I
]REVIEW
Stress Ulcer Prophylaxis in the Critically I11:A Meta-Analysis DEBORAH J. COOK, M.D., Hamilton, Ontario, Canada, LANA G. WlTT, Ph.D., Stanford, California, RICHARD J. COOK, M.Sc., GORDON H. GUYATT, M.D., Hamilton, Ontario, Canada
PURPOSF~To e~rnmine the differential effect of stress ulcer prophylaxis on overt bleeding, clinically important bleeding, and mortality in critically ill patients. DATA IDENTIFICATION:Computerized bibliographic search of published and unpublished research. STUDY SELECTION:Independent review of 168 articles identified 42 relevant randomized trials for inclusion. DATAABSTRACTION:The validity, population, intervention, and outcomes of each trial were evaluated. RESULTS: Stress ulcer prophyl~xi~ with antacids (odds ratio 0.40 [95% confidence interval (CI) 0.20 to 0.79]) or hist~mlne-2-receptor antagonists (odds ratio 0~9 [95% CI 0.17 to 0.45]) decreases the incidence of overt gastrointestinal bleeding. Hist~mine-2-receptor antagonists are more effective than antacids at reducing overt hemorrhage (odds ratio 036 [95% CI 0.33 to 0.97]). A significant reduction in clinically important gastrointestlnai hemorrhage is evident only with hist~mlne-2-receptor antagonist therapy. There is a trend favoring antacids over sucralfate in the outcome of clinically important bleeding (odds ratio 0.65 [95% CI 0.16 to 2.49]); however, there are insufficient data to evaluate histamine-2-receptor antagonists versus sucralfate. No difference in mortality between treated and untreated patients was found. CONCLUSIONS: O v e r t gastrointestinal bleeding in critically ill patients is reduced by prophylaxis with antacids or hist~mine-2-receptor antagonists. Hist~mlne-2-receptor antagonists are more effective than antacids at decreasing overt bleeding and are more effective than no treatFrom the Department of Medicine (DJC), Division of Critical Care, and the Department of Clinical Epidemiology and Biostatistics (DJC, RJC, GHG), McMaster Faculty of Health Sciences, Hamilton, Ontario, Canada, and the Department of Pharmacy (LGW), Stanford Medical Center, Stanford, California. Dr. Deborah Cook is a St. Joseph's Hospital Research Scholar. Drs. Cook and Guyatt are Career Scientists of the Ontario Ministry of Health. Requests for reprints should be addressed to Deborah J. Cook, M.D., Department of Medicine, Division of Critical Care, St. Joseph's Hospital, 50 Charlton Avenue East, Hamilton, Ontario L8N 4A6, Canada. Manuscript submitted January 4, 1991, and accepted in revised form June 5, 1991.
ment at reducing the incidence of clinically important bleeding. Mortality rates in the intensive care unit are not decreased by stress ulcer prophylaxis.
inor upper gastrointestinal bleeding due to stress ulceration is common in morbidly ill hospitalized patients [1]. The incidence of serious hemorrhage, which is accompanied by a high degree of morbidity and mortality, is substantially lower [2-4]. The pathogenesis of stress ulceration in the critically ill involves disruption of the usual mechanisms of gastric mucosal integrity. Hypotension and systemic acidosis decrease gastric blood flow, resulting in impaired turnover of gastric epithelium, loss of the protective mucous and bicarbonate barrier, and back diffusion of hydrogen ions across the gastric mucosa [5]. Prophylactic therapy against stress ulceration has focused on cytoprotection (with sucralfate), and reduction (histamine-2-receptor antagonists) or neutralization (antacids) of gastric acid. The positive results of randomized trials in stress ulcer prophylaxis have led to recommendations that prophylaxis be administered to a large proportion of critically ill patients [4,6-13]. As such, control of gastric pH has been described as standard practice in intensive care units in North America and Europe [14-17]. However, many of these trials have included occult bleeding (positive results of guaiac tests of either gastric contents or stool) in their definition of stress ulcer bleeding. Studies employing a guaiacpositive nasogastric aspirate as a criterion for stress ulceration will therefore assign all bleeding to stress ulcers, including trauma from a nasogastric tube. In addition, results of guaiac testing are nonspecific [18]; cimetidine itself may produce a false-positive occult blood test [19-21]. Moreover, occult bleeding, even in the absence of prophylactic therapy, rarely progresses to overt bleeding [11] or is clinically important [13,22-25]. Even in studies using overt bleeding (such as hematemesis) in the definition of bleeding, the profile of a patient with overt bleeding may range from a transiently bloody nasogastric aspirate to hypovo-
M
November 1991 The American Journal of Medicine Volume 91
519
PREVENTIONOF GASTROINTESTINALBLEEDING/ COOKET AL
lemic shock. Accordingly, there may only be a small proportion of critically ill patients who actually develop clinically important bleeding. No single trial of prophylactic therapy has shown a reduction in clinically important bleeding. We therefore addressed the issue of whether ulcer prophylaxis in the critically ill results in a reduction in clinically important bleeding.
REVIEW OF THE LITERATURE Several descriptive review articles of stress ulcer prophylaxis have been published [26-29]. One scientific overview combined the data from 16 randomized trials to evaluate the efficacy of antacid and cimetidine therapy [1]. Both agents were more effective than placebo in preventing overall, occult, and overt bleeding (p
]REVIEW
Stress Ulcer Prophylaxis in the Critically I11:A Meta-Analysis DEBORAH J. COOK, M.D., Hamilton, Ontario, Canada, LANA G. WlTT, Ph.D., Stanford, California, RICHARD J. COOK, M.Sc., GORDON H. GUYATT, M.D., Hamilton, Ontario, Canada
PURPOSF~To e~rnmine the differential effect of stress ulcer prophylaxis on overt bleeding, clinically important bleeding, and mortality in critically ill patients. DATA IDENTIFICATION:Computerized bibliographic search of published and unpublished research. STUDY SELECTION:Independent review of 168 articles identified 42 relevant randomized trials for inclusion. DATAABSTRACTION:The validity, population, intervention, and outcomes of each trial were evaluated. RESULTS: Stress ulcer prophyl~xi~ with antacids (odds ratio 0.40 [95% confidence interval (CI) 0.20 to 0.79]) or hist~mlne-2-receptor antagonists (odds ratio 0~9 [95% CI 0.17 to 0.45]) decreases the incidence of overt gastrointestinal bleeding. Hist~mine-2-receptor antagonists are more effective than antacids at reducing overt hemorrhage (odds ratio 036 [95% CI 0.33 to 0.97]). A significant reduction in clinically important gastrointestlnai hemorrhage is evident only with hist~mlne-2-receptor antagonist therapy. There is a trend favoring antacids over sucralfate in the outcome of clinically important bleeding (odds ratio 0.65 [95% CI 0.16 to 2.49]); however, there are insufficient data to evaluate histamine-2-receptor antagonists versus sucralfate. No difference in mortality between treated and untreated patients was found. CONCLUSIONS: O v e r t gastrointestinal bleeding in critically ill patients is reduced by prophylaxis with antacids or hist~mine-2-receptor antagonists. Hist~mlne-2-receptor antagonists are more effective than antacids at decreasing overt bleeding and are more effective than no treatFrom the Department of Medicine (DJC), Division of Critical Care, and the Department of Clinical Epidemiology and Biostatistics (DJC, RJC, GHG), McMaster Faculty of Health Sciences, Hamilton, Ontario, Canada, and the Department of Pharmacy (LGW), Stanford Medical Center, Stanford, California. Dr. Deborah Cook is a St. Joseph's Hospital Research Scholar. Drs. Cook and Guyatt are Career Scientists of the Ontario Ministry of Health. Requests for reprints should be addressed to Deborah J. Cook, M.D., Department of Medicine, Division of Critical Care, St. Joseph's Hospital, 50 Charlton Avenue East, Hamilton, Ontario L8N 4A6, Canada. Manuscript submitted January 4, 1991, and accepted in revised form June 5, 1991.
ment at reducing the incidence of clinically important bleeding. Mortality rates in the intensive care unit are not decreased by stress ulcer prophylaxis.
inor upper gastrointestinal bleeding due to stress ulceration is common in morbidly ill hospitalized patients [1]. The incidence of serious hemorrhage, which is accompanied by a high degree of morbidity and mortality, is substantially lower [2-4]. The pathogenesis of stress ulceration in the critically ill involves disruption of the usual mechanisms of gastric mucosal integrity. Hypotension and systemic acidosis decrease gastric blood flow, resulting in impaired turnover of gastric epithelium, loss of the protective mucous and bicarbonate barrier, and back diffusion of hydrogen ions across the gastric mucosa [5]. Prophylactic therapy against stress ulceration has focused on cytoprotection (with sucralfate), and reduction (histamine-2-receptor antagonists) or neutralization (antacids) of gastric acid. The positive results of randomized trials in stress ulcer prophylaxis have led to recommendations that prophylaxis be administered to a large proportion of critically ill patients [4,6-13]. As such, control of gastric pH has been described as standard practice in intensive care units in North America and Europe [14-17]. However, many of these trials have included occult bleeding (positive results of guaiac tests of either gastric contents or stool) in their definition of stress ulcer bleeding. Studies employing a guaiacpositive nasogastric aspirate as a criterion for stress ulceration will therefore assign all bleeding to stress ulcers, including trauma from a nasogastric tube. In addition, results of guaiac testing are nonspecific [18]; cimetidine itself may produce a false-positive occult blood test [19-21]. Moreover, occult bleeding, even in the absence of prophylactic therapy, rarely progresses to overt bleeding [11] or is clinically important [13,22-25]. Even in studies using overt bleeding (such as hematemesis) in the definition of bleeding, the profile of a patient with overt bleeding may range from a transiently bloody nasogastric aspirate to hypovo-
M
November 1991 The American Journal of Medicine Volume 91
519
PREVENTIONOF GASTROINTESTINALBLEEDING/ COOKET AL
lemic shock. Accordingly, there may only be a small proportion of critically ill patients who actually develop clinically important bleeding. No single trial of prophylactic therapy has shown a reduction in clinically important bleeding. We therefore addressed the issue of whether ulcer prophylaxis in the critically ill results in a reduction in clinically important bleeding.
REVIEW OF THE LITERATURE Several descriptive review articles of stress ulcer prophylaxis have been published [26-29]. One scientific overview combined the data from 16 randomized trials to evaluate the efficacy of antacid and cimetidine therapy [1]. Both agents were more effective than placebo in preventing overall, occult, and overt bleeding (p
