JAMA February 18, 2020

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

Paul J. Young et al. (The PEPTIC Investigators)

Bottom Line

In mechanically ventilated ICU patients, a default strategy of stress ulcer prophylaxis with proton pump inhibitors compared with histamine-2 receptor blockers reduced upper gastrointestinal bleeding but showed a non-significant trend toward increased 90-day in-hospital mortality.

Key Findings

1. In-hospital mortality at 90 days occurred in 18.3% (2,459/13,415) of patients in the PPI group compared to 17.5% (2,333/13,356) in the H2RB group (RR 1.05 [95% CI, 1.00-1.10]; absolute risk difference 0.93 percentage points; P=0.054).

2. Clinically important upper gastrointestinal bleeding was significantly lower in the PPI group (1.3%) compared to the H2RB group (1.8%) (RR 0.73 [95% CI, 0.57-0.92]; absolute risk difference -0.51 percentage points; P=0.009).

3. An estimated 20.1% of patients randomized to the H2RB group crossed over to receive PPIs, while 4.1% of patients in the PPI group received H2RBs.

4. Rates of Clostridioides difficile infection, ICU length of stay, and hospital length of stay did not differ significantly between the two treatment strategies.

Study Design

Design

Cluster-Crossover RCT

Open-Label

Sample

26,828

Patients

Duration

90 days

Median

Setting

50 ICUs, 5 countries

Population Adult ICU patients requiring invasive mechanical ventilation within 24 hours of ICU admission.

Intervention Default ICU strategy of stress ulcer prophylaxis using proton pump inhibitors (PPIs).

Comparator Default ICU strategy of stress ulcer prophylaxis using histamine-2 receptor blockers (H2RBs).

Outcome All-cause in-hospital mortality within 90 days.

Study Limitations

• High rate of protocol crossover, with 20.1% of patients assigned to the H2RB group actually receiving PPIs, which may have diluted any potential differences between the groups.

• Open-label cluster crossover design, which could introduce prescribing bias, although the primary outcome of mortality is objective.

• Clinically important gastrointestinal bleeding was determined by treating clinicians without centralized, blinded adjudication.

• The pragmatic use of registry data meant limited granularity in collecting specific patient-level details and minor side effects.

Clinical Significance

The PEPTIC trial challenges the routine preference for PPIs over H2RBs for stress ulcer prophylaxis in mechanically ventilated ICU patients. Although PPIs were more effective at preventing gastrointestinal bleeding (absolute reduction of 0.51%), this benefit did not improve survival and was accompanied by a concerning, albeit statistically non-significant, signal for increased mortality (absolute increase of 0.93%). Consequently, many clinicians and guidelines now favor H2RBs as the default prophylactic agent in critical illness to avoid the potential harms of profound acid suppression, reserving PPIs for patients with specific, high-risk indications.

Historical Context

Stress ulcer prophylaxis has been a cornerstone of intensive care for decades, historically relying on H2RBs. As PPIs became widely available, their superior acid-suppressing capabilities led to a widespread shift in practice favoring their use. However, emerging observational data raised concerns that profound acid suppression might increase the risk of ventilator-associated pneumonia, C. difficile infection, and overall mortality. The PEPTIC trial was launched as a massive, pragmatic study to definitively compare the safety (mortality) and efficacy of these two ubiquitous drug classes in a real-world ICU population.

Guided Discussion

High-yield insights from every perspective

Med Student

Medical Student

What are the differing mechanisms of action between proton pump inhibitors (PPIs) and histamine-2 receptor antagonists (H2RAs), and why are mechanically ventilated patients at high risk for stress-related mucosal disease?

Key Response

PPIs irreversibly inhibit the H+/K+ ATPase pump in gastric parietal cells, while H2RAs reversibly block the histamine H2 receptors on these cells. Mechanically ventilated ICU patients face extreme physiologic stress, leading to splanchnic hypoperfusion, breakdown of the gastric mucosal barrier, and subsequent stress ulcers, which is why prophylaxis is traditionally utilized.

Resident

Based on the PEPTIC trial findings, how should you approach the choice between a PPI and an H2RA for a newly intubated ICU patient without a prior history of GI bleeding?

Key Response

The trial showed PPIs decrease upper GI bleeding compared to H2RAs (1.3% vs 1.8%) but have a concerning, though statistically borderline, trend toward increased 90-day mortality (18.3% vs 17.5%, p=0.054). For a standard patient without specific high-risk indications for PPIs (like recent peptic ulcer), an H2RA may be preferred to avoid potential unmeasured harms of profound acid suppression, balancing bleeding risk against overall survival.

Fellow

How does the cluster-crossover design of the PEPTIC trial impact the interpretation of the results, particularly regarding the separation of the primary outcome (mortality) and the secondary outcome (clinically important GI bleeding)?

Key Response

A cluster-crossover design randomizes ICUs, not individual patients, to a default strategy. This reflects real-world practice but can introduce crossover bias if protocols are not strictly adhered to or if unit-level case mix changes. The discrepancy between improved bleeding outcomes but a potential mortality increase with PPIs suggests that profound acid suppression might cause systemic harms (e.g., microbiome alteration, pneumonia, or cardiovascular effects) that outweigh the localized benefit of preventing rare GI bleeds.

Attending

Given the borderline significance of the mortality harm signal (p=0.054) for PPIs in the PEPTIC trial, how do we reconcile the drive to prevent clinically obvious complications (GI bleeding) with the potential for obscure, systemic harms that may affect overall survival?

Key Response

This highlights a classic critical care paradigm: intervening to fix a surrogate or specific organ-level problem (bleeding) might induce iatrogenic harm (mortality). As attendings, we must teach that 'more' (stronger acid suppression) is not always better. The NNT to prevent one GI bleed with PPIs is high, while the potential NNH for mortality, even if small, applies to a massive population, arguing for a minimalist approach in patients without compelling indications for PPIs.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD

The PEPTIC trial utilized a cluster-crossover randomization with a waiver of informed consent. What are the statistical and ethical implications of using unit-level default strategies for comparative effectiveness research, particularly concerning the power to detect a 1% absolute difference in mortality?

Key Response

Cluster-crossover designs are highly pragmatic and reduce selection bias, but they suffer from intra-cluster correlation (ICC) which dilutes statistical power compared to individual randomization. The study was powered for a large sample (26,000+ patients) to detect small mortality differences, but the p-value of 0.054 (RR 1.05, 95% CI 1.00-1.10) sits right at the traditional alpha threshold. Methodologically, this raises questions about whether the ICC was accurately estimated a priori and if unmeasured unit-level confounders during the crossover periods influenced the marginal mortality signal.

Journal Editor

In reviewing the PEPTIC trial, how does the lack of strict individual-level protocol enforcement (i.e., using a 'default' ICU strategy where clinicians could override the assigned treatment) affect the validity of the intention-to-treat analysis, and how should this influence the editorial framing of the mortality finding?

Key Response

Since clinicians could prescribe off-protocol, treatment contamination is a major threat. About 20% of patients in the H2RA group received PPIs. This non-adherence biases the results toward the null. A tough reviewer would highlight that the intention-to-treat analysis might actually underestimate the true mortality harm of PPIs. Consequently, the editorial framing must carefully balance the pragmatic nature of the 'default strategy' with the dilution of the treatment effect.

Guideline Committee

The Surviving Sepsis Campaign and other critical care guidelines have historically recommended stress ulcer prophylaxis for mechanically ventilated patients without strongly preferring PPIs over H2RAs. How should the PEPTIC trial's finding of a mortality point estimate of 1.05 (upper CI 1.10) for PPIs influence the strength of recommendation and specific phrasing in the next iteration of SUP guidelines?

Key Response

Guidelines must weigh the certainty of evidence and the balance of benefits and harms. While PPIs clearly reduce GI bleeding, the PEPTIC trial's signal of increased mortality shifts the balance of risks. Guidelines should likely issue a conditional recommendation favoring H2RAs over PPIs for routine SUP in the ICU, reserving PPIs only for patients with specific, high-risk indications (e.g., active bleeding), emphasizing that the value of mortality avoidance outweighs minor bleeding prevention.

Clinical Landscape

Noteworthy Related Trials

1998

Canadian Critical Care Trials Group Trial

n = 1,200 · NEJM

Tested

Ranitidine (H2RA)

Population

Mechanically ventilated ICU patients

Comparator

Sucralfate

Endpoint

Clinically important gastrointestinal bleeding

Key result: Ranitidine significantly reduced the rate of clinically important gastrointestinal bleeding compared to sucralfate, with no significant differences in mortality.

2018

SUP-ICU Trial

n = 3,298 · NEJM

Tested

Pantoprazole 40 mg IV daily

Population

Adult ICU patients at risk for GI bleeding

Comparator

Placebo

Endpoint

90-day mortality

Key result: There was no significant difference in 90-day mortality or clinically important gastrointestinal bleeding between the pantoprazole and placebo groups.

2024

REVISE Trial

n = 4,821 · NEJM

Tested

Pantoprazole

Population

Mechanically ventilated critically ill patients

Comparator

Placebo

Endpoint

Clinically important upper GI bleeding and 90-day mortality

Key result: Pantoprazole resulted in a significantly lower risk of clinically important upper gastrointestinal bleeding than placebo, but 90-day mortality was not significantly different.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis