What the REVISE Trial Changed About Stress Ulcer Prophylaxis

Stress-related mucosal disease can cause upper gastrointestinal bleeding during critical illness. Acid suppression has been routine for mechanically ventilated patients for decades, yet the modern absolute benefit was uncertain because feeding, resuscitation, and intensive-care practice had changed. Concern also persisted that suppressing gastric acid might increase pneumonia or Clostridioides difficile infection.

The international REVISE trial addressed that trade directly. It randomized 4,821 adults receiving invasive mechanical ventilation to intravenous pantoprazole 40 milligrams daily or matching placebo, and pantoprazole substantially reduced clinically important upper gastrointestinal bleeding and did not produce a statistically significant change in 90-day mortality.

Why stress ulcers occur in critical illness#

Shock, vasopressors, inflammation, and impaired mucosal perfusion can disrupt gastric and duodenal defenses. Mechanical ventilation and coagulopathy have historically identified patients at higher risk of clinically important bleeding. Minor occult blood or a small fall in hemoglobin is different from bleeding that causes hypotension, transfusion, intervention, or interruption of care.

Proton pump inhibitors reduce gastric acid secretion. Histamine-2 receptor blockers do the same through a different pathway. Both can reduce bleeding, but acid is also a barrier to ingested organisms. The clinical question is net benefit, not merely whether gastric pH rises. Modern enteral nutrition may lower risk, and clinically important bleeding has become less common; when baseline risk falls, the absolute benefit of treating everyone falls even if relative efficacy remains.

The REVISE design#

The REVISE trial enrolled adults in 68 intensive-care units. Participants were expected to receive invasive mechanical ventilation beyond the calendar day after randomization. They received pantoprazole or placebo during ventilation up to 90 days, with protocol stopping conditions.

Randomization and blinding reduce confounding and differential assessment. The placebo comparison asks whether prophylaxis adds benefit beyond contemporary ICU care. The protocol and prespecified analysis plan make outcome definitions and methods visible before the final results. People with active gastrointestinal bleeding, a clear clinical need for acid suppression, or certain contraindications could not ethically be randomized. This protects participants but limits application to those already needing treatment for another reason.

The primary efficacy result#

Clinically important upper gastrointestinal bleeding occurred in 25 of 2,385 participants assigned pantoprazole, 1.0 percent, and 84 of 2,377 assigned placebo, 3.5 percent. The hazard ratio was 0.30 with a 95 percent confidence interval from 0.19 to 0.47.

The absolute risk reduction was about 2.5 percentage points. Its reciprocal gives a rough number needed to treat of 40 during the study period to prevent one event, although time-to-event analysis, confidence intervals, and variation in baseline risk make the exact NNT context-dependent. That is a clinically meaningful benefit because the outcome required consequential bleeding rather than only occult blood, and the relative reduction was large, while the absolute result shows that most untreated participants did not bleed.

Mortality was not significantly changed#

At 90 days, 696 of 2,390 participants in the pantoprazole group had died, 29.1 percent, compared with 734 of 2,379 in placebo, 30.9 percent. The hazard ratio was 0.94, with a 95 percent confidence interval from 0.85 to 1.04 and P value 0.25.

The interval includes a modest mortality benefit and a small harm. “No significant difference” means the trial did not establish either, not that the treatments are mathematically identical. Mortality in critical illness has many causes, and preventing a relatively uncommon bleed may not shift overall survival detectably. Mortality was the trial's primary safety outcome, which is the right choice, and it is also the reason not to let a reduced bleeding endpoint be rewritten as improved survival.

Patient-important bleeding and other outcomes#

REVISE also reported patient-important upper gastrointestinal bleeding, a broader clinically consequential outcome, and found fewer events with pantoprazole. Other secondary outcomes, including ventilator-associated pneumonia, C difficile infection, renal replacement therapy, duration of ventilation, and ICU or hospital stay, were similar.

Secondary outcomes vary in ascertainment and statistical power. A nonsignificant difference in a rare infection does not prove zero risk. It does make a large common harm less likely within the populations and practices studied.

Competing risks complicate interpretation. A patient who dies early has less time to bleed or acquire infection. Time-to-event methods and prespecified definitions help, but look at the event counts and the follow-up yourself.

How REVISE relates to SUP-ICU#

The SUP-ICU trial randomized 3,298 ICU patients at risk for gastrointestinal bleeding to pantoprazole or placebo. It found no significant difference in 90-day mortality. Clinically important gastrointestinal bleeding occurred in 2.5 percent with pantoprazole and 4.2 percent with placebo.

SUP-ICU and REVISE point in the same general direction: pantoprazole reduces important bleeding without a demonstrated survival effect. REVISE was larger and focused on invasive ventilation, providing a more precise estimate in that group.

The baseline bleeding rate and enrollment criteria differ, so the absolute reductions should not be pooled casually at the bedside; a person with coagulopathy, prior ulcer, shock, or no enteral feeding may have a different baseline risk from the trial average.

PPI versus H2 blocker is another question#

REVISE compared pantoprazole with placebo. It did not compare pantoprazole with an H2 blocker. The PEPTIC trial used a cluster crossover design across ICUs that preferentially used a proton pump inhibitor or an H2 blocker for ventilated patients.

PEPTIC did not show a statistically significant difference in its primary in-hospital mortality outcome at the prespecified threshold. Clinically important bleeding was lower with the PPI strategy. Interpretation is complicated by crossover, open-label cluster practice, and the fact that some patients received the nonpreferred agent. A class comparison cannot be inferred from a placebo comparison. Choice between PPI and H2 blocker includes bleeding efficacy, kidney and liver function, drug interactions, thrombocytopenia concerns, local formulary, and enteral access.

What the updated evidence synthesis adds#

An updated systematic review incorporated REVISE into the broader randomized evidence. Meta-analysis can estimate effects across populations, but differences in risk, feeding, comparator, drug, and outcome definition need attention.

The consistent reduction in clinically important bleeding strengthens confidence that acid suppression has efficacy in higher-risk critically ill patients. Mortality remains essentially unchanged in pooled evidence. Infection effects are less certain, particularly for rare outcomes and local microbiology. Certainty differs by endpoint, so you can hold high confidence in the bleeding reduction and low confidence about a rare adverse effect without contradicting yourself.

What changed in practice interpretation#

Before REVISE, one defensible position was that modern bleeding rates were too low to justify routine PPI prophylaxis during ventilation, and the trial showed an untreated clinically important bleeding risk of 3.5 percent and a sizeable reduction. That supports prophylaxis for patients resembling the trial population.

The result does not support acid suppression for every person admitted to a ward, everyone briefly receiving noninvasive ventilation, or anyone with a mildly elevated illness score. The inclusion criterion and bleeding-risk period define the evidence boundary. It also shifts attention from whether to start toward when to stop. A benefit during invasive ventilation can become unnecessary long-term use after extubation.

Deprescribing at the transition of care#

Stress-ulcer prophylaxis is frequently continued on transfer from ICU and at discharge without a documented indication. Continued therapy may be appropriate for peptic ulcer disease, recent gastrointestinal bleeding, severe erosive esophagitis, selected antithrombotic combinations, or another condition. The reason should be explicit.

At extubation or resolution of critical illness, review the medication. Ask whether your patient was already on a PPI before admission, whether an indication emerged during the stay, and whether a stop order is actually in place. Medication reconciliation should distinguish prophylaxis from treatment. Abrupt discontinuation after long-term outpatient use can produce rebound acid symptoms in some people, but short ICU prophylaxis often does not require a taper. The plan depends on duration and indication.

Avoiding common overinterpretations#

First, the 70 percent relative hazard reduction does not mean 70 percent of patients benefit. The absolute difference was about 25 events per 1,000 treated. Second, the trial did not show that PPIs save lives. Third, it did not establish zero infection risk. Fourth, it did not test oral outpatient pantoprazole for chronic prevention.

The trial also should not be used to deny treatment to someone with active bleeding. Prophylaxis prevents an event in an at-risk person; active gastrointestinal hemorrhage requires resuscitation, diagnosis, and therapeutic management. Those are two different jobs, and the trial speaks to only one of them.

Applying the result at the bedside#

Confirm that your patient is invasively ventilated and lacks an exclusion or a separate treatment indication. Assess bleeding risk, enteral feeding, coagulopathy, prior bleeding, and expected ventilation duration. Choose agent and route using current local guidance.

Monitor for overt bleeding, hemoglobin change, hemodynamics, transfusion, infection, and medicine interactions. Reassess daily. Place a stop or review point at extubation, ICU transfer, and discharge.

The best evidence-based protocol includes initiation and discontinuation. A start-only order set converts a time-limited benefit into avoidable chronic treatment.

References#

1. REVISE randomized trial
2. REVISE protocol and pilot
3. REVISE statistical analysis plan
4. SUP-ICU trial
5. PEPTIC trial
6. Updated systematic review