Multimodal Management of Upper Gastrointestinal Bleeding Caused by Stress Gastropathy

Multimodal Management of Upper Gastrointestinal Bleeding
Caused by Stress Gastropathy
Vlad Denis Constantin1, Sorin Paun2, Victor V. Ciofoaia1, Vlad Budu1, Bogdan Socea1
1) St. Pantelimon Hospital; 2) Floreasca Emergency Hospital, “Carol Davila” University of Medicine and Pharmacy
Bucharest, Romania
Abstract
Introduction
Background: The 1999 ASHP (American Society of
Although stress-related mucosal disease (SRMD) is
Health-System Pharmacists) recommendation regarding the very common in critically ill patients, significant upper
prevention of stress-related mucosal disease and bleeding gastrointestinal (GI) bleeding from stress-related mucosal
in critical care patients by using PPI and H2RA still holds. disease is not [1]. Within the first 24 hours after ICU
We tried to compare the results obtained by our group with admission, approximately 75% to 100% of critically ill
the international data available and determine the benefits patients have some endoscopic evidence of gastroduodenal
of this prophylactic therapy. Methods: The present paper or upper GI lesions [2]. Clinical significant stress-related
presents a retrospective single center report of 36 patients mucosal bleeding (gastroduodenal bleeding associated
with upper gastrointestinal (GI) bleeding caused by stress with clinically important complications, i.e. hemodynamic
gastritis. Despite prophylaxis, the patients included in this instability, need for blood transfusion or need for surgery)
study who were admitted in the ICU during a five year period is a relatively infrequent event even when taking into
(2003-2008) with various underlying conditions, had clinical consideration only ICU patients, with the incidence declining
and endoscopic diagnoses of bleeding from stress-related in the last 30 years [3] mainly due to the modern intensive
gastric mucosal disease. The initial treatment focused on care settings [4] and apparently independent of the use of
patient stabilization first by medical intervention aimed at prophylaxis. Implementation of a stress ulcer prophylaxis
maintaining an elevated intragastric pH, in association with risk stratification scheme for ICU patients is necessary, since
haemostatic drugs and blood transfusions; complementary once this condition occurs, the mortality and morbidity
endoscopic or surgical haemostatic therapy was employed associated are extremely high and the management options
for patients unresponsive to the initial management. become very limited [5]. Prophylaxis of stress ulcers was
Results: Despite prophylactic acid suppressive therapy, shown to reduce the frequency of occurrence of major
upper GI bleeding findings were consistently present in high bleeding, but has not yet been shown to improve survival
risk patients, 69.4% presenting hematemesis and 55.6% [6]. The cause of death is usually related to the underlying
presenting coffee-ground gastric content. Conclusions: Each medical or surgical condition or to the multiple organ failure
institution needs to have guidelines in place to establish the aggravated by the bleeding itself.
patients that actually have sufficient risk factors to justify
The present paper analyses the diagnosis, the treatment
stress gastritis prophylaxis.
and the outcome in the management of patients with upper
GI bleeding caused by SRMD in patients on mechanical
Keywords
ventilation for more than 48 hours.
Stress-related upper GI bleeding – stress-related
Methods
gastric mucosal disease – stress gastritis – multivalent
management.
This study is a retrospective single center case series
report of 36 patients hospitalized in the St. Pantelimon
Received: 27.04.2009 Accepted: 17.06.2009
University Emergency Hospital in Bucharest. All patients
J Gastrointestin Liver Dis
included in this study have received prophylaxis (acid
September 2009 Vol.18 No 3, 279-284
Address for correspondence:
Vlad Denis Constantin
suppressive therapy) for SRMD.

St. Pantelimon Hospital, General
The following terms were used to define the study

Surgery Clinic, 340 Pantelimon Street,
group: the critically ill patients at risk for developing

73561 Bucharest, Romania
SRMD are the patients admitted to the ICU and requiring

Email: constantindenis@yahoo.com;

---

280
Constantin et al
mechanical ventilation or having coagulopathy or an ICU
Table I. Diagnosis of underlying conditions
stay longer than 1 week, with or without a history of active
peptic ulcer disease in the previous year or corticosteroid
Frequency (%)
therapy, regardless of the underlying condition (including
Acute myocardial infarction
1 (2.8%)
sepsis, massive burn injury, polytrauma, severe trauma and
Motor vehicle accident
21 (58.3%)
multiorgan failure, head trauma associated with increased
Work related trauma
2 (5.6%)
intracranial pressure). Upper GI bleeding was identified
Neoplasm
7 (19.4%)
by clinical findings, laboratory data and endoscopic
Severe pneumonia
2 (5.6%)
confirmation using multiple criteria: guaiac-positive stool
Acute renal failure
2 (5.6%)
and nasogastric aspirate, frank hematemesis or melena,
Entero-mesenteric infarction
1 (2.8%)
defined as any episode of coffee-ground emesis requiring
lavage, hematemesis or melena with or without a change in
Total
36 (100.0%)
hemoglobin levels or hematocrit, accompanied or not by a
decrease in hemoglobin level, a drop in blood pressure or
need for blood transfusion. Severe (clinically important)
gastrointestinal bleeding was identified by a spontaneous
decrease in systolic blood pressure of 20 mm Hg or more,
an increase in heart rate more than 20 beats per minute,
or a decrease in hemoglobin level more than 2 g/dL and
subsequent transfusion of blood after which hemoglobin
levels do not increase by a value defined as the number of
units transfused minus 2.
In order to assess the presence and the severity of upper
GI bleeding in these patients, relevant clinical findings and
laboratory data were gathered and documented. Once the
clinical diagnosis of upper GI bleeding was established,
the endoscopic examination was employed to confirm the
presence of the gastrointestinal mucosal lesions and the
etiology. The timeframe in which endoscopic examination
was performed ranged from 15 minutes to 12 hours after
the clinical diagnosis of upper GI bleeding. This represented
approx. 12 hours to 5 days after admission to the ICU and it
identified the hemorrhagic stigmata in patients with increased Fig 1. Mechanical ventilation (average duration).
risk of continued or resuming bleeding. A subset of patients
could not be managed medically or endoscopically and
surgical intervention was warranted.
Table II. Previous GI patient history
Cases (N=36)
male
female
N (%)
Results
Previous GI bleeding
1
1
2 (5.6%)
The study included 36 critical care patients, admitted to
Acid suppressive therapy
5
3
8 (22.2%)
the critical care ward with a severe polytrauma diagnosis
Previous peptic ulcer disease
2
0
2 (5.6%)
in 23 patients (21 motor vehicle accidents, 2 work related
(PUD) history
trauma), an underlying neoplasm with other than gastric
Surgical therapy attempted
3
1
4 (11.1%)
location following surgical intervention in 7 patients or other
Dyspeptic syndrome history
5
1
6 (16.7%)
causes (Table I).
All patients were mechanically ventilated for at
Twenty-one patients (58.33%) received histamine 2
least 48h and in one case for 18 days (Fig. 1). None receptor antagonists (H2RA) prophylaxis and 15 patients
had any underlying coagulopathy, such as a preexisting (41.66%) received proton pump inhibitors (PPIs). The
thrombocytopenia. Screening for hereditary coagulopathies average acid suppressive therapy duration was 2-3 days and
was not performed.
in 4 patients - 7 days.
In this group (Table II), only 2 (patients 5.55%) had a
The initial upper GI bleeding diagnosis was clinical,
history of upper GI bleeding and 8 patients (22.22%) were followed by endoscopic confirmation. In 20 patients the
receiving anti-acid medication at the time of bleeding. Patient nasogastric tube demonstrated coffee-ground-like gastric
history was significant for a previous episode of upper GI content; 6 patients presented with hematemesis and coffee-
bleeding in 2 patients, previous history of peptic ulcer in ground-like emesis. A subgroup of 7 patients presented with
2 male patients and intermittent dyspeptic syndrome in 6 an association of hematemesis and melena/coffee ground
patients.
emesis or melena and 7 patients presented with melena,

---

Upper GI bleeding caused by stress gastropathy
281
without hematemesis. Four patients presented hematochezia.
Other clinical signs were: significant skin pallor, tachycardia
(heart rate more than 100bpm) and decreased blood pressure
(systolic BP less than 90 mmHg), thirst, diaphoresis,
decreased diuresis, diffuse abdominal pain, dysphagia,
dyspepsia or jaundice (Table III).
Table III. Clinical presentation
Sex
Total
N (%)
male
female
N (% of total N)
N (% of total N)
Hematemesis
13 (52.0%)
12 (48.0%)
25 (69.4%)
HR>100 bpm,
16 (64.0%)
9 (36.0%)
25 (69.4%)
systolic
BP<90 mmHg
Coffee ground-
9 (45.0%)
11 (55.0%)
20 (55.6%)
like content
Diffuse
7 (77.8%)
2 (22.2%)
9 (25.0%)
abdominal pain
Decreased
5 (62.5%)
3 (37.5%)
8 (22.2%)
diuresis
Fig 2. Time to endoscopy after start of the gastrointestinal
Marked skin
4 (50.0%)
4 (50.0%)
8 (22.2%)
bleeding.
pallor
tamponade). Various other substances such as ethanol,
Melena
4 (57.1%)
3 (42.9%)
7 (19.4%)
polidocanol and sodium tetradecyl sulfate were employed
Coffee ground
5 (83.3%)
1 916.7%)
6 (16.7%)
in 4 cases, with good results in 3 cases: one patient required
emesis
surgical intervention.
Hematochezia
2 (50%)
2 (50%)
4 (11.1%)
Surgery was attempted in 4 cases (11.11%) that did not
Diaphoresis
2 (50.0%)
2 (50.0%)
4 (11.1%)
respond to the medical or endoscopic therapy. The registered
Dysphagia
1 (33.3%)
2 (66.7%)
3 (8.3%)
mortality was 7 cases (19.44%).
Significant
2 (66.7%)
1 (33.3%)
3 (8.3%)
thirst
Discussion
Jaundice
1 (50.0%)
1 (50.0%)
2 (5.6%)
Currently there are many terms used to identify the stress-
Dyspepsia
1 (50.0%)
1 (50.0%)
2 (5.6%)
related gastric damage in critically ill patients (stress ulcers/
ulceration, stress erosions, stress gastritis, hemorrhagic
The endoscopic examination was performed as soon as gastritis, erosive gastritis and SRMD). The upper GI bleeding
the patient was hemodynamically stable, regardless of the that can result from this condition can also be identified and
presence of the endotracheal tube, with an average period of graded according to multiple criteria (guaiac-positive stool
6-8h after the start of the GI bleeding. The timeframe within and nasogastric aspirate, frank hematemesis or melena,
which endoscopic examination was performed ranged from accompanied or not by a decrease in hemoglobin level, drop
approx. 15 minutes to 12 hours after establishing the clinical in blood pressure, need for blood transfusion). Spontaneous
diagnosis of upper GI bleeding, which represented a period decrease in systolic blood pressure of 20 mm Hg or more,
of 12 hours to 5 days after admission to the ICU. In 4 cases an increase in heart rate > 20 bpm, a decrease in systolic
immediate endoscopy was performed (Fig. 2).
blood pressure > 10 mmHg on sitting up, and a decrease
Endoscopy revealed erosions or ulcers, often with in hemoglobin level more than 2 g/dL and subsequent
stigmata of recent hemorrhage, such as adherent clots or transfusion of blood after which hemoglobin levels do not
petechiae. These lesions are usually more shallow, more increase by a value defined as the number of units transfused
diffuse and more numerous than those of peptic ulcers. In 3 minus 2, all identify clinically important GI bleeding and this
cases (83%), the esophago-gastro-duodenoscopy identified definition has been used in previous studies [7].
alongside the gastric lesions, associated duodenal lesions
According to a landmark multicenter prospective cohort
(erythema and erosions). The site of the gastric lesions was study on 2,252 patients by Cook et al [7], approx. 2% to 6% of
the upper 1/3 of the stomach in 14 patients (27.8%), the ICU patients had had overt bleeding (defined as any episode
whole stomach in 17 patients (47.2%) and the antropyloric of coffee-ground emesis requiring lavage, hematemesis, or
region in 10 patients (27.8%). The endoscopic examination melena with or without a change in hemoglobin levels or
revealed gastric ulcerations in 36 patients.
hematocrit). Similar results were obtained in another study
The techniques we used included thermocoagulation, [8]. Once the upper GI bleeding occurs, the mortality rate
epinephrine injection (0,5-1 ml dilution 1:10000), absolute in patients with endoscopic evidence of ulcer, bleeding, or
alcohol and saline injections (saline used in order to achieve both within 18 hours of admission to a medical ICU was 57%

---

282
Constantin et al
compared with 24% in patients with either a normal mucosa, the etiology of SRMD, associating also other two dreaded
only nonhemorrhagic erosions or petechial changes [9]. complications - GI hypomotility and diarrhea [19]. The
The mortality for the critically ill patients with GI bleeding inflammatory response in severely ill patients can severely
is 50% to 77%, and the cause of death is not that much the affect the patient’s gastric mucosa perfusion, even without
bleeding itself, but the underlying medical condition or the influencing peripheral oxygen saturation levels [20].
aggravated multiple organ failure.
Luminal acid secretion can be increased by psychological
According to the 1994 Cook study [7], prophylaxis factors and leads to multiple erosive lesions, independent
against stress ulcers can be safely withheld from critically ill of previous H.pylori infection status or NSAIDs exposure,
patients unless they have coagulopathy or require mechanical enhancing the aggressive factors that predispose to peptic
ventilation, since few critically ill patients have clinically ulcer or exacerbating preexisting ulcer disease. The identified
important GI bleeding. The prophylaxis and the treatment for risk factors for clinically important GI bleeding include
the GI bleeding determined by stress gastritis are guided by a major burns, head injury, previous peptic ulcer disease
well defined therapeutic protocol realized in 1999 by ASHP in the last 6 weeks, organ transplants, upper GI bleeding
(The American Society of Health-System Pharmacists). in the previous 42 days [7], respiratory failure requiring
This protocol defines the indications for medical, surgical, mechanical ventilation for longer than 48h, the presence of
respiratory, and pediatric ICU patients and still holds [10]. a coagulopathy [7], and as probable risk factors older age,
Stress ulcer prophylaxis is recommended for adults admitted repair of abdominal aortic aneurysms, severe burns, multiple
to the ICU who have coagulopathy, require mechanical organ failure, neurological trauma, sepsis or septic shock,
ventilation for more than 48 hours, have a history of GI and high-dose corticosteroid therapy (intravenous or oral
ulceration or bleeding within 1 year before admission or ≥40 mg/day) [2].
have at least two of the following risk factors: sepsis, ICU
Currently, the indications for prophylactic treatment
stay longer than 1 week, occult bleeding lasting 6 days or are to some degree arbitrary. Since hypotension is a known
longer and use of more than 250 mg hydrocortisone or the risk for decreased mucosal perfusion, maintenance of the
equivalent [11]. The ASHP recommendations do not cover hemodynamic stability is very important. It is extremely
single system injuries such as head trauma, spinal cord injury important to prevent gastric injury by prompt and appropriate
or thermal injury [12].
therapy of the underlying systemic disease, including
Currently, in strong contrast to the initial 1994 and 1999 reversal of hypoxemia and hypotension and by avoidance
recommendation, the use of acid suppressive therapy as part of gastrotoxic medications such as aspirin or other NSAIDs
of SRMD and associated upper GI bleeding prophylaxis [21]. In 2002, 86% of critically ill patients admitted to the
has become increasingly more common in general patients, ICU received some sort of stress ulcer prophylaxis [22].
with little to no evidence to support it. As many as 71% of The agents used are H2RA, sucralfate and PPIs [2], with a
patients in general medicine wards are receiving some sort of tendency towards using PPI more frequently. Proton pump
acid-suppressive therapy without an appropriate indication, inhibitors provide more consistent pH control than H2RA.
although this practice is currently not recommended or There is no consensus on the drug of choice for stress ulcer
supported [10, 11], and despite the fact that any form of prophylaxis with several meta-analyses providing conflicting
prophylaxis against stress ulcers is expensive [13, 14] and results on the superiority of any medication [19]. Now,
can have adverse effects [15].
H2RA are the most widely used class of agents for SMRD
Stress-related mucosal disease encompasses two types prophylaxis, although PPIs are gaining in acceptance and
of mucosal lesions and the continuum that can be described are proved to be as effective as H2RA [5]. H2RA are more
between these two: (1) stress-related injury, which is diffuse, effective than sucralfate [23].
superficial mucosal damage, and (2) discrete stress ulcers,
It has not been established whether enteral feeding
which are deep focal lesions that penetrate the mucosa to the is protective or just a marker for a patient that has a less
submucosa. Both types are caused by mucosal ischemia [16] severe condition, although it is established that it optimizes
and both show a propensity for the acid-producing corpus splanchnic distribution and lessens macroscopic ulceration
and fundus [17].
[5]. Proper randomization is difficult to perform in the
The stress-related erosive syndrome can be caused studies that address enteral feeding and can theoretically
by multiple factors, ranging from sepsis, massive burn invalidate the consistent result in various studies that early
injury, polytrauma, head trauma associated with increased enteral feeding is associated with a less likely tendency to
intracranial pressure, and multiorgan failure [7]. The bleed than in patients who are not tolerating enteral feeding.
pathophysiology is multifactorial and includes major A remarkable result is presented in a study by Raff et al
factors such as reduced blood flow, mucosal ischemia, [24], who concluded that enteral feeding initiated within
hypoperfusion, and reperfusion injury [11]. Still, despite 12 h of trauma (if tolerated by the patient and this in itself
the postulated pathogenetic role of ischemia in stress is a marker of the patient’s condition) was as effective in
gastritis, selective vasopressin infusion in the left gastric reducing the risk of clinical significant upper GI bleeding
artery achieves hemostasis in 80% of cases [18]. In patients as H2RA therapy and/or anti-acids [25].
on mechanical ventilation, positive pressure-induced
The incidence of stress related mucosal bleeding is
splanchnic hypoperfusion appears to play a central role in decreasing [5]. This is attributed to both improvements in

---

Upper GI bleeding caused by stress gastropathy
283
critical care and to the prophylactic administration of acid
60.
suppressive therapy [26, 27]. Recent trials [27-29] emphasize 5. Sesler JM. Stress-related mucosal disease in the intensive care unit: an
that prophylactic therapy seems not to alter the already low
update on prophylaxis. AACN Adv Crit Care 2007; 18: 119-126.
incidence of stress-related clinically important bleeding even 6. Spirt MJ. Stress-related mucosal disease: risk factors and prophylactic
in high risk patients or to justify the cost of the treatment;
therapy. Clin Ther 2004; 26: 197-213.
this raises the questions whether there will be a time when 7. Cook DJ, Fuller HD, Guyatt GH, et al. Risk factors for gastrointestinal
bleeding in critically ill patients. Canadian Critical Care Trials Group.
routine prophylaxis for SRMD will no longer be necessary
N Engl J Med 1994; 330: 377-381.
[5] and whether there is a need for a review of the 1999 8. Schuster DP, Rowley H, Feinstein S, McGue MK, Zuckerman GR.
ASHP recommendation.
Prospective evaluation of the risk of upper gastrointestinal bleeding
In our institution, the underlying management principle
after admission to a medical intensive care unit. Am J Med 1984;
for upper GI bleeding caused by stress-gastritis in the
76: 623-630.
critically ill patients is prophylaxis, by critically ill patients 9. Peura DA, Johnson LF. Cimetidine for prevention and treatment of
identifying the patients admitted to the ICU and requiring
gastroduodenal mucosal lesions in patients in an intensive care unit.
mechanical ventilation or having coagulopathy or ICU
Ann Intern Med 1985; 103: 173-177.
stay longer than 1 week. A history of active peptic ulcer 10. Grube RR, May DB. Stress ulcer prophylaxis in hospitalized patients
disease in the previous year or corticosteroid therapy are
not in intensive care units. Am J Health Syst Pharm 2007; 64: 1396-
also considered indications for prophylaxis. Gastric contents
1400.
11. Spirt MJ. Acid suppression in critically ill patients: what does the
pH monitoring provides an objective variable to determine
evidence support? Pharmacotherapy 2003; 23: 87S-93S.
the current status of the patient and the efficiency of the 12. ASHP Therapeutic Guidelines on Stress Ulcer Prophylaxis. ASHP
therapy, but it is not routinely performed. The target pH value
Commission on Therapeutics and approved by the ASHP Board of
should be greater than 4.0. Anything less should prompt the
Directors on November 14, 1998. Am J Health Syst Pharm 1999;
clinician to double the dose of the agent used if the patient
56: 347-379.
was previously on prophylaxis.
13. Smythe MA, Zarowitz BJ. Changing perspectives of stress gastritis
The prophylaxis of upper GI bleeding caused by stress
prophylaxis. Ann Pharmacother 1994; 28: 1073-1085.
gastropathy should be applied by first identifying the major 14. Maier RV, Mitchell D, Gentilello L. Optimal therapy for stress
risk patients and then by taking into account economic
gastritis. Ann Surg 1994; 220: 353-360.
factors, since these determine the futility of large scale 15. Driks MR, Craven DE, Celli BR, et al. Nosocomial pneumonia in
implementations. For each patient, the therapeutic decision
intubated patients given sucralfate as compared with antacids or
histamine type 2 blockers. The role of gastric colonization. N Engl
should be individualized according to the severity of the
J Med 1987; 317: 1376-1382.
presenting hemorrhage and the coexisting morbidities. We 16. Shiotani A, Graham DY. Pathogenesis and therapy of gastric and
observed that this approach still does not manage all cases
duodenal ulcer disease. Med Clin North Am 2002; 86: 1447-1466.
of this rare condition, raising the question whether there 17. Spirt MJ. Stress-related mucosal disease. Curr Treat Options
is an economic justification to the current prophylactic
Gastroenterol 2003; 6: 135-145.
symptomatic approach.
18. Baum S. Angiographic diagnosis and treatment of gastrointestinal
bleeding. In: Baum SPM (ed.). Abram’s angiography: interventional
Conclusion
radiology. Boston: Little, Brown; 1997: 389-421.
19. Mutlu GM, Mutlu EA, Factor P. Prevention and treatment of
We recommend that each institution has guidelines in
gastrointestinal complications in patients on mechanical ventilation.
place to determine what patients actually have sufficient
Am J Respir Med 2003; 2: 395-411.
risk factors to require stress gastropathy prophylaxis. The 20. Spirt MJ, Guth PH, Randall G, Leung FW. Gastroduodenal perfusion
type of therapy applied must be tailored to the patient’s
and mortality in mechanical ventilation-dependent patients with
presentation, taking into consideration the gravity of
systemic inflammatory response syndrome. Dig Dis Sci 2004; 49:
bleeding, the associated morbidities and the general status
906-913.
of the patients.
21. Cappell MS, Friedel D. The role of esophagogastroduodenoscopy in
the diagnosis and management of upper gastrointestinal disorders.
Conflicts of interest
Med Clin North Am 2002; 86: 1165-1216.
22. Barletta JF, Erstad BL, Fortune JB. Stress ulcer prophylaxis in trauma
No conflict to declare.
patients. Crit Care 2002; 6: 526-530.
23. Cook D, Guyatt G, Marshall J, et al. A comparison of sucralfate and
References
ranitidine for the prevention of upper gastrointestinal bleeding in
patients requiring mechanical ventilation. Canadian Critical Care
1. Harty RF, Ancha HB. Stress ulcer bleeding. Curr Treat Options
Trials Group. N Engl J Med 1998; 338: 791-797.
Gastroenterol 2006; 9: 157-166.
24. Raff T, Germann G, Hartmann B. The value of early enteral nutrition
2. Spirt MJ, Stanley S. Update on stress ulcer prophylaxis in critically
in the prophylaxis of stress ulceration in the severely burned patient.
ill patients. Crit Care Nurse 2006; 26: 18-28.
Burns 1997; 23: 313-318.
3. Yang YX, Lewis JD. Prevention and treatment of stress ulcers in 25. Fennerty MB. Pathophysiology of the upper gastrointestinal tract in
critically ill patients. Semin Gastrointest Dis 2003; 14: 11-19.
the critically ill patient: rationale for the therapeutic benefits of acid
4. Laine L, Takeuchi K, Tarnawski A. Gastric mucosal defense and
suppression. Crit Care Med 2002; 30(6 Suppl): S351-355.
cytoprotection: bench to bedside. Gastroenterology 2008; 135: 41-
26. Kupfer Y, Cappell MS, Tessler S. Acute gastrointestinal bleeding in

---

284
Constantin et al
the intensive care unit. The intensivist’s perspective. Gastroenterol 28. Kantorova I, Svoboda P, Scheer P, et al. Stress ulcer prophylaxis
Clin North Am 2000; 29: 275-307.
in critically ill patients: a randomized controlled trial.
27. Kahn JM, Doctor JN, Rubenfeld GD. Stress ulcer prophylaxis in
Hepatogastroenterology 2004; 51: 757-761.
mechanically ventilated patients: integrating evidence and judgment 29. Faisy C, Guerot E, Diehl JL, Iftimovici E, Fagon JY. Clinically
using a decision analysis. Intensive Care Med 2006; 32: 1151-
significant gastrointestinal bleeding in critically ill patients with
1158.
and without stress-ulcer prophylaxis. Intensive Care Med 2003; 29:
1306-1313.

---