Skip to main content

Foundational Trials

REVISE

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Stress Ulcer Prophylaxis during Invasive Mechanical Ventilation
  • Acronym: REVISE
  • Year: 2024
  • Journal published in: The New England Journal of Medicine
  • Citation: Cook D, Deane A, Lauzier F, Zytaruk N, Guyatt G, Saunders L, et al; REVISE Investigators. Stress ulcer prophylaxis during invasive mechanical ventilation. N Engl J Med. 2024 Jul 4;391(1):9-20.

Context & Rationale

  • Background
    • Stress-related mucosal disease can cause upper gastrointestinal bleeding in critical illness; historically this was clinically consequential but contemporary incidence has fallen with modern ICU practices (earlier enteral nutrition, improved haemodynamic resuscitation, lower ventilator days, and changes in sedation and anticoagulation).
    • Acid suppression (PPIs or histamine-2 receptor antagonists) reduces gastric acidity and can reduce bleeding, but may plausibly increase infections (e.g., aspiration pneumonia, Clostridioides difficile) via gastric colonisation and altered microbiota.
    • In placebo-controlled critical care RCTs before REVISE, pantoprazole reduced clinically important gastrointestinal bleeding but effects on mortality and infectious harms remained uncertain, in part because these outcomes were less frequent and trials were variably powered for them.1
    • Comparative effectiveness evidence (PPI vs histamine-2 receptor antagonists) was debated after a large cluster crossover trial that did not show a clear mortality difference overall, but did not resolve concerns about net benefit in the lowest bleeding-risk patients and could not completely exclude differential harms in key subgroups.2
    • Guideline recommendations before REVISE generally supported pharmacologic prophylaxis for patients with higher bleeding risk, but were conditional and explicitly emphasised the importance of targeting therapy rather than routine universal use in all ICU patients.3
  • Research Question/Hypothesis
    • Does stress ulcer prophylaxis with pantoprazole, compared with placebo, reduce clinically important upper gastrointestinal bleeding among invasively mechanically ventilated ICU adults?
    • Does pantoprazole increase 90-day all-cause mortality (a safety endpoint) in this population?
    • Do key putative harms (ventilator-associated pneumonia, C. difficile infection, kidney injury requiring renal replacement therapy) differ between strategies?
  • Why This Matters
    • Stress ulcer prophylaxis is common, yet the absolute benefit depends heavily on baseline bleeding risk; overuse risks exposing large numbers of low-risk patients to uncertain harms.
    • A definitive, contemporary, double-blind placebo-controlled trial was needed to quantify both bleeding prevention and mortality safety in the modern ICU context, across multiple health systems.
    • Clinicians and guideline developers required higher certainty about whether PPIs should be routine in ventilated ICU patients, and if so, which patients derive meaningful net benefit.

Design & Methods

  • Research Question:
    • In invasively mechanically ventilated adult ICU patients, does intravenous pantoprazole (40 mg once daily) compared with placebo reduce clinically important upper gastrointestinal bleeding during ICU care, without increasing 90-day all-cause mortality?
  • Study Type:
    • International, multicentre, randomised, double-blind, placebo-controlled, parallel-group pragmatic trial conducted in 68 ICUs across 8 countries.
    • Investigator-initiated, with central web-based randomisation and blinded outcome adjudication.
    • Protocol and statistical analysis plan were prospectively published.45
  • Population:
    • Setting: Adult ICUs.
    • Inclusion: Adults (≥18 years) receiving invasive mechanical ventilation and expected to remain ventilated beyond the calendar day after randomisation.
    • Key exclusions: Invasive mechanical ventilation >72 hours during the admission; active gastrointestinal bleeding or high-risk bleeding indication requiring acid suppression (e.g., recent peptic ulcer bleed); receipt in ICU of >1 daily dose equivalent of PPI or histamine-2 receptor antagonist; dual antiplatelet therapy; combined antiplatelet agent plus therapeutic anticoagulation; contraindication to pantoprazole; palliative care/anticipated withdrawal of life-sustaining therapy; pregnancy; prior enrolment/prohibited co-enrolment; patient/proxy/physician refusal.
  • Intervention:
    • Pantoprazole 40 mg intravenously once daily, started soon after randomisation and continued while invasively mechanically ventilated in ICU (stopping when ventilation ceased and not restarted, death, withdrawal, or end of follow-up as prespecified).
    • Open-label acid suppression was discouraged while receiving study drug unless clinically indicated.
  • Comparison:
    • Matching placebo intravenously once daily, with identical administration and stopping rules.
    • Open-label acid suppression was discouraged unless clinically indicated (same as intervention group).
  • Blinding:
    • Double-blind (participants, clinicians, research staff, adjudicators, and analysts).
    • Study drug preparation processes were designed to preserve masking across sites and regions.
  • Statistics:
    • Power calculation: A total sample size of 4,800 was planned to detect a 1.5 percentage point absolute reduction in clinically important upper gastrointestinal bleeding (from 3.0% to 1.5%) with 85% power at a two-sided 5% significance level.
    • Analysis: Modified intention-to-treat (excluding participants who withdrew consent post-randomisation as required); time-to-event methods (Cox proportional-hazards models) for primary endpoints; secondary outcomes analysed with multiplicity-adjusted p values (Holm method) as prespecified.
  • Follow-Up Period:
    • 90 days after randomisation for survival status; in-ICU follow-up for bleeding and several secondary endpoints.

Key Results

This trial was not stopped early. Enrolment completed after achieving the prespecified sample size; one interim analysis (n=2400) did not trigger stopping.

Outcome Pantoprazole Placebo Effect p value / 95% CI Notes
Clinically important upper gastrointestinal bleeding (primary efficacy) 25/2385 (1.0%) 84/2377 (3.5%) HR 0.30 95% CI 0.19 to 0.47; P<0.001 Absolute difference 2.5 percentage points (95% CI 1.6 to 3.3)
90-day all-cause mortality (primary safety) 696/2390 (29.1%) 734/2379 (30.9%) HR 0.94 95% CI 0.85 to 1.04; P=0.25 Model adjusted for baseline APACHE II score
Patient-important upper gastrointestinal bleeding (secondary) 36/2385 (1.5%) 100/2377 (4.2%) HR 0.36 95% CI 0.25 to 0.53; P<0.001 P values for secondary outcomes were multiplicity-adjusted
Ventilator-associated pneumonia in ICU (secondary) 556/2385 (23.3%) 567/2377 (23.8%) HR 1.00 95% CI 0.89 to 1.12; P=0.93 Defined by Clinical Pulmonary Infection Score ≥6
Clostridioides difficile infection in hospital (secondary) 28/2385 (1.2%) 16/2377 (0.7%) HR 1.78 95% CI 0.96 to 3.29; P=0.50 Low event counts; multiplicity-adjusted p value
Initiation of renal replacement therapy in ICU (secondary) 146/2385 (6.1%) 142/2377 (6.0%) HR 1.04 95% CI 0.83 to 1.31; P=0.98 Multiplicity-adjusted p value
Total units of red blood cells transfused in ICU (tertiary) 4850 total (2.3/100 patient-days) 5521 total (2.6/100 patient-days) RR 0.87 95% CI 0.74 to 1.02; P=0.51 Negative binomial model; multiplicity-adjusted p value
Peak serum creatinine in ICU (mg/dL) (tertiary) 1.8 (IQR 1.0–3.4) 1.8 (IQR 1.0–3.4) Difference 0.03 95% CI −0.08 to 0.15; P=0.91 Multiplicity-adjusted p value
Duration of invasive mechanical ventilation in ICU (days) (tertiary) 6.0 (IQR 3.0–11.0) 6.0 (IQR 3.0–11.0) Difference 0.0 95% CI −0.5 to 0.6; P=0.73 Multiplicity-adjusted p value
Duration of ICU stay (days) (tertiary) 10.0 (IQR 6.0–17.0) 10.0 (IQR 6.0–17.0) Difference 0.1 95% CI −0.3 to 0.5; P=0.48 Multiplicity-adjusted p value
Duration of hospital stay (days) (tertiary) 20.0 (IQR 12.0–36.0) 20.0 (IQR 12.0–37.0) Difference 0.2 95% CI −0.2 to 0.7; P=0.47 Multiplicity-adjusted p value
  • Pantoprazole substantially reduced clinically important upper gastrointestinal bleeding (HR 0.30; 95% CI 0.19 to 0.47) and reduced patient-important bleeding (HR 0.36; 95% CI 0.25 to 0.53), with an absolute reduction of 2.5 percentage points for the primary bleeding outcome.
  • There was no statistically significant difference in 90-day mortality (HR 0.94; 95% CI 0.85 to 1.04); prespecified severity subgroup estimates were HR 0.85 (95% CI 0.74 to 0.98) for APACHE II <25 vs HR 1.04 (95% CI 0.89 to 1.20) for APACHE II ≥25 (interaction P=0.27).
  • Secondary “harm” outcomes were broadly similar (ventilator-associated pneumonia HR 1.00; 95% CI 0.89 to 1.12), but C. difficile infection was numerically higher with pantoprazole (28 vs 16 events; HR 1.78; 95% CI 0.96 to 3.29) with low event counts and multiplicity-adjusted P=0.50.

Internal Validity

  • Randomisation and Allocation:
    • Central web-based 1:1 randomisation with allocation concealment.
    • Stratified by centre and pre-hospital acid suppression; variable block sizes (reducing predictability and selection bias).
  • Drop out or exclusions (post-randomisation):
    • 4,900 randomised; 79 (1.6%) excluded post-randomisation (36 pantoprazole; 43 placebo), primarily because consent was declined/withdrawn after randomisation, plus a small number deemed ineligible after randomisation.
    • Vital status at 90 days was obtained for 4,769/4,821 (98.9%) included participants.
  • Performance/Detection Bias (blinding feasibility):
    • Double-blind design; masking extended to clinicians, participants, outcome adjudicators, and analysts.
    • No reported requests for unblinding.
  • Protocol Adherence and Separation of the Variable of Interest:
    • Study drug started within 24 hours of screening in 2,260/2,417 (93.5%) vs 2,230/2,404 (92.8%).
    • Study drug administered for a median of 86.7% (IQR 63.4–97.5) vs 84.6% (IQR 57.1–97.2) of ventilator days.
    • Received study drug for ≥80% of ventilator days: 1,802/2,417 (74.6%) vs 1,684/2,404 (70.0%).
    • Open-label PPI use while ventilated: 107/2,417 (4.4%) vs 129/2,404 (5.4%); open-label histamine-2 receptor antagonist use: 20/2,417 (0.8%) vs 32/2,404 (1.3%).
    • Median duration of invasive ventilation was identical: 6.0 (IQR 3.0–11.0) days in both groups.
  • Baseline Characteristics:
    • Groups were well balanced: mean age 58.4 vs 58.0 years; mean APACHE II 21.7 vs 21.7; female sex 36.5% vs 36.2%.
    • Key exposures that can influence bleeding risk were similar: pre-hospital acid suppression 23.6% vs 23.7%; prophylactic anticoagulation in ICU 62.9% vs 62.5%; enteral nutrition in ICU 92.0% vs 91.8%.
  • Heterogeneity:
    • International, multicentre setting increases clinical heterogeneity; treatment effects for primary outcomes were broadly consistent across prespecified subgroups.
  • Timing and Dose:
    • Early initiation was achieved in most participants (as above); pantoprazole dose (40 mg IV daily) reflects standard prophylactic dosing in many ICUs.
  • Outcome Assessment:
    • Primary bleeding outcome was adjudicated and incorporated haemodynamic/transfusion consequences (clinically important bleeding), improving objectivity and clinical relevance.
    • Ventilator-associated pneumonia definition relied on a Clinical Pulmonary Infection Score threshold, which is less objective than microbiology-confirmed pneumonia and may misclassify some cases.
  • Statistical Rigor:
    • Prespecified dual primary endpoints (efficacy and safety) with time-to-event analyses and an interim analysis plan.
    • Multiplicity-adjustment for secondary outcomes reduces false-positive inference but decreases power for uncommon harms (notably C. difficile).

Conclusion on Internal Validity: Strong overall: allocation concealment and blinding were robust, groups were balanced, protocol delivery achieved good separation with low open-label crossover, and follow-up was high; the main internal validity constraint is the small post-randomisation exclusion fraction and limited precision for rare harms.

External Validity

  • Population Representativeness:
    • Enrolled a broad adult mechanically ventilated ICU population across 8 countries with common admission diagnoses (sepsis, trauma, cardiac arrest, pneumonia) and substantial organ support use.
    • Mean APACHE II ~22 suggests moderate-to-high acuity typical of many tertiary ICUs.
  • Key Exclusions and Applicability:
    • Findings apply best to ventilated ICU adults without a clear pre-existing indication for acid suppression and without dual antiplatelet therapy or combined antiplatelet plus therapeutic anticoagulation.
    • Intravenous pantoprazole is widely available in high-resource systems; external validity is lower where IV PPI supply, pharmacy preparation, or infection control diagnostics are constrained.
    • Because the baseline bleeding risk in modern ICUs varies, the absolute benefit will be larger in higher-risk case-mix settings and smaller in low-risk settings (even if relative effects are similar).

Conclusion on External Validity: Generalisability is high for contemporary adult ventilated ICU practice in similar health systems, but is more limited for populations with mandated acid suppression indications, dual antiplatelet/therapeutic anticoagulation combinations, and settings with markedly different baseline bleeding risk or resource constraints.

Strengths & Limitations

  • Strengths:
    • Large, international, double-blind, placebo-controlled design with pragmatic bedside delivery and broad inclusion.
    • Dual primary endpoints explicitly addressed both efficacy (clinically important bleeding) and safety (90-day mortality).
    • Clinically and patient-centred bleeding outcomes: both “clinically important” and “patient-important” bleeding were captured and showed concordant benefit.
    • High follow-up for 90-day mortality and prespecified multiplicity handling for secondary endpoints.
  • Limitations:
    • Post-randomisation exclusions due to consent withdrawal (and a small number of ineligible participants) resulted in a modified intention-to-treat analysis.
    • Rare harms (notably C. difficile) had low event counts; confidence intervals do not exclude clinically important increases or decreases, and multiplicity adjustment further reduces power.
    • Ventilator-associated pneumonia ascertainment used a score-based definition (CPIS ≥6), which may be insensitive/specific compared with adjudicated microbiology-based criteria.
    • Findings are less directly applicable to patients with strong acid-suppression indications (e.g., recent peptic ulcer bleed) or higher bleeding-risk antithrombotic regimens excluded from enrolment.

Interpretation & Why It Matters

  • Bleeding prevention is real and clinically relevant
    • Pantoprazole reduced clinically important bleeding (1.0% vs 3.5%) and patient-important bleeding (1.5% vs 4.2%), supporting that stress ulcer prophylaxis still prevents consequential events in a modern ventilated ICU population.
  • Mortality safety signal was not observed
    • 90-day mortality was similar (29.1% vs 30.9%; HR 0.94), including no evidence of excess mortality in the highest acuity subgroup (APACHE II ≥25 HR 1.04; interaction P=0.27).
  • Harms remain a “rare event” problem
    • Infectious harms were uncommon; the numerically higher C. difficile incidence (1.2% vs 0.7%) highlights that the net benefit question is still risk-stratified and that low-risk patients may experience minimal absolute benefit.

Controversies & Subsequent Evidence

  • Pre-REVISE evidence established that pantoprazole reduces clinically important gastrointestinal bleeding versus placebo, but left uncertainty about mortality and infectious harms because those outcomes were less frequent and effect estimates imprecise in individual trials.1
  • Practice variation persisted because comparative evidence (PPI vs histamine-2 receptor antagonist) did not fully resolve whether PPIs were preferable for all ventilated patients or only for those at higher baseline bleeding risk, and because plausible infection mechanisms created concern about routine universal PPI exposure in low-risk patients.2
  • Guideline methodology before REVISE explicitly framed prophylaxis as a risk-targeted intervention (rather than universal), in part because the balance of benefit and harm changes sharply as baseline bleeding risk falls.3
  • REVISE’s design choices (dual primary endpoints; prespecified analyses; contemporary definitions for clinically important and patient-important bleeding) were anchored in a published protocol and statistical analysis plan, strengthening interpretability and limiting analytical flexibility.45
  • The accompanying editorial highlighted the tension between “uncertain answers” for rare harms and the tangible prevention of bleeding events, and reinforced the practical imperative to avoid unnecessary continuation of PPIs once ventilation-related bleeding risk resolves.6
  • Post-trial synthesis in network meta-analysis frameworks supports a consistent reduction in stress-ulcer-related bleeding with acid suppression in critical illness, while emphasising ongoing uncertainty for uncommon harms and the need for risk-based deployment rather than automatic universal use.7
  • Updated SCCM/ASHP guidelines (post-REVISE) recommend stress ulcer prophylaxis for critically ill adults at elevated bleeding risk and emphasise reassessment and discontinuation when risk factors abate, aligning implementation with a targeted net-benefit framework.8

Summary

  • In invasively mechanically ventilated ICU adults, pantoprazole reduced clinically important upper gastrointestinal bleeding from 3.5% to 1.0% (HR 0.30; 95% CI 0.19 to 0.47).
  • There was no significant difference in 90-day all-cause mortality (29.1% vs 30.9%; HR 0.94; 95% CI 0.85 to 1.04).
  • Patient-important bleeding was also reduced (1.5% vs 4.2%; HR 0.36; 95% CI 0.25 to 0.53), supporting a clinically meaningful downstream benefit.
  • Ventilator-associated pneumonia rates were similar; C. difficile infection was numerically higher with pantoprazole but uncommon and statistically inconclusive after multiplicity adjustment.
  • Trial conduct demonstrated strong internal validity (concealed randomisation, robust blinding, good protocol adherence and separation), with precision limits chiefly for rare harms.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Secondary and tertiary outcomes in the main trial publication used multiplicity-adjusted p values (Holm method), so p values for uncommon harms (e.g., C. difficile) are conservative; confidence intervals remain central for clinical inference.

Overall Takeaway

REVISE is a contemporary, international, double-blind placebo-controlled evaluation of stress ulcer prophylaxis in ventilated ICU adults, demonstrating a clear reduction in clinically important and patient-important upper gastrointestinal bleeding without evidence of increased 90-day mortality. Its principal contribution is shifting the evidence base from assumption-driven routine prophylaxis to quantifiable benefit with persisting uncertainty for rare harms, supporting a risk-targeted approach with active discontinuation when bleeding risk resolves.

Overall Summary

  • Pantoprazole reduced clinically important upper GI bleeding (HR 0.30) and patient-important bleeding (HR 0.36) versus placebo in ventilated ICU adults.
  • No significant difference in 90-day mortality (HR 0.94) and no excess mortality in the highest severity prespecified subgroup.
  • Rare harms (especially C. difficile) remain the key uncertainty; implementation should be risk-based with prompt cessation when risk abates.

Bibliography