracted varied in different jurisdictions, which likely contributed to geographical differences in study findings. For instance, higher rates of GI bleeding identified in Canada compared to Australia and New Zealand may due to more sensitive data collection methods

The Bottom Line

  • Although a default strategy of PPI for stress-ulcer prophylaxis in ICU resulted in significantly lower rates of GI bleed (absolute risk reduced by 0.5%), it was also associated with an increase in 90-day hospital mortality compared with H2RB (absolute risk increased by 0.8%).
  • The 95% confidence interval for the relative risk of mortality included 1.0 (1.00-1.10) so the finding is reported as not “statistically significant”, however it is more likely that PPI prophylaxis increases mortality than reduces mortality compared with H2RB
  • This study supports H2RB as the default agent for stress ulcer prophylaxis in ICU, despite the uncertainty noted above. PPI therapy remains appropriate for patients with evidence of upper GI bleeding

External Links



Summary author: Nick Tarmey
Summary date: 23rd January 2020
Peer-review editor: Chris Nickson @precordialthump

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.