In summary, for patients who received proton pump inhibitors:

Benefits in NNT

  • 1.6% lower risk of clinically important gastrointestinal bleeding (compared to placebo)
  • No one was helped (no death was prevented)
  • 1 in 63 were helped (clinically important gastrointestinal bleeding was prevented) when compared to placebo
  • No one was helped (no death was prevented) when compared to placebo

Harms in NNT

    No evidence of harm (compared to placebo)
    No evidence of harm (compared to placebo)

View As:

Source: Alhazzani W, Alshamsi F, Belley-Cote E, et al. Correction to: Efficacy and safety of stress ulcer prophylaxis in critically ill patients: a network meta-analysis of randomized trials. Intensive care medicine 2018;44:277-8.

Alhazzani W, Alshamsi F, Belley-Cote E, et al. Efficacy and safety of stress ulcer prophylaxis in critically ill patients: a network meta-analysis of randomized trials. Intensive care medicine 2018;44:1-11.

Study Population: 7293 critically ill adult patients from 57 trials

Efficacy Endpoints: Prevention of clinically important gastrointestinal bleeding and all-cause mortality

Harm Endpoints: Pneumonia and Clostridium difficile infection (the study was unable to state a network estimate for this outcome)

Narrative: Stress ulceration is a term coined to explain a form of upper gastrointestinal (GI) bleeding seen in critically ill patients, who are commonly defined as individuals admitted to an intensive care unit. Stress ulceration was first noted in the 1960s when a series of post-mortem examinations performed on critically ill individuals revealed gastric mucosal lesions.1 Proposed mechanisms for the formation of stress ulceration include reduced blood flow in the vicinity of ulcer formation, ischemia of the gastrointestinal mucosa, and reperfusion injury.1 Earlier studies indicated that overt GI bleeding occurred in up to 25% of critically ill patients and more recent literature estimates the incidence of clinically important GI bleeding to be in the range of 1 to 4%.1, 2

The network meta-analysis discussed here3, 4 was designed to evaluate the safety and efficacy of medications commonly used for stress ulcer prophylaxis in critically ill patients with a focus on the prevention of clinically important and overt GI bleeding, all-cause mortality, the development of pneumonia, and the development of Clostridium difficile infections.

This network meta-analysis focused on both clinically important GI bleeding and overt GI bleeding.3, 4 A network meta-analysis differs from a conventional meta-analysis in that therapies that have not been directly compared with each other in randomized controlled trials could be compared (direct and indirect comparison). Clinically important GI bleeding was defined as “evidence of upper GI bleeding with any of the following: significant hemodynamic changes not explained by other causes, need for transfusion of more than two units of blood, significant decrease in hemoglobin level, evidence of bleeding on GI endoscopy, or need for surgery to control the bleeding.”3, 4 Overt GI bleeding was defined as having coffee-ground emesis or aspirate, melena, hematemesis, or hematochezia.3, 4 However, in this evidence-based review, we only report the efficacy data for clinically important bleeding.

When compared to placebo, the use of PPIs in critically ill patients was associated with a significant reduction in the risk of clinically important gastrointestinal bleeding in this network meta-analysis (Odds ratio [OR]: 0.24, 95% confidence interval [CI] 0.10 – 0.60; Absolute risk difference [ARD]: 1.6%; Number needed- to treat [NNT]: 63; moderate quality evidence).3, 4 However, the administration of PPIs was not associated with a significant reduction in all-cause mortality.

This network meta-analysis also compared other therapies including histamine-2-receptor antagonists (H2RA) and sucralfate with PPIs. The analysis showed the following results for reduction in clinically significant gastrointestinal bleeding: PPI vs H2RA [OR 0.38, 95% CI 0.20 – 0.73; NNT 125; moderate quality evidence] and PPI vs sucralfate [OR 0.30, 95% CI 0.13 – 0.69; NNT 83; moderate quality evidence].3, 4 In this network analysis, PPIs were ranked first for the prevention of clinically important GI bleeding.

The authors of this study mentioned that the use of PPIs in critically ill patients is likely to be associated with a higher risk of developing pneumonia. PPIs were associated with a statistically significant increase in developing pneumonia when compared to sucralfate [OR 1.65, 95% CI 1.20 – 2.27; NNH 28].3, 4 In addition, the data trended towards a higher risk of developing pneumonia when comparing PPIs to H2RAs and to placebo but this data is not statistically significant as the associated confidence intervals cross 1.0.3,4 The study authors’ expressed their concern regarding an increased risk of developing pneumonia in critically ill patients treated with PPIs by citing a potential 3.1 % absolute increase.3, 4 Nevertheless, the presented data does not support an increased risk of developing pneumonia when comparing PPIs to H2RAs or to placebo.

In this study, the data pertaining to the risk of developing a C. difficile infection was not sufficient for analysis.3,4 Though this study included a total of 57 trials, only one of them reported C. difficile infections so a network estimate for this outcome could not be made.3, 4

Overall, this large network meta-analysis found moderate quality evidence that prophylaxis with PPIs or H2RAs reduced clinically important GI bleeding, when compared to no prophylaxis.3, 4 Additionally, PPIs appear to be more efficacious than both H2RAs and sucralfate in terms of reducing clinically important gastrointestinal bleeding. In this study, PPIs were associated with an increased chance of developing pneumonia when compared to sucralfate but not when compared to H2RAs or placebo. In addition, there was insufficient data to evaluate the risk of developing a C. difficile infection. The PPIs used in the studies included in this network meta-analysis were omeprazole, pantoprazole, esomeprazole, rabeprazole, and lansoprazole and these medication were administered intravenously, orally, or via nasogastric tube.

A recent study published in the New England Journal of Medicine studied the use of pantoprazole versus placebo in adult intensive care unit patients at risk for GI bleeding.5 This study concluded that the number of clinically important events (myocardial ischemia, clinically important GI bleeding, pneumonia, and Clostridium difficile infection) and the 90 day mortality (primary outcome of the study) were similar between the study and the control groups.5 An editorial in the same journal maintained that prophylaxis with a PPI should be reserved for patients that are significantly ill and are at exceedingly high risk of deterioration from clinically significant GI bleeding.6

Caveats: Although a total of 57 trials were included in the primary meta-analysis by Alhazzani et al.,3, 4 only 16 trials had a low risk of bias. Of the remaining trials, 30 had a high risk of bias and 11 had an unclear risk of bias. However, the overall quality of the evidence was rated as moderate.

The categorization and diagnosis of pneumonia in this study was a limiting factor as well.3, 4 The majority of the included trials used a combination of clinical, radiographic, and microbiological criteria to diagnose pneumonia, but no standard was set and the definition of pneumonia varied among the studies. Also, some of the trials that analyzed sucralfate predated the widespread use of pneumonia prevention stategies in critically ill patients, thus casting doubt on the validity of the results in the context of current ICU management.

An additional caveat of this study relates to Clostridium difficile infections (CDIs).3, 4 Recent studies have suggested that PPIs are associated with an increased risk of Clostridium difficile infection.7 The primary study only had one trial that included CDIs, and thus was unable to state a network estimate for this outcome. Of the 214 randomized patients in this single trial, only one patient developed a C. difficile infection which indicates that there is insufficient power to address this topic.8

Another limiting factor of this meta-analysis is the heterogenitey in defining GI bleeding, as each study had different inclusion criteria. It is also important to note that study subjects included in this network meta-analysis likely had different baseline risks for GI bleeding. Although this network meta-analysis includes a variety of critically ill patients in various intensive care settings, the efficacy of the PPIs might be different in higher risk subgroups of patients.

In conclusion, administration of PPIs for stress ulcer prophylaxis in critically ill patients is associated with reduced clinically important upper gastrointestinal bleeding without improving survival. Due to the aforementioned, we are assigning a yellow (unclear if benefits outweigh harms) color recommendation. Additional high quality studies are needed which compare the effects of these agents in order to develop more robust evidence regarding the efficancy and harm endopoints discussed.

The original manuscript was published in Academic Emergency Medicine as part of the partnership between TheNNT.com and AEM.

Author: Justin Putnam, MD; Allan B. Wolfson, MD
Supervising Editors: Shahriar Zehtabchi, MD

Published/Updated: January 17, 2020

  1. The Title Bar

    The title bar is color-coded with our overall recommendation.

    • Green: Benefits outweigh risks.
    • Yellow: Unclear risk/benefit profile.
    • Red: Benefits do not outweigh risks.
    • Black: Obvious harms, no clear benefits.
  2. Tip content...