Current issues of ACP Journal Club are published in Annals of Internal Medicine


Prophylaxis of stress ulcer bleeding: a meta-analysis

ACP J Club. 1992 Jan-Feb;116:9. doi:10.7326/ACPJC-1992-116-1-009

Source Citation

Tryba M. Prophylaxis of stress ulcer bleeding. A meta-analysis. J Clin Gastroenterol. 1991 Jul;13(Suppl 2):S44-55.



To evaluate the efficacy of antacids, H2-antagonists, pirenzepine, and sucralfate in preventing stress ulcer bleeding and to compare the incidence of pneumonia and death in patients receiving these agents.

Data sources

A computerized bibliographic database search; a manual search for original articles about stress bleeding prophylaxis published before April 1990, in German, English, French, Italian, and Spanish; and requests to pharmaceutical companies.

Study selection

Randomized, controlled trials of ≥ 10 adult patients in intensive care units that evaluated stress-ulcer prophylactic agents and measured macroscopic bleeding frequency were included. 44 of 66 studies identified were selected.

Data extraction

Data on drug regimens, bleeding events, and pneumonia and mortality rates were extracted. Data were pooled for comparisons of agents.

Main results

Prophylaxis with antacids reduced the rate of macroscopic bleeds compared with no prophylaxis (n = 607 patients; 6% vs. 15% for occurrence of bleeding, respectively; typical [common] odds ratio [COR], 0.34; 95% CI, 0.20 to 0.59). When cimetidine or ranitidine was compared with no treatment (n = 1405), H2 antagonists had lower rates of bleeding (5% vs. 16%; COR, 0.29; CI, 0.20 to 0.41). There was no overall difference in event rates between prophylaxis with antacids and antagonists (COR, 0.82; CI, 0.50 to 1.35). Pirenzepine reduced bleeding frequency compared with no treatment (n = 184; 4% vs. 24%) and compared with H2-antagonists (n = 928; 5% vs. 13%; COR, 0.33; CI, 0.20 to 0.53). Sucralfate had lower rates of bleeding in comparison with H2-antagonists (n = 563; 4% vs. 10%; COR, 0.38; CI, 0.19 to 0.76) and in comparison with antacids (n = 962; 3% vs. 4%; COR, 0.58; CI, 0.28 to 1.19), although the latter was not statistically significant. All studies comparing sucralfate with antacids or H2-antagonists found a lower risk for pneumonia with sucralfate (n = 488; 18% vs. 38% occurrence of pneumonia; COR, 0.37; CI, 0.25 to 0.55). Mortality rates for patients receiving H2-antagonists or antacids were similar to control rates (n = 1290; 16% vs. 19%; COR, 0.94; CI, 0.69 to 1.30). 9 studies including 882 patients, at least half of whom were ventilated, had fewer deaths associated with sucralfate (24%) than with antacids or H2-antagonists (33%; COR, 0.67; CI, 0.51 to 0.87).


Antacids, H2-antagonists, and pirenzepine reduced bleeding in comparison with placebo. Pirenzepine and sucralfate were more effective than H2-antagonists in preventing bleeding. Antacids and H2-antagonists were associated with higher rates of pneumonia and death than was sucralfate.

Source of funding: No external funding.

Address for article reprint: Dr. M. Tryba, Department of Anesthesiology, Intensive Care Medicine, and Pain Therapy, University of Bochum, Bergmannsheil, Gilsingstrasse 14, D-4630 Bochum 1, Germany.


Gastric pH and Nosocomial Pneumonia: A Meta-analysis

Stress ulcer bleeding has been found to affect mainly patients on mechanical ventilation or with coagulopathy or the multiple organ system failure syndrome. For these conditions, the pathophysiology of stress ulceration and resultant gastrointestinal (GI) bleeding may not be easily amenable to prophylaxis. Worse still, there is now concern about increased gastric microbial growth associated with alteration of gastric pH. A consequence of gastric microbial colonization is transmission of organisms to the tracheobronchial tree by aspiration, leading to nosocomial pneumonia.

Parallel to the elucidation of the mechanisms involved in nosocomial pneumonia, important advances have been made in its diagnosis in mechanically ventilated patients. In the past, nosocomial pneumonia has been diagnosed clinically with criteria of fever, leukocytosis, purulent sputum (usually with pathogens isolated by culture), and changing infiltrates on the chest radiograph. Recent studies have definitively shown that these clinical criteria are not accurate in ventilated patients (1). Accurate diagnosis requires quantitative cultures of the protected specimen brush (PSB) or bronchoalveolar lavage (BAL) in diagnosing nosocomial pneumonia (2, 3).

What has this to do with the reviewed articles? Both articles are meta-analyses relating to the efficacy and consequences of stress ulcer prophylaxis, mainly in mechanically ventilated patients. Meta-analysis is an increasingly used tool that applies statistical principles to the evaluation of quantitative results of study outcomes. These two studies nicely demonstrate what I consider to be the "up" side and "down" side of this type of analysis. Meta-analysis is often helpful when the results can indeed be quantitated. Tryba's review of 40 studies indicates that any form of prophylaxis is better than nothing, and sucralfate appears to be better than antacids or H2-antagonists for pneumonia and mortality. Cook and associates also analyzed the effect of drugs for stress ulcer prophylaxis on the development of nosocomial pneumonia. These authors have extensive experience in this type of analysis and the abstracted article reflects this experience. Unfortunately for both meta-analyses, nosocomial pneumonia was diagnosed clinically in all the studies reviewed. Therefore, although the methods of the meta-analyses may be excellent, the studies they review are flawed. The clinical diagnosis of nosocomial pneumonia is an unacceptable substitute for diagnosis using quantitative cultures of PSB or BAL. It is impossible, in my opinion, to either verify or refute the conclusions of the meta-analyses because the true frequency of nosocomial pneumonia cannot be found in the reviewed data. Therefore, I feel strongly that clinicians should not let fear of inducing nosocomial pneumonia affect decisions about stress ulcer prophylaxis on the basis of these reviews.

I heartily agree with Cook and colleagues that further information regarding the effect of stress ulcer medication on the development of nosocomial pneumonia should include studies in which appropriate diagnostic measures are used.*

Susan K. Pingleton, MD
The University of Kansas Medical Center Kansas City, Kansas

*Inaccurate diagnosis might have obscured the relationship between prophylaxis and pneumonia in the study by Cook and colleagues, but, in reply to Dr. Pingleton's commentary, Dr. Tryba points out that this is not a plausible explanation for better efficacy of sucralfate in these studies for either pneumonia or death. Thus, sucralfate may be the preferred form of stress ulcer prophylaxis.—The Editor