Pantoprazole in Patients at Risk for Gastrointestinal Bleeding in the ICU

Krag M, Marker S, Perner A et al.

NEJM 2018; 379: 2199-2208

Primary Hypothesis

Amongst adults patients with an increased risk of GI bleeding requiring emergency ICU admission, pantoprazole is associated with a lower rate of GI bleeding but with higher rates of nosocomial infections and MI than placebo.

Design and Location

Large randomised controlled trial conducted in 33 ICUs across six different countries (UK, Denmark, Finland, the Netherlands, Norway and Switzerland) between Jan 2016 to Oct 2017.


Inclusion criteria:

  • >18yo
  • Emergency ICU admission
  • >= 1 risk factors for GI bleeding including shock, use of anticoagulant, renal replacement therapy, mechanical ventilation, history of liver disease and ongoing coagulopathy

Exclusion criteria:

  • Contraindications to PPI
  • Already on PPI/H2 antagonist prior to admission
  • Pregnancy
  • Organ transplant
  • Diagnosis of peptic ulcer disease during current hospital admission
  • Withdrawal of treatment


  • Treatment Group: IV pantoprazole 40mg in 10ml 0.9% NaCl daily from randomisation until ICU discharge/death for maximum 90 days
  • Placebo Group: 10ml of 0.9% NaCl matching treatment group

Primary Outcome

Death by 90-days after randomisation

Secondary Outcomes

Incidence of GI bleed, new pneumonia, C.diff, severe adverse reactions and percentage of days alive without organ-support within the 90-day period.


3350 patients required to have 90% power to detect a difference of 5% points between groups in primary outcome.


3298 patients enrolled, 1645 in pantoprazole and 1653 in placebo group. Similar baseline characteristics.

ICU median stay 6 days, trial agent for median of 4 days in each group.

Primary outcome:

  • Pantoprazole group: 510/1642
  • Placebo group: 499/1640
  • P=0.76, CI 0.91-1.13
  • No heterogeneity in patients with liver disease, coagulopathy, shock, mechanical ventilation or type of ICU admission.

Secondary outcomes:

No P values for secondary outcomes obtained in the study.

  • GI bleeding: 2.5% in pantoprazole vs. 4.2% in placebo (RR 0.58; 95% CI 0.40-0.86)
  • Pneumonia or C. difficile infection: 16.8% vs. 16.9% (RR 0.99; 95% CI 0.84-1.16)
  • Severe adverse reactions: 0% vs. 0%
  • Median percentage of days without organ support: 92% vs. 92% of days


  • No difference between pantoprazole and placebo in 90-day mortality
  • Fewer patients in pantoprazole group had clinically important GIB on ICU but lack of p-value
  • No increase in c.diff and MI in pantoprazole group as anticipated

Stated Limitations from the Study

  • Total number of patients enrolled did not meet target enrollment to detect 5% difference in primary outcome.
  • Clinical diagnosis of GI bleeding could not differentiate source of bleeding (stress ulcer vs other sources of bleeding).
  • Overall low risk populations for GI bleed identified in study- >=one risk factor, NNT=60.
  • Patient feeding status not assessed
  • No serious adverse events reported suggestive of under-reporting

Discussion from Journal Club Meeting

  • Stress ulcer prophylaxis is frequently prescribed on our ICU as most of our ICU emergency admissions are deemed at high risk of GI bleed.
  • We thought the 90-day mortality being the primary outcome of the study is slightly inappropriate as the aim of the study was to determine whether pantoprazole would reduce incidence of GI bleeding rather than mortality.
  • Secondary outcomes did not contain p-values in the study which made the data difficult to interpret the statistical significance.
  • A few elective cases have been included in the case mix without any further explanation as to whether they had emergency ICU admission following elective procedure.
  • This study may benefit from including randitidine as well as PPI.

Summary by Dr M Zou. Journal Club Meeting 24 January 2019.

Comments are closed.

Blog at

Up ↑

%d bloggers like this: