Académique Documents
Professionnel Documents
Culture Documents
A SHP REPORT
the literature to be reviewed because they involved One problem with both the grading systems9,10 on
meta-analysis and distinctions based on confidence which this system was based is the relative inability to
intervals, homogeneity, and heterogeneity. These distinguish the quality of individual investigations.
methods were modified slightly and expanded to in- Two RCTs may each be ranked similarly, but one may
clude a category D that represents a panel consensus have been better designed and reported than the other.
based on the clinical experience of the panel members Therefore, in addition to listing levels of evidence for
and a paucity of quality supporting literature. Before the various trials, the superscript wd was used to signify
reviewing the literature, the panel of experts agreed a well-designed and well-reported trial when the guide-
that a meta-analysis carried more weight than a ran- lines were sent to the panel of experts and other review-
domized, controlled trial (RCT) in determining a rec- ers to aid them in evaluating the literature presented.
ommendation unless at least two RCTs had similar More than 30 criteria, some of which have been pub-
results or one RCT was designed with a sample large lished, were used in the analysis of the investiga-
enough to detect a difference of 50% or less in bleeding tions.13,14 The criteria represented a nonnumerical as-
rates (considered the primary outcome) between the sessment and were not used as a scoring system for the
treatment and control groups with a power of 80% formal statistical analysis.15
and an of 0.05. The levels of evidence were used to evaluate those
The recommendations in this document were catego- studies that assessed the frequency of clinically impor-
rized according to the strength of the evidence. Category tant bleeding (i.e., gastroduodenal bleeding associated
A corresponds to levels of evidence I+, I, and I; category with hemodynamic compromise or the need for blood
B to levels II+, II, and II; and category C to levels III+, III, transfusion or surgery) and nosocomial pneumonia.
IV+, IV, and V. Table 1 includes definitions for each level Analysis of the number and type of risk factors associat-
of evidence. The level of evidence was evaluated for each ed with bleeding by a rigid evidence-based evaluation
trial. When there were multiple RCTs of comparable may be misleading for a number of reasons, including
power pertaining to a particular recommendation and the following: studies used various definitions of bleed-
their levels of evidence differed, the level of evidence for ing and risk factors associated with bleeding, various
the least stringent trial was used in determining the risk factors were studied, various risk factors were used
strength of evidence for a recommendation. Recommen- as inclusion criteria, the analysis of risk factors often
dations were assigned to category D if they represented was not a study objective, and, in many studies, multi-
the expert opinion of panel members. For a recommen- variate analysis was not used to determine which risk
dation based on RCTs to be assigned to category A, there factors might be independent predictors of bleeding.
had to be at least two RCTs or one RCT with a sample size Observational trials with large numbers of patients are
sufficient for detecting a difference of 50% or less in rates often used appropriately to assess risk factors for bleed-
of clinically important bleeding with a power of 0.8 and ing, although the strength of evidence would never be
an of 0.05. higher than C.
Table 1.
System Used for Defining Levels of Evidence in Stress Ulcer Prophylaxis Guidelinesa
Level of
Evidence Definition
I+ Meta-analysis of randomized, controlled trials (RCTs) with homogeneityb of results and a 95% confidence interval (CI)
that lies entirely on one side of the numerical threshold for clinically important benefit
I Meta-analysis of RCTs with homogeneity of results but with a 95% CI that includes the threshold for clinically
important benefit
I RCT in which the entire 95% CI lies on one side of the threshold for clinically important benefit
II+ Meta-analysis of RCTs with heterogeneity of results and a 95% CI that lies entirely on one side of the threshold for
clinically important benefit
II Meta-analysis of RCTs with heterogeneity of results and a 95% CI that includes the threshold for clinically important
benefit
II RCT in which the 95% CI includes the threshold for clinically important benefit
III+ Nonrandomized concurrent cohort study in which the entire 95% CI lies on one side of the threshold for clinically
important benefitc
III Nonrandomized concurrent cohort study in which the 95% CI includes the threshold for clinically important benefit
IV+ Nonrandomized historic cohort study in which the entire 95% CI lies on one side of the threshold for clinically
important benefit
IV Nonrandomized historic cohort study in which the 95% CI includes the threshold for clinically important benefit
V Case series suggesting clinically important benefit
a
Adapted from references 9 and 10.
b
Various definitions of homogeneity were used in the literature (e.g., the difference between the risk reductions or odds ratios between the two most
disparate trials is less than 20% and the difference between 95% confidence intervals for the two most disparate trials is less than 5%); the fact that
homogeneity was tested for is more important than the precise method used.
c
This assumes the availability of comparative data for calculating 95% CIs.
Pathophysiology of stress-induced lesions that become more acidic within a few hours of birth.27
Stress ulceration is a form of hemorrhagic gastritis Gastric pH values in infants begin to approach those of
that may occur in patients who have had a major healthy adults during the first six months of life. How-
stressful event such as trauma, surgery, organ failure, ever, stress ulcer lesions may occur at any age, including
sepsis, or thermal injury. In general, the histological in preterm infants, lending additional support to a
and macroscopic appearance of stress-induced fundal multifactorial process.28
lesions are similar to those associated with nonsteroidal The frequency of ulceration, the number of ulcers, and
anti-inflammatory drugs (NSAIDs) seen in the antrum the sites of ulceration may be different between children
and the body of the stomach. However, in contrast to and adults, depending on the stress-related event. In one
NSAID-induced lesions, stress-induced lesions often retrospective study involving 93 children less than 10
cause more congestion and bleeding, and they eventu- years of age and 178 adults who died of burn-related
ally involve multiple sites in the upper GI tract.16 Cush- injuries, 13 (14.0%) of the children had duodenal ulcer-
ings ulcer, which is associated with CNS injury, often ation on postmortem examination compared with 13
presents as a single deep lesion in the duodenum or in (7.3%) of the adults. In addition, 21 (22.6%) of the
the stomach.16 Curlings ulcer, which is associated with children had gastric ulceration compared with 20
thermal injury, is morphologically similar to other (11.2%) of the adults.29 No relationship between sex and
stress-induced lesions but may appear in the esopha- ulceration was found. Another review of infants less than
gus, the stomach, the small intestine, or the colon.16 one year of age with stress ulceration found that single
The pathogenesis of stress-induced lesions is proba- duodenal ulcers were common, in contrast to the multi-
bly multifactorial.17 Acid hypersecretion is often stated ple gastric erosions usually found in adults.30
as being particularly important in the formation of
ulcers associated with head and thermal injuries.18 Frequency of bleeding, efficacy
However, acid hypersecretion is not a universal finding of prophylaxis, and risk factors for bleeding
in such patients, and higher pH values (>4.5) do not in general (medical, surgical, respiratory,
ensure avoidance of lesions.19-21 During stressful events, pediatric) ICU populations
normal protective mechanisms are altered, including Acute bleeding in critically ill patients has been
epithelial turnover in the gastric mucosa and secretions described in the medical literature since the 1800s. To
of mucus and bicarbonate. These events, combined determine the true frequency of, and risk factors associ-
with the release of various mediators (e.g., arachidonic ated with, stress-induced bleeding in general ICU pop-
acid metabolites, cytokines, oxygen free radicals), result ulations, studies that contained at least one defined
in erosions that may progress to ulceration and bleed- group of patients who did not receive prophylaxis were
ing.22 Reductions in mucosal blood flow may also be evaluated.31-63 As will be seen in the following discus-
important in ulcer formation. Alterations in blood flow sion, determining a true frequency of clinically impor-
at the microcirculatory level may initiate a series of tant bleeding is complicated by the difficulty in mea-
changes resulting in erosions that could progress to suring endpoints, the variability in endpoints used, the
ulceration and bleeding. Necrosis and erosions of the variability in definitions of endpoints, and the hetero-
gastric membrane have occurred within 45 minutes of geneity of the patient populations.
hemorrhagic shock in experiments in rats.23 In adult In general, the studies included in this section of the
humans, erosions may occur very quickly (i.e., during document on general ICU populations determined the
the first 24 hours of hospital admission) in association frequency of stress ulcer lesions and the risk factors
with a stressful event, although the time from the associated with clinically important bleeding in pa-
inciting event to the diagnosis of bleeding may be tients in medical, surgical, or respiratory ICUs. Only
much longer (e.g., >1014 days). two of the studies involved a relatively homogeneous
Other less well-studied factors may be involved in the sample of patients with a single-system disease or sur-
pathogenesis of stress-related lesions. For example, ele- gery. One study was conducted in patients with respira-
vated concentrations of anti-Helicobacter pylori immuno- tory disease, and the other involved patients undergo-
globulin A have been found in the serum of critically ill ing abdominal surgery.36,44 These studies were included
patients.24 Although the association between H. pylori because the patient populations were similar to those
and chronic ulceration has been established, few studies enrolled in other investigations. Studies enrolling pa-
have been performed in acute care settings. tients with single-system injuries (e.g., head or spinal
It is not known whether differences in basal or cord surgery), patients with thermal injuries, or pa-
maximal gastric acid production seen in various age tients undergoing transplantation were not included
groups affect the frequency of ulceration and clinically (see the section on special populations) because of
important bleeding. For example, premature infants presumed differences in the pathophysiology or fre-
typically produce less gastric acid than full-term in- quency of stress-induced bleeding.
fants, albeit with substantial interinfant variability.25,26 Frequency of bleeding. In all studies conducted
Newborns often have slightly acidic gastric pH values before 1978, the frequency of clinically important
bleeding and related complications ranged from 5.3% The study was designed to have 80% power to detect a
to 33%. From 1978 on, substantial differences in the 75% treatment-related reduction in bleeding. Patients
frequency of stress-induced bleeding have been record- could be randomly assigned to a control (no placebo)
ed among trials. For example, in an RCT published by group, a pH-adjusted cimetidine group, or a sucralfate
Pinilla et al.42 in 1985, the frequency of bleeding was group. There were no significant differences between
1.6% (1 of 61 patients), and it was not clear whether this groups with respect to clinically important bleeding
patient had clinically important bleeding. In contrast, episodes (including some patients whose bleeding did
an RCT published by Peura and Johnson43 in the same not clear by lavage): 6 of 100 patients in the control
year, involving adult patients in a medical ICU, found group, 5 of 100 patients in the cimetidine group, and 5
a 39% frequency (7 of 18 patients) of endoscopically of 100 patients in the sucralfate group. In addition,
confirmed, clinically important, stress-induced bleed- there were no significant differences between groups
ing in patients not receiving prophylaxis. Articles pub- when respiratory failure and coagulopathy were stud-
lished from 1984 to 1994 on the frequency of clinically ied as risk factors. If the different findings between this
important bleeding in patients not receiving prophy- trial and that of Martin et al. can be attributed to
laxis indicate that the average frequency of bleeding is differences in the populations being studied, it might
6% (0.139%).64 be reasonable to consider withholding prophylaxis in
With regard to the pediatric population, it is difficult general medical ICU patients. However, it should be
to estimate the frequency of clinically important bleed- stressed that the trial by Ben-Menachem et al. was not
ing because of the small numbers of patients enrolled in designed to detect a difference of less than 75% in
published investigations and the use of endpoints oth- bleeding between groups, even though a smaller differ-
er than clinically important bleeding (e.g., endoscopic ence might be clinically important.
evidence of bleeding or overt bleeding). There are im- Another study enrolled 90 intensive care patients
portant exceptions, albeit with differing results. In one with septic shock, severe head injury, and trauma.67 The
study involving 140 children from birth to 20 years of patients were randomly assigned to receive pirenzepine
age, the frequency of clinically important bleeding was 30 mg/day, omeprazole 80 mg/day, or no prophylaxis.
20% in a control group and 5.7% in a combined treat- A rating scale of zero (normal mucosa) to four (>25
ment group.65 In contrast, the frequency of clinically hemorrhages, erosions, or invasive ulcers) was used to
important bleeding (not defined in the published ab- evaluate all patients at baseline and after five days of
stract) was 1.6% in a study of 1006 admissions to a prophylaxis. Although all patients were at high risk for
pediatric ICU, although it was not stated whether the stress ulcer hemorrhage, no patients developed severe
children received any form of stress ulcer prophylaxis.66 bleeding from stress lesions, and no difference in endo-
Efficacy. Prophylaxis versus no prophylaxis. Two scopic evaluations was noted at baseline or at five days
studies conducted in the 1990s can be used to illustrate among the three groups.
the disparate results of the more than 20 RCTs of Because individual trials have demonstrated con-
prophylaxis conducted in general ICU populations. flicting results with respect to the frequency of bleeding
The first trial, conducted in 1993 by Martin et al.,58 was with or without prophylaxis, several meta-analyses
a multicenter, double-blind, placebo-controlled study have sought to resolve this issue (Table 2).5,68-71 Ab-
involving general medical and surgical ICU patients stracts and non-English-language articles were fre-
with an expected length of stay of at least 36 hours and quently included in the data sets of these analyses, as
at least one risk factor (of eight studied) for bleeding. were reports of studies involving populations that other
The study had 90% power to detect a 75% difference in studies specifically excluded (e.g., patients with head or
bleeding between placebo and pH-adjusted cimetidine thermal injuries). One meta-analysis was published in
therapy. The study was terminated early when an inter- 1989, and three were published in 1991, so the number
im analysis revealed a significant difference in bleeding of trials combined for each analysis was limited.68-71
rates between the control and cimetidine groups. In the Despite this difference in time frame, there was definite
control group, 16 (24.2%) of 66 patients had overt and overlap among the meta-analyses in the studies that
clinically important bleeding (some bleeding was not were included. Only one of the meta-analyses investi-
cleared by lavage, but the patients did not require gated the difference in bleeding rates with sucralfate
transfusion), compared with 5 (7.7%) of 65 patients in compared with no prophylaxis.5
the treatment group (p = 0.009). Six (10.0%) of 66 The meta-analysis conducted by Lacroix et al.68 in-
patients in the control group and 4 (6.2%) of 65 in the cluded randomized studies only. The frequency of
treatment group had clinically important bleeding. Sta- bleeding in patients receiving no prophylaxis (or place-
tistical testing of this endpoint was not performed. bo) ranged from 3.4% to 52.7%. Use of cimetidine and
The second study, a single-blind RCT, was conduct- antacids was associated with a significantly lower fre-
ed in 1994 by Ben-Menachem et al.59 In that trial, only quency of bleeding compared with no prophylaxis.
medical ICU patients who were expected to have a Among the limitations of this meta-analysis were that
length of ICU stay of at least 24 hours were enrolled. the evaluated studies had methodological problems,
demonstrated heterogeneous effects, and investigated receiving antacids versus 14.9% with no prophylaxis. A
the frequency of occult or overtbut not necessarily second meta-analysis by Tryba71 in 1991 included only
clinically importantbleeding, defined as hemody- studies with clinically important bleeding. The results
namic instability or the need for blood transfusion. were similar to those of Trybas previous meta-analysis.
In a meta-analysis published by Tryba69 in 1991, Two meta-analyses published by Cook et al.5,70 in
overt and clinically important bleeding (i.e., need for 1991 and 1996 included only randomized, controlled
blood transfusion) were combined in the definition of trials. The investigations looked at clinically important
upper GI bleeding. Randomized and nonrandomized bleeding, defined as overt bleeding with hemodynamic
studies were included. The odds ratios (ORs) and 95% changes (a 20-mm Hg decrease in blood pressure within
confidence intervals (CIs) calculated for H2-receptor 24 hours or a 10-mm Hg decrease plus a 20-beat/min
antagonists or antacids compared with no prophylaxis increase in heart rate on standing) or the need for
were all less than one (i.e., treatment was effective in transfusion (a decrease in hemoglobin of 2 g/dL plus
preventing bleeding). The data were presented only in transfusion of two units of blood in 24 hours). In the
the form of figures, so the exact ORs and CIs are not 1991 analysis, H2-receptor antagonists were found to be
available. However, the frequency of bleeding was 5.3% more efficacious in preventing clinically important
in the group receiving H2-receptor antagonists versus bleeding than no prophylaxis.
16.1% with no prophylaxis and 5.6% in the group The 1996 meta-analysis was performed to resolve
Table 2.
Meta-analyses Investigating the Risk of Bleeding with and without Stress Ulcer Prophylaxis
Reference Comparisona No. Trials Endpoint Resultsb
68 Cimetidine versus 7 Overt bleeding OR = 0.32, CI = 0.210.49 (lower rate of
control bleeding with cimetidine)
Antacids versus 9 Overt bleeding OR = 0.12, CI = 0.080.19 (lower rate of
control bleeding with antacids)
Cimetidine versus 10 Overt bleeding OR = 1.61, CI = 0.972.65
antacids
69 Antacids versus 6 Overt or clinically important OR and CI < 1 (lower rate of bleeding with
control bleeding antacids)
H2-RAs versus 14 Overt or clinically important OR and CI < 1 (lower rate of bleeding with H2-
control bleeding RAs)
H2-RAs versus 14 Overt or clinically important CI included 1
antacids bleeding
Sucralfate versus 6 Overt or clinically important OR and CI < 1 (lower rate of bleeding with
H2-RAs bleeding sucralfate)
Sucralfate versus 10 Overt or clinically important CI included 1
antacids bleeding
71 Sucralfate versus 9 Clinically important bleeding OR = 0.532, CI = 0.3030.933 (lower rate of
H2-RAs bleeding with sucralfate)
Sucralfate versus 10 Clinically important bleeding OR = 0.868, CI = 0.4521.667
antacids
70 Antacids versus 3 Clinically important bleeding OR = 0.35, CI = 0.081.33
control
H2-RAs versus 6 Clinically important bleeding OR = 0.35, CI = 0.150.76 (lower rate of
control bleeding with H2-RAs)
H2-RAs versus 9 Clinically important bleeding OR = 0.84, CI = 0.451.56
antacids
Antacids versus 5 Clinically important bleeding OR = 0.65, CI = 0.162.49
sucralfate
H2-RAs versus 1 Clinically important bleeding OR = 0.95, CI = 0.0615.40
sucralfate
5 Antacids versus 3 Clinically important bleeding OR = 0.35, CI = 0.091.41
control
H2-RAs versus 10 Clinically important bleeding OR = 0.44, CI = 0.220.88 (lower rate of
control bleeding with H2-RAs)
H2-RAs versus 10 Clinically important bleeding OR = 0.86, CI = 0.461.59
antacids
Sucralfate versus 1 Clinically important bleeding OR = 1.26, CI = 0.1212.87
control
Sucralfate versus 5 Clinically important bleeding OR = 1.49, CI = 0.435.27
antacids
Sucralfate versus 4 Clinically important bleeding OR = 1.28, CI = 0.276.11
H2-RAs
a
H2-RAs = histamine H2-receptor antagonists.
b
OR = odds ratio, CI = 95% confidence interval. Exact values provided when available; some values were extrapolated from figures.
some of the inconsistencies associated with the previ- come. A significant (p < 0.05) difference between the
ously conducted meta-analyses and to incorporate active-treatment groups and the no-treatment group
more recent studies.5 The methodology was similar to was achieved only if the results for the three active-
that of the earlier meta-analysis. Only H2-receptor an- treatment groups were combined. Two studies focused
tagonists were found to have a 95% CI of <1 for prevent- on a neonatal population but did not use clinically
ing clinically important bleeding compared with no important bleeding as an endpoint.49,63 In one of these
prophylaxis. However, only three trials with an antacid studies that had a control group, the risk of endoscop-
group and one with a sucralfate group were available for ically diagnosed gastric lesions in mechanically venti-
comparison with no prophylaxis. Because the 1996 lated infants was found to be lower with ranitidine
meta-analysis was the most recent and comprehensive, prophylaxis (OR = 0.03; 95% CI, 0.0030.178). Surfac-
its results are used in this document whenever the tant therapy, which was used to treat infants with
findings from previous meta-analyses are conflicting. respiratory distress syndrome, was also associated with
To date, the study with the largest number of pro- a lower rate of formation of lesions (OR = 0.083; 95% CI,
spectively enrolled patients was published by Cook et 0.0090.788).63 The mechanism for the apparent pro-
al. in 1994.61 This multicenter follow-up trial enrolled tective effect of surfactant on the GI tract remains to be
2252 patients older than 16 years of age who were elucidated. No RCT conducted to date in pediatric pa-
admitted to medical and surgical ICUs. Approximately tients has conclusively demonstrated a significant dif-
half of the enrolled patients were admitted for cardio- ference in clinically important bleeding when prophy-
vascular surgery. Some subgroups (head or spinal cord laxis was compared with no prophylaxis.
injuries, multiple trauma or thermal injuries, and trans- Risk factors. In Cooks prospective observational
plantation) had few patients. This limits the generaliz- study, two risk factors were found to be significant
ability of the results to these populations. Lack of predictors of stress-induced bleeding: respiratory fail-
resources at three of the four hospitals prohibited en- ure, as defined by mechanical ventilation for at least 48
rollment of consecutive patients as intended, and phy- hours (OR = 15.6; CI not reported, p < 0.001), and
sicians were not required to withhold prophylaxis. coagulopathy, defined as a platelet count of <50,000
Various prophylactic medications were used, and the mm3, an International Normalized Ratio of >1.5, or a
dosage and duration of prophylaxis were not con- partial thromboplastin time of >2 times the control
trolled. Clinically important bleeding occurred in 10 value (OR = 4.3; CI not reported, p < 0.001).61 The
(0.6%) of 1578 patients not receiving prophylaxis and frequency of bleeding was 3.7% (95% CI, 2.55.2%) if
23 (3.4%) of 674 patients receiving prophylaxis (22 of one or both of these risk factors were present and 0.1%
the 33 bleeding episodes were confirmed). This suggests (95% CI, 0.020.5%) if neither factor was present. Pa-
that, in this observational design, physicians were tients who had received prophylaxis and those who
probably able to predict who was at risk for bleeding had not were included. Simple regression analysis
and administer prophylaxis; it does not necessarily showed that sepsis, renal insufficiency, hepatic failure,
represent failure of prophylaxis. enteral feeding, and the use of glucocorticoids, heparin,
Prophylaxis in pediatric patients. The number of stress- and warfarin were risk factors for bleeding; multiple
induced lesions progressing to clinically important regression analysis showed that only mechanical venti-
bleeding with or without prophylaxis is not well stud- lation and coagulopathy were independently predic-
ied in pediatric patients. Lacroix et al.48 compared pro- tive of clinically important bleeding.
phylaxis with no prophylaxis in 40 patients from birth The finding that enteral feeding was a risk factor for
to 18 years of age. When the results were reported, overt bleeding (OR = 3.8, p < 0.001) is of interest because
bleeding was not distinguished from clinically impor- previous retrospective studies in adults have found
tant bleeding, and no differences in upper GI bleeding various effects of enteral feedings on pH and the fre-
were noted between the cimetidine group (9 of 19 quency of bleeding.72-75 These differences may be due to
patients) and the placebo group (8 of 21 patients). Six differences in location of the enteral feeding tubes. For
patients had massive bleeding, defined as the presence example, feedings placed into the stomach may alkalin-
of nasogastric blood with changes in blood pressure or ize stomach contents, and jejunal feedings may stimu-
hemoglobin levels. It was not stated which groups these late gastric acid secretion. The fact that patients are able
patients were in, and it was noted that five of these to tolerate tube feedings may mean they are less ill than
patients had bleeding at other sites due to disseminated patients who cannot tolerate such feedings, and this
intravascular coagulation. The researchers concluded may determine the differences noted. Other factors
that routine prophylaxis in medical pediatric patients is may be specific to certain types of injury, such as the
not warranted. development of GI edema after thermal injury.
In another study, 165 pediatric patients admitted to Conflicting associations with total parenteral nutri-
an ICU were randomly assigned to one of three active- tion (TPN) and bleeding have been noted. Ben-Men-
treatment groups or a no-treatment group.65 The inves- achem et al.59 found that the frequency of bleeding was
tigators used clinically important bleeding as an out- not related to the use of TPN. Ruiz-Santana et al.52
concluded that TPN was protective, but their sample mechanically ventilated preterm and full-term infants
size was small, and their findings require confirmation found that 20 (80%) of 25 patients had stress-induced
in larger randomized trials. lesions as determined by endoscopic evaluation when
The most recently published large-sample retrospec- no prophylaxis was administered.63 That study and
tive study of risk factors in ICU trauma patients involved others demonstrated that gastric lesions due to stress
2574 patients.76 The only patients receiving prophylaxis may occur in the very young despite their possible
(antacids) were those with a history of gastroduodenal physiological differences from adults.29,63
ulcerations. The published abstract does not reveal how The rate of progression of these lesions to clinically
many of the 2574 patients received antacids. Clinically important bleeding is not known. In a study in 881
important bleeding was not analyzed separately from GI pediatric ICU patients (1006 admissions), 16 (1.6%)
bleeding. GI bleeding was noted in 60 (2.3%) of the clinically important bleeding episodes occurred.66 Be-
patients. The risk factors associated with a higher risk of cause this was an observational investigation, physi-
bleeding included male sex, spinal cord injury, respirato- cians were not hindered from prescribing medications
ry failure, acute renal failure, and severe sepsis. for stress ulcer prophylaxis. Among the 110 patients
Of the other trials in which risk factors were studied, who received prophylaxis, there were 10 clinically im-
a majority of those conducted in adults showed a portant bleeding episodes. The authors noted that the
higher risk of bleeding as the number of risk factors three strongest independent risk factors for clinically
increased.37-39,42,45-47,54,55,58 The results of most trials are important bleeding were respiratory failure (OR = 10.2),
consistent with those of the 1994 study by Cook et al.61 coagulopathy (OR = 9.3), and a Pediatric Risk of Mortal-
in that mechanical ventilation (as a measure for respira- ity Score of 10 (OR = 4.0). Of the 16 patients with
tory failure)36,41,59 and coagulopathy36,41,53 are the prima- clinically significant bleeding, 9 (56%) had all three risk
ry risk factors for bleeding. Other risk factors associated factors and 14 (88%) had at least two.
with bleeding in adult patients in some studies include Summary. The reported frequency of stress-in-
sepsis,35 length of ICU stay greater than one week,41 duced bleeding has varied substantially since approxi-
presence of occult bleeding for at least six days,53 and mately 1978 and depends on the number and type of
administration of high doses of corticosteroids (>250 risk factors. Data from large prospective, observational
mg/day of hydrocortisone or equivalent).54,59 However, investigations in children and adults have found that
the studies by Cook et al. did not find use of corticoster- general medical and surgical ICU patients who have
oids, heparin, warfarin, or NSAIDs to be independent either coagulopathy or respiratory failure requiring me-
risk factors for bleeding when respiratory failure and chanical ventilation have a substantially higher risk of
coagulopathy were considered.53,61 clinically important bleeding. The studies often exclud-
Prophylaxis does not necessarily prevent bleeding in ed neurosurgery, burn, and transplant patients. Anoth-
patients with documented risk factors, and the effec- er risk factor for children is a Pediatric Risk of Mortality
tiveness of prophylaxis varies in different popula- Score of 10. A number of other risk factors have been
tions.59 Some potential risk factors for bleeding have associated with overt bleeding in isolated studies (pro-
not been adequately studied, and it is unclear whether longed ICU stay or occult bleeding and administration
they are independent predictors of bleeding. A majority of high-dose corticosteroids in adults; prolonged shock
of investigations excluded patients with a history of GI or surgery and trauma in children), but it is unknown
disease, particularly ulcers and bleeding. It is unknown whether these are independent predictors of clinically
whether a history of ulceration or bleeding increases important bleeding, except in the case of coagulopathy
the risk of acute, stress-induced bleeding. Similarly, or respiratory failure. Several studies that included both
ulcerogenic medications with well-documented associ- medical and surgical ICU patients suggest that, as the
ations with bleeding problems, such as long-term use of number of risk factors increases, the frequency of clin-
NSAIDs, have not been adequately studied as risk fac- ically important bleeding increases. It has been pro-
tors for stress-induced bleeding. Typically, patients re- posed that adult medical ICU patients may not neces-
ceiving such medications either were not discussed in sarily need stress ulcer prophylaxis, even if they have
the methods section of study reports or were prohibited respiratory failure or coagulopathy. However, there are
from enrolling in the studies. concerns related to inadequate sample sizes in the
In pediatric patients (birth to 19 years of age), coag- studies that resulted in this conclusion. Overall, accord-
ulopathy,48,57 shock,56 surgery lasting for longer than ing to RCTs, about 6% (range, 0.139%) of adult medi-
three hours,56 trauma,56,57 and pneumonia57 were associ- cal and surgical patients at risk for stress ulcers who do
ated with an increased risk of bleeding, while the use of not receive prophylaxis have clinically important
enteral nutrition reduced the risk of bleeding.57 The bleeding.64 Observational studies involving large num-
results were not stratified by age. One study in pediatric bers of pediatric or adult patients suggest that the risk
patients (median age, 34.9 months) found that a Pedi- may be substantially lower.61,66
atric Risk of Mortality Score greater than 10 was associ- Recommendation (adults): Risk factors for ICU patients have
ated with a greater risk of bleeding.57 Another study in been delineated in trials comparing prophylaxis with no
one study investigated the effect of no prophylaxis in The frequency of, and risk factors for, clinically
pediatric patients only.88 All 477 children had thermal important bleeding in patients with single-system inju-
injuries involving 13% or more of their BSA. Thirty-six ries such as head trauma, spinal cord injury, and ther-
(7.5%) of the children had clinically important bleed- mal injuries; patients undergoing transplantation; and
ing when no prophylaxis was given (or at least was not patients with multiple trauma have not been well stud-
mentioned). Some 32 of the 36 children (89%) with ied. These populations were usually excluded from
clinically important bleeding required blood transfu- studies in which patients might be randomly assigned
sions, and 5 patients (14%) required surgery; 1 death to receive no prophylaxis. It is not known whether
(2.8%) was associated with GI bleeding. respiratory failure or coagulopathy would serve as inde-
Another defined population studied for stress ulcer pendent predictors of bleeding in these populations.
prophylaxis includes patients (at least 14 years of age Recommendation (adults): Prophylaxis is recommended for
in published studies) undergoing renal or hepatic ICU patients with a Glasgow Coma Score of 10 (or the
transplantation or hepatic surgery.92-97 Cimetidine or inability to obey simple commands) or thermal injuries to
antacids were used for prophylaxis in all the trials >35% of their BSA. (Strength of evidence = B) ICU patients
except one investigation that used ranitidine.97 Of the with partial hepatectomy may also benefit from prophylaxis.
(Strength of evidence = C) Prophylaxis may also be indicated
studies evaluated, one was randomized. In all the in ICU patients with multiple trauma (e.g., Injury Severity
investigations, both randomized and nonrandomized, Score of 16), transplantation patients in the ICU perioper-
overt and clinically important bleeding were com- atively, ICU patients with hepatic failure, and ICU patients
bined. The frequency of bleeding was lower with ci- with spinal cord injuries. (Strength of evidence = D) (Table 3)
metidine prophylaxis in one study involving renal Recommendation (pediatrics): For pediatric patients (one
transplant patients but not in the other.93,94 month of age or older) with thermal injuries, prophylaxis is
Prophylaxis with ranitidine was associated with a recommended, but there is insufficient evidence to recom-
significantly lower frequency of endoscopically con- mend prophylaxis based on any given percentage of BSA.
(Strength of evidence = D) For other pediatric surgery or
firmed stress ulceration in the one available study of
trauma patients, insufficient evidence is available to allow
hepatic transplant patients, but the frequency of clini- recommendations about prophylaxis to be made.
cally important bleeding was not delineated.97 One RCT
involved patients with hepatic failure,95 and one pro- Medications used for prophylaxis
spective trial enrolled patients undergoing partial hepa- In general, medications used for preventing stress-
tectomy.96 The frequency of bleeding was lower with induced bleeding exert their pharmacologic effect
prophylaxis in both studies, but only the hepatectomy through one of the following mechanisms: inhibition
study contained adequate information about clinically of gastric acid secretion, neutralization of gastric acid
important bleeding to be able to conclude that it was secretions, and protective mechanisms unrelated to
significantly lower with prophylaxis. None of the stud- acid secretion or neutralization.103 The H2-receptor an-
ies looking at renal or hepatic surgery or disease inves- tagonists, prostaglandin analogues, and proton-pump
tigated the number or type of other risk factors poten- inhibitors act primarily as antisecretory agents by in-
tially associated with bleeding. Thus, direct evidence hibiting gastric acid secretion. But, other explanations
that concomitant organ disease or surgery increases the for the beneficial actions of these agents have been
risk of clinically important bleeding is lacking. proposed. For example, H2-receptor antagonists have
Several studies have included defined patient popu- immunomodulatory actions at the cellular and media-
lations other than those previously mentioned.98-102 tor levels.104 However, mediation of cytokine concen-
Only one of these investigations was randomized. trations is a complex process, and the cytokine concen-
That study involved patients with tetanus who had trations fluctuate widely in critically ill patients, often
tracheostomies, a majority of whom were not mechan- not correlating with more global physiological de-
ically ventilated.100 The frequency of overt bleeding rangements.105 Similarly, the prostaglandin analogue
was similar between the patients who did and did not misoprostol prevents gastric mucosal damage by irri-
receive prophylaxis, and there were no episodes of tants that do not appear to depend on acid or acid
clinically important bleeding. The frequency of bleed- secretion and is sometimes referred to as a cytoprotec-
ing and risk factors associated with bleeding were tive agent.106 There is no convincing evidence to estab-
evaluated in one retrospective study of 33,637 trauma lish that cytoprotection with misoprostol plays an
patients, most of them adults.101 It is unknown how important role in preventing stress-induced lesions.107
many of these patients received prophylaxis, and overt Although alternative explanations for the effectiveness
and clinically important bleeding episodes were com- of antacids in preventing stress-induced bleeding have
bined (18 of 33,367 patients bled). Risk factors associ- been proposed (e.g., stimulation of prostaglandin re-
ated with bleeding included an Injury Severity Score of lease), neutralization of gastric acid is the primary
16, a Glasgow Coma Score of <8, and spinal cord mechanism of action.108 In contrast, sucralfate has mul-
injury. Elderly patients and patients with infections tiple actions on the GI tract besides providing a direct
and organ failure were also more likely to bleed. protective barrier, including modulation of pepsin and
mucus activity, arachidonic acid metabolism, bicarbon- addition, lansoprazole has FDA-approved labeling for
ate secretion, and tissue growth or repair.109 Which, if administration by nasogastric tube with some fruit
any, of these mechanisms is of primary importance for juices.116 One open-label study involving 75 mechani-
stress ulcer prophylaxis is not clear. cally ventilated surgery patients found a compounded
The H2-receptor antagonists that are available com- omeprazole suspension effective in preventing stress-
mercially in the United States are cimetidine, famoti- induced bleeding.114 Larger comparative studies are
dine, nizatidine, and ranitidine. Only cimetidine ad- needed to determine efficacy and assess the potential
ministered as a continuous intravenous infusion has for adverse effects such as pneumonia (from increased
FDA-approved labeling for the prevention of bleeding gastric pH, as postulated for the H2-receptor antago-
in critically ill patients, although there is no reason to nists) and other unanticipated problems.
suspect any substantial difference in prophylactic effi- Although medications may have different mecha-
cacy when other H2-receptor antagonists are given in nisms for their protective effects, there is no evidence
equipotent doses.110 However, other reasons may pre- that combination therapy is more effective than a
clude their use, such as the lack of an i.v. dosage form single agent for initial prophylaxis, but the potential
(as with nizatidine). No RCTs of adequate sample size risk of adverse effects would increase with combination
have compared dosage regimens (e.g., intermittent ver- therapy. Therefore, combination therapy for initial
sus continuous i.v. administration) or routes (e.g., i.v. prophylaxis cannot be recommended. Combination
versus oral) of H2-receptor antagonists by using clinical- therapy for the prevention of recurrent bleeding is
ly important bleeding as a study endpoint. discussed further in this document (see Prevention of
One concern with using oral H2-receptor antagonists recurrent bleeding).
and proton-pump inhibitors in critically ill patients is
the potential for malabsorption. Because these agents Efficacy of prophylactic agents: Prevention
must be absorbed to exert their effects, the patient must of bleeding
have a functioning GI tract. A functioning GI tract is A substantial number of clinical trials, a majority of
usually presumed when tube feedings are tolerated which were randomized, have involved comparisons of
without nausea, vomiting, abdominal distention, or medications used for stress ulcer prophylaxis.117-174 In
diarrhea. In patients with nasogastric tubes, the clini- general, no significant differences between medica-
cian may monitor the gastric residual volume when tions used for the prophylaxis of clinically important
feeding formulas are administered into the stomach. bleeding have been found. However, in a majority of
Residuals refer to the feedings and gastric secretions studies, the sample size needed to detect a difference
that accumulate with GI dysfunction. A volume of between treatment groups in clinically important
more than 200 mL or an amount greater than the bleeding was not calculated. Therefore, the number of
volume infused over a two-hour period is usually indic- patients enrolled in any given study was often insuffi-
ative of dysfunction.111 cient to preclude a Type II statistical error. Other studies
Although widely used for stress ulcer prophylaxis, have found substantial differences in bleeding between
sucralfate is not labeled for this indication. Because su- groups, but the frequency of bleeding was unexpectedly
cralfate appears to exert its predominant effects locally, it high for one of the groups. For example, Phillips et al.175
must be administered either orally or by a nasogastric, reported clinically important bleeding (the definition
orogastric, or gastrostomy tube. Sucralfate should not be of clinically important was different than the definition
administered with nasogastric tube feedings (see Adverse used in these guidelines) in 4 (16%) of 25 patients
effects of prophylactic agents). Neither should it be ad- receiving ranitidine by continuous i.v. infusion and 1
ministered by feeding duodenal or jejunostomy tubes (3%) of 33 patients receiving omeprazole suspension.
because the medication will miss the site of action. Compared with the results of other studies, 16% is a
Similar to sucralfate, antacids have local actions and high rate of bleeding with ranitidine.64,176
must be administered directly into the stomach. A In the largest RCT published to date, 1200 patients
number of antacid products have been routinely used who required mechanical ventilation for at least 48
for stress ulcer prophylaxis. Several formulations carry hours were assigned to receive either nasogastric sucral-
FDA-approved labeling for this indication. fate slurry (1 g every 6 hours) and i.v. placebo or i.v.
Misoprostol is available in tablet form, while proton- ranitidine (50 mg every 8 hours) and nasogastric place-
pump inhibitors such as omeprazole and lansoprazole bo.176 The patients in this multicenter study were ad-
are available in delayed-release capsules. These medica- mitted to medical and surgical services (approximately
tions have not been well studied for stress ulcer prophy- 60% and 40%, respectively). There were relatively few
laxis and are not labeled for this indication. The de- patients from special populations: CNS surgery, 49
layed-release dosage forms of the proton-pump inhibi- patients; trauma, 158; burns, 12; and transplantation,
tors create administration problems when oral intake is 19. These special populations were not discussed sepa-
not possible, although some interesting compounding rately. Although the estimated sample size was based
techniques for omeprazole have been reported.112-115 In on the power to detect differences in rates of pneumo-
nia (power, or 1 = 75%), the rates of clinically impor- H2-receptor antagonists were found to be significantly
tant bleeding and mortality were additional study ob- more effective than antacids in preventing overt bleed-
jectives, and the trials power to detect a difference in ing but not clinically important bleeding. Sucralfate
bleeding rate was 90%. A significant difference in favor was similar to antacids or H2-receptor antagonists for
of the H2-receptor antagonist (ranitidine) compared prophylaxis of both overt bleeding and clinically im-
with sucralfate was noted. Ten (1.7%) of the 596 pa- portant bleeding.
tients in the ranitidine group and 23 (3.8%) of 604 While useful, this meta-analysis raises some impor-
patients in the sucralfate group had clinically impor- tant questions.5 For example, the definitions for overt
tant bleeding (relative risk, 0.44; 95% CI, 0.210.92; p = and clinically important bleeding are very explicit,
0.02). Endoscopic confirmation of an upper GI source which contrasts to the definitions used in many of the
of bleeding was obtained for 3 patients in the ranitidine randomized investigations that were analyzed. It is not
group and 10 in the sucralfate group. An additional possible to derive rates of clinically important bleeding
patient in the ranitidine group had an upper GI source from many of the trials, particularly if a rigid definition
of bleeding confirmed by angiography and red blood is used. The meta-analysis also included RCTs from a
cell scanning. Because invasive diagnostic testing was variety of populations that have usually been studied as
left to the discretion of participating physicians, the distinct populations (e.g., patients with head and ther-
source of bleeding was not found for 19 patients. In mal injuries). Even if one assumes that the pathogenesis
addition, no differences were noted in the overall of bleeding is similar, there may be differences in the
length of stay or mortality. Therefore, the largest RCT frequencies of bleeding compared with populations
conducted to date has not definitively determined the studied more frequently (e.g., general medical and sur-
agent of choice for the prevention of clinically impor- gical ICU patients). If the frequency of bleeding has
tant GI bleeding. decreased over time as a result of improvements in the
Five meta-analyses have been conducted to resolve care of the ICU patient (which is difficult to verify
some of the discrepancies noted between RCTs, includ- because many recent studies have not included a no-
ing the issue of inadequate sample size.5,68-71 None of treatment group), this could have also influenced the
these meta-analyses included the recently published results of the meta-analysis. Finally, the results of this
RCT by Cook et al. (1998) and the Canadian Critical meta-analysis are somewhat inconsistent with the find-
Care Trials Group (Table 2).176 The other major limita- ings of the large RCT recently published by the same
tions of these analyses have been discussed previously. lead authors with regard to possible differences in med-
In general, the meta-analyses found H2-receptor antag- ication efficacy.176
onists and antacids to be more effective than no pro- Recommendation (adults): Given the conflicting results of
phylaxis, but in some cases the 95% CIs were wide, several meta-analyses and a recent RCT (both with strengths
indicating substantial variation in the data. Only one of evidence = A), the choice among antacids, H2-receptor
trial compared sucralfate with no prophylaxis, and this antagonists, and sucralfate for use as prophylactic agents to
trial was included in the 1996 meta-analysis by Cook et prevent clinically important bleeding associated with stress
in adult patients admitted to general medical and surgical
al.5 The results of these meta-analyses were conflicting ICUs should be made on an institution-specific basis. This
when medications were compared. In the meta-analy- choice should take into account concerns regarding adminis-
ses by Lacroix et al.68 and Tryba,69 H2-receptor antago- tration (e.g., functioning GI tract), adverse-effect profile, and
nists were compared with antacids, but the 95% CIs total costs. (Strength of evidence = D) Insufficient data on
included 1, indicating that the two regimens did not misoprostol or the proton-pump inhibitors are available to
differ significantly. In the 1991 meta-analysis by Cook allow any recommendation about these agents to be made.
et al.,70 H2-receptor antagonists were significantly more Recommendation (pediatrics and special populations): The
effective than antacids for preventing overt bleeding, lack of comparative trials of these agents in pediatric and
special populations (e.g., patients with burns, trauma pa-
but the therapies were similar in preventing clinically
tients, patients undergoing neurosurgical procedures or with
important bleeding. neurologic disorders, and transplantation patients) precludes
Tryba71 found sucralfate to be significantly more definitive recommendations as to the agent of choice in these
effective than H2-receptor antagonists and similar in situations. The choice of agent should be made on an institu-
efficacy to antacids (OR = 0.532; 95% CI, 0.3030.933). tion-specific basis and should take into account concerns
Cook et al.70 (1991) found sucralfate to be similar to H2- about administration (e.g., functioning GI tract), adverse-
receptor antagonists or antacids for preventing overt effect profile, and total costs.
and clinically important bleeding.
Given the disparities noted in the previous meta- Efficacy of prophylactic agents: Mortality
analyses, Cook et al.5 (1996) conducted a meta-analysis Many of the available studies did not include mortal-
to resolve the apparent discrepancies. This meta-analy- ity outcomes, and still fewer specifically associated
sis included studies that were not available when the mortality with bleeding. In the largest RCT to date,
initial meta-analyses were completed. In addition to there was not a significant difference in mortality be-
being significantly more effective than no prophylaxis, tween ranitidine recipients and sucralfate recipients
sorption of other medications). When intermittent liver. Although a number of interactions with some of
feedings are used, interactions can be minimized by these acid-suppressing agents are of potential concern
giving the feedings and medications at different times. (e.g., warfarin, ketoconazole, phenytoin, theophylline),
When continuous feedings are used in conjunction few are clinically important when patients are adequate-
with any medications, the feedings should be tempo- ly monitored.200 However, there may be other reasons for
rarily stopped. The tube should be flushed with water avoiding particular medications. For example, cimeti-
before the medication is administered. After the medi- dine may increase serum creatinine concentrations with-
cation is administered, the tube should again be flushed out affecting the glomerular filtration rate by competi-
with water to ensure passage through the tube. It is tion for tubular secretion in the kidneys.201 This could be
probably unnecessary to discontinue postpyloric (jeju- interpreted as renal dysfunction in patients receiving
nal) feedings for administration of sucralfate, because nephrotoxic medications, leading to inappropriate re-
sucralfate binds to the stomach mucosa. For more de- ductions in the dosage of the suspected medication.
tails about enteral feedingmedication interactions and Other interactions may occur as a result of the acid-
methods to prevent them, the reader is referred to suppressing properties of these two classes of agents.
published reviews.111,186 Sucralfate and antacids can affect the absorption of
Case reports suggest that sucralfate may contribute other medications. Although most problems can be
to esophageal and GI bezoar formation, especially avoided if doses of sucralfate and antacids are separated
when given in conjunction with enteral feedings in from doses of potentially interacting substances, this
patients with motility disorders.187-190 In addition to practice is often not used.202 Potentially interacting med-
technical problems related to the removal of bezoars, ications should be administered one to two hours before
complications such as intestinal obstruction and perfo- the sucralfate dose. Furthermore, medications may inter-
ration have been reported in pediatric patients. For act with sucralfate and antacids despite separation (e.g.,
example, intestinal obstruction and perforation were the bioavailability of quinolone antimicrobials has been
reported with sucralfate in a low-birth-weight infant substantially reduced by sucralfate even when the two
who was given 250 mg every eight hours by nasogastric products were separated by two or more hours).203,204
tube.191 Also, bezoars, hypermagnesemia, and intestinal Test interference is a cause for concern with all
perforation have been reported with liquid antacids medications used for stress ulcer prophylaxis. This con-
used in premature infants and newborns.192,193 There- cern also applies to agents that do not alter pH (e.g.,
fore, sucralfate and antacids should be used with cau- sucralfate interferes with at least one method of occult
tion, particularly in premature infants. blood testing).205
The H2-receptor antagonists are associated with a A full discussion of all potential drugdrug interac-
number of adverse effects that occur infrequently but tions with these agents is beyond the scope of these
may lead to a consultation with a specialist, a diagnos- guidelines; however, comprehensive reviews have been
tic test, a new medication to treat the adverse effect, or published.206
a switch to another medication for prophylaxis. Poten- The proton-pump inhibitors have few clinically im-
tial adverse effects include hepatitis, cytopenia, and portant adverse effects. The most common concerns are
cardiovascular and CNS toxicities. The exact frequency mild GI or CNS toxicities.207 The prostaglandin ana-
of each of these problems is difficult to determine, but logue misoprostol is rarely used clinically for prophy-
estimates range from 0.0023% for cytopenia to 1.9% for laxis and is associated with substantial adverse effects
CNS toxicity.194-199 The low rate of these adverse effects compared with placebo (nausea, abdominal pain, diar-
in adults and limitations associated with postmarketing rhea).208
surveillance studies prohibit accurate comparisons of Lansoprazole (pregnancy risk category B) and ome-
the frequencies of such effects among medications in prazole (FDA pregnancy risk category C) should be used
this class. The lack of studies in the pediatric population in pregnant or nursing mothers only when the poten-
precludes an accurate assessment of the frequency of tial benefits exceed the potential risks.116,209 Similar
particular adverse effects. There is a possibility that caution is advised for the H2-receptor antagonists (preg-
pediatric patients may be more susceptible to some nancy risk category B for cimetidine, famotidine, and
adverse effects (e.g., thrombocytopenia associated with ranitidine; pregnancy risk category C for nizatidine)
cimetidine49) than adults. Similarly, the lack of infor- and sucralfate (pregnancy risk category B).110,210-213 Mis-
mation related to pediatric dosing could result in either oprostol is an abortifacient, may result in miscarriage
underdosing with diminished efficacy or overdosing (pregnancy risk category X), and is generally contrain-
with increased adverse effects. dicated in pregnant women, women of childbearing
H2-receptor antagonists and proton-pump inhibitors age, and nursing mothers.214
have the potential for drugdrug, drugnutrient, and Pneumonia. The frequency of nosocomial pneu-
drugtest interactions through a variety of mechanisms. monia associated with stress ulcer prophylaxis has been
For example, some drugdrug interactions are mediated widely debated. In addition to patient morbidity and
through effects on the cytochrome P-450 systems in the economic concerns, mortality rates associated with
pneumonia may be as high as 70%.215 There are multi- or antacids.223 Patients gastric and respiratory tract
ple mechanisms by which bacteria may enter the lungs, contents were cultured for routine aerobic bacteria and
including inhalation, hematogenous spread, contigu- fungi before surgery and daily throughout the hospital-
ous spread, direct inoculation, aspiration of secretions, ization until the patient was discharged or died. Clini-
and translocation from the GI tract.216 The vast majority cal endpoints included upper GI tract bleeding (grossly
of studies investigating a possible connection between bloody fluid in nasogastric aspirate that failed to clear
pneumonia and stress ulcer prophylaxis have been with saline lavage), pneumonia, wound infection, and
conducted in adults. In the few studies involving pedi- urinary tract infection. New gastric microorganisms
atric patients receiving stress ulcer prophylaxis, pneu- appeared in 24% of the patients receiving sucralfate and
monia was usually not studied. When pneumonia was 51% of the patients receiving antacids. The rates of
reported, it was typically not a primary endpoint.65 gastric bleeding, pneumonia, and nosocomial infection
Early reports of adult patients receiving antacids and were not significantly different.
cimetidine demonstrated that the stomach and upper Other risk factors for nosocomial pneumonia have
airways had similar bacterial flora, suggesting retro- been identified and ranked according to supporting evi-
grade transmission from the stomach.217 Gastric coloni- dence.224 In addition to host factors (e.g., gastric hypoa-
zation was uniformly seen in patients with alkalinized cidity) and environmental factors, the use of a nasogas-
stomach contents.218 However, a positive correlation tric tube to administer medications (with the attendant
between increased gastric pH and colonization with risks of aspiration and guaiac-positive stools from naso-
bacteria does not mean there is a linear relationship gastric manipulation), along with the volume and pH of
between acid-suppressing drugs and pneumonia. medications and feedings administered, determines the
It is also important to keep in mind that the tech- frequency of pneumonia. However, despite these other
niques used to diagnose pneumonia have continued to risk factors, a committee associated with the Centers for
evolve. Bronchoscopy, which is now a widely used and Disease Control and Prevention has recommended that
accepted diagnostic tool in research, was not performed non-pH-altering medications (e.g., sucralfate) be used for
in older studies of medication-associated pneumonia. prophylaxis in mechanically ventilated patients (ages
Although the sensitivity and specificity of bronchosco- not specified).225 The recommendation was suggested
py for diagnosis of infection have been improved by for implementation in many hospitals. While sucralfate
newer techniques such as use of a protected-specimen may not contribute to translocation or colonization of
brush, morbidity and economic concerns have been the airways by bacteria, animal models of aspiration
expressed about its widespread clinical use.219 Other pneumonia have demonstrated that the volume of fluid
methods are less invasive but often have lower sensitiv- administered with sucralfate and the drugs minimal
ity and specificity, particularly when not performed buffering activity may increase the risk of acid aspiration
appropriately. and pulmonary edema.226
Agents that do not alter gastric pH, such as sucralfate, In the RCT by Cook et al.176 that involved 1200
may be less likely to result in pneumonia through retro- patients, the determination of ventilator-associated
grade transmission or translocation of bacteria.220,221 pneumonia was based on criteria modified from the
When protected specimens, bronchoalveolar lavage, or Centers for Disease Control and Prevention. The study
both were used, one prospective randomized, double- had sufficient power to detect a 25% difference in rates
blind study involving 141 patients showed no differ- of pneumonia (power = 75%). Bronchoalveolar lavage
ence in the effect of antacids or sucralfate on gastric or protected specimen brush sampling was used for
acidity or rates of bacterial colonization or ventilator- diagnostic confirmation in patients with pneumonia
associated pneumonia.145 Decreased gastric acidity was suspected by clinical criteria. One potentially con-
observed in conjunction with greater colonization of founding variable was the use of enteral feedings by a
the stomach but not of the upper respiratory tract. In majority (approximately 70%) of patients in each
another prospective investigation in which samples group. Pneumonia was diagnosed in 114 (19.1%) of 596
were obtained by bronchoscopy, mechanically venti- patients in the ranitidine group and 98 (16.3%) of 604
lated patients were randomly assigned to receive either patients in the sucralfate group, a difference that was
antacids or sucralfate and then studied for risk factors not significant (relative risk, 1.18; 95% CI, 0.921.51; p
associated with colonization and pneumonia.222 As in = 0.19). It is possible that a significant, albeit small,
the previous study, gastric colonization was not associ- difference in favor of sucralfate would be found if a
ated with the development of pneumonia, but duration larger trial were conducted. There were five patients
of mechanical ventilation and upper-respiratory-tract with definite pneumonia in the ranitidine group
colonization were found to be risk factors. A more (0.8%) but none in the sucralfate group (95% CI, 0.1
recent study on this issue had a double-blind, double- 1.6; p = 0.03).
dummy design and compared the growth of gastric Five meta-analyses of the frequency of nosocomial
pathogens in 140 patients undergoing elective surgery pneumonia associated with stress ulcer prophylaxis
who were randomly assigned to receive either sucralfate have also been conducted (Table 5). Two of the investi-
Table 7.
Studies of Stress Ulcer Prophylaxis in Pediatric Populationsa
Study Design and Population No. with Clinically
Reference (No. Patients or Admissions) Age Medication Important Bleeding
88 Retrospective; thermal injury 35 days to 13 yr; None stated 36
to 1393% (mean, 29%) mean, 4.3 yr
BSA; (477)
230 Retrospective; thermal injury 4 wk to 15 yr Milk or feeding formula plus 2
to 592% BSA (582) diazepam
48 Prospective, randomized; Birth to 18 yr; mean, Cimetidine 5 mg/kg i.v. NA
general ICU (40) 1.85 yr every 6 hr (1 g/day
maximum)
231 Prospective, randomized; 3041.5 wk Cimetidine 6 mg/kg i.v. NA
general ICU (33) every 6 hr
232 Prospective, randomized; Birth to 18 yr; mean, Ranitidine 24 mg/kg NG NA
general ICU (40) 2.64.3 yr (varied by every 12 hr or 0.751.5
group) mg/kg i.v. every 6 hr
49 Prospective, randomized; Full-term neonates Cimetidine 3.755 mg/kg NA
general ICU (100) i.v. every 6 hr
65 Prospective, randomized; Birth to 20 yr; mean, Ranitidine 1.5 mg/kg i.v. 7/35 in control group
general ICU (140) 4.6 yr every 6 hr, sucralfate 0.5 versus 6/105 in
g (<10 kg) or 1 g (>10 combined medica-
kg) every 6 hr, antacid tion groups (p <
(almagate) 0.250.5 mL/ 0.05); no significant
kg every 2 hr differences for any
single medication
versus control group
56 Prospective, observational; Mean S.D., None None
general ICU (208) 38 53.5 mo
57 Prospective, observational; Mean S.E., Various (per discretion of 2/698 with no prophy-
general ICU (984) 62.6 2.1 mo physician) laxis versus 2/286
with prophylaxis
168 Prospective, randomized; 35 days to 9.85 yr; Famotidine or pirenzepine NA
cardiac surgery (79) mean, 2.62 yr 1 mg/kg/day i.v. (3 doses
if <10 kg), 0.5 mg/kg/day
if serum creatinine > 3
mg/dL
63 Prospective, randomized; Neonates, mostly Ranitidine 1.67 mg/kg i.v. NA
general ICU (53) preterm; mean, every 8 hr
32 wk gestation
233 Prospective, observational; 2 wk to 264 mo; Ranitidine 26 mg/kg/day None
general ICU (45) median, 36 mo i.v. (mean, 3.3 mg/kg/
day) given at intervals of
6, 8, or 12 hrb
66 Prospective, observational; 3 days to 18 yr; mean, Various (per discretion of 6/990 with no prophy-
general ICU (1006) 61.5 mo physician) laxis versus 10/110
with prophylaxis
a
Only studies with at least 25 patients were included. BSA = body surface area, ICU = intensive care unit, NA = not available, NG = by nasogastric tube.
b
Administered every eight hours in 89% of patients.
larger volumes of distribution than adults and that on pulmonary vasculature.242 In general, the use of ant-
doses should be given either by continuous infusion or acids and sucralfate should be avoided in premature
more frequently than those recommended for adults. infants because of the possibility of complications (see
Studies addressing the proper dosing of agents for stress Adverse effects of prophylactic agents).
ulcer prophylaxis in pediatrics are needed. Doses of H2-receptor antagonists are often reduced
In general, there is little dosing information for the (or the intervals extended) for patients with renal dys-
neonatal population, in particular, premature neo- function, because all these medications have slower
nates. Dosing in premature neonates is further compli- clearances and longer half-lives in this population.243
cated by decreased renal function secondary to disease These alterations in dose or interval have been based on
or lack of maturation. Caution should be exercised expected serum drug concentrations, because well-con-
when H2-receptor antagonists are used in combination trolled studies have not investigated various dosage
with tolazoline. Tolazoline may still be used in some regimens in preventing stress-induced bleeding.
institutions to treat pulmonary hypertension in neo- Whether doses or intervals of H2-receptor antagonists
nates and has been associated with gastric bleeding. Case need to be adjusted in patients with severe hepatic
reports have implicated both cimetidine and ranitidine dysfunction or other diseases that might affect the
with reversal of tolazolines histamine-mediated effects pharmacokinetic properties of the agents is not known.
Table 8.
Pediatric Dosage Recommendations for H2-Receptor Antagonists and Sucralfate
Dosagea
Medication Neonates Infants Children
Cimetidine48,49,231,236,237 10 (520) mg/kg/day i.v. in 2040 mg/kg/day i.v. 3040 mg/kg/day i.v.
2 or 3 divided dosesb in 3 or 4 divided doses (1200 mg/day maximum)
in 6 divided doses
Famotidine168,240,241 Inadequate information 1.2 (12) mg/kg/day i.v. 1.6 (12) mg/kg/day i.v.
available to make dosage in 2 or 3 divided doses (40 mg/day maximum)
recommendations in 3 divided doses
Ranitidine63,65,232,233,235,238,239,244,245 Continuous i.v. infusion: 4.5 (36) mg/kg/day i.v. 6 (36) mg/kg/day i.v.
0.06250.2 mg/kg/hr in 3 divided doses (200 mg/day maximum)
Intermittent i.v. doses: If full in 4 divided doses
term, 1.53 mg/kg/day in 3
divided doses; if premature,
0.5 mg/kg/day in 2 divided
doses
Sucralfate65,234 Not recommendedc 40 mg/kg/day p.o. or NG 4080 mg/kg/day p.o. or NG
in 4 divided dosesd (lower end of range for
younger children, 4000
mg/day maximum)
in 4 divided dosesd
a
Dosages are for patients with normal renal function. NG = by nasogastric tube. There is a lack of data to make definitive dosage adjustments in pediatric
patients for renal dysfunction, but decreases in the total daily dosage of cimetidine, famotidine, and ranitidine would be logical. This could be done either by
decreasing individual doses while keeping the dosage interval the same as for patients with normal renal function or by extending the dosage interval and
keeping individual doses the same as for patients with normal renal function. No adjustment in the dosage of sucralfate is required for patients with renal
dysfunction.
b
In general, half-lives were longer in neonates (23.4 hours), more than likely because of immature renal function. There is inadequate dosage information
on premature neonates with respect to stratification by birth weight or gestational age. Thus, a wide range of dosages have been reported in the literature.
Many dosage handbooks and institutions use 510 mg/kg/day divided every 12 hours in premature neonates.
c
Inadequate information is available. Not recommended because of reports of bezoar formation in neonates (particularly premature neonates). Amount of
aluminum in adult doses of sucralfate may contribute to hypophosphatemia in neonates.
d
Dosage not well established. Bezoar formation with obstruction has been reported. Caution is advised when sucralfate is administered to any pediatric
patient, especially in cases of impaired gastric motility and dehydration.
even if dosages of antisecretory agents are individual- may be harmful. In the case of acid neutralizers, increas-
ized.256,257 Gastric pH can also be increased by the pres- ing gastric pH allows for bacterial overgrowth, with
ence of blood and bile in the GI tract. Although newer subsequent spread to other organs.261
methods of pH monitoring, such as noninvasive, intra- As discussed previously, the effect of enteral feedings
mucosal pH measurements, may prove useful for pre- on the frequency of stress-induced bleeding and pneu-
dicting stress-induced bleeding, such measurements monia has not been established through controlled
have been studied primarily as indicators of systemic trials. Although gastric acid appears to play a role in
tissue oxygenation and hence of overall mortality rates stress-induced bleeding, and while it is clear that micro-
in critically ill patients.258 bial growth is inhibited at low gastric pH, it has not
Checking for microscopic bleeding by occult blood been conclusively demonstrated that continuous or
testing techniques has similar limitations. Most impor- intermittent enteral feedings consistently raise gastric
tant, there is no proven relationship between micro- pH.262 The site of enteral feeding administration (stom-
scopic bleeding and outcomes such as clinically impor- ach or sites distal to the pylorus) varied in published
tant bleeding, although the results of one prospective studies, and potential site-specific effects remain to be
study suggest that occult bleeding for at least six days elucidated. The possibility that enteral feedings offer
increases the risk for overt bleeding.53 Overt bleeding is protection against bleeding and pneumonia through
defined as the presence of blood or coffee-ground other mechanisms (e.g., maintenance of a positive ni-
material in nasogastric aspirates, hematemesis, hema- trogen balance that is important for normal reparative
tochezia, or melena. Overt bleeding that results in functions of the gastric mucosa) should be explored.
hemodynamic compromise or the need for blood trans- Appropriate supportive care, such as administration
fusions or surgery is clinically important. Therefore, of fluids and blood products, nutritional support, ven-
patients at risk for stress-related complications should tilatory support, and timely definitive surgery, may
be routinely followed for overt bleeding. preclude the need for stress ulcer prophylaxis in many
When overt bleeding occurs, the patient should be patients.263,264 Before the widespread use of H2-receptor
closely monitored for decreases in blood pressure (e.g., antagonists for stress ulcer prophylaxis, a number of
a decrease of 20 mm Hg or more in the diastolic pressure regimens were used to prevent acute upper GI bleeding
during a 24-hour period), hemoglobin (e.g., 3 g/dL in burn patients, including milk and diazepam given to
during a 24-hour period), and hematocrit (e.g., 9% children and vitamin A injection given to adults.230,264
during a 24-hour period), all of which suggest clinically Although corticosteroids have been studied as inciting
important GI bleeding in the absence of other causes. factors for GI bleeding, other evidence suggests that
When overt bleeding, with or without hemodynamic corticosteroids may protect against bleeding in some
compromise, continues despite conservative therapies populations, including patients with shock.265
such as saline lavage, endoscopy is usually indicated. Combinations of allopurinol and dimethyl sulfox-
Endoscopy can be used to confirm a GI source of ide have been studied in two RCTs in patients with leg
bleeding. or pelvic fractures who had clinical evidence of
Recommendation (adults and pediatrics): It is recommended shock.166,266 These agents were studied because of their
that all patients receiving medications for stress ulcer pro- ability to decrease free-radical formation and, there-
phylaxis be monitored for bleeding and adverse drug effects fore, potentially prevent gastric mucosal injury. Both
(see Adverse effects of prophylactic agents). Paper techniques trials showed significant (p < 0.01) decreases in endo-
for measuring gastric pH have questionable validity and scopically confirmed lesions. Both of these studies are
reliability, and there is no evidence that adjusting the dosage
of pH-altering medications (antacids, H2-receptor antago-
promising but require confirmation in different popu-
nists, proton-pump inhibitors) on the basis of these measure- lations. Comparisons with existing agents that include
ments influences patient morbidity or mortality. Despite the economic evaluations of the options are also needed.
lack of supporting data, pH monitoring for antacids may be Recommendation (adults and pediatrics): It is premature to
appropriate (goal pH of >3.54). Such monitoring may also recommend the use of novel therapies (e.g., free-radical
be useful for H2-receptor antagonists when standard dosage scavengers) in place of conventional agents for stress ulcer
regimens might not be appropriate (e.g., in cases of renal prophylaxis, although the limited number of studies have
dysfunction, for increased dosages due to perceived failure of had promising results. (Strength of evidence = D)
therapy, in pediatric patients). (Strength of evidence = D)
ously without intervention. In others, the patients ically important bleeding upon extubation or the pa-
bleeding often stopped and did not recur when conser- tients discharge from the ICU. Thus, it is reasonable to
vative measures were instituted, usually an increase in assume that prophylaxis can be stopped once risk fac-
the dosage of the current medication, a switch to a tors have resolved.
different medication, or the addition of another agent, Recommendation (adults): Stress ulcer prophylaxis is not
plus gastric lavage. No large controlled trials have com- recommended for adult patients in non-ICU settings. (Strength
pared the various medication options for preventing a of evidence = B for general medical and surgical patients with
recurrence of stress-induced bleeding. However, when fewer than two risk factors for clinically important bleeding;
combination therapy for prophylaxis is being consid- strength of evidence = D for patients with two or more risk
factors) (Table 3)
ered in a patient who bled while receiving single-agent
prophylaxis, use of drugs with different mechanisms of Recommendation (pediatrics): Stress ulcer prophylaxis is not
recommended for pediatric general medical and surgical
action would seem logical. Potential interactions be-
patients or special populations (e.g., transplantation) in non-
tween prophylactic agents from different classes have ICU settings if fewer than two risk factors for bleeding are
not been well studied but could occur (e.g., reduced present. (Strength of evidence = D). Data are insufficient to
activity of sucralfate with neutral or alkaline gastric pH, allow recommendations to be made about the use of prophy-
binding of agents by sucralfate). Providing any defini- laxis in pediatric patients with two or more risk factors. If
tive recommendation regarding this issue is difficult. prophylaxis is given, it should be discontinued once risk
factors have resolved. (Strength of evidence = D)
The assumption is, of course, that any medication is of
benefit in preventing recurrent bleeding.
Economic analysis
Recommendation (adults and pediatrics): The lack of avail-
Inappropriate stress ulcer prophylaxis can be costly,
able trials prohibits definitive recommendations for pre-
venting recurrent bleeding after an episode of stress-induced but implementing guidelines may decrease the inap-
GI bleeding, although consideration could be given to in- propriateness and the expense.270,271 Published econom-
creasing the dosage of the prophylactic agent, adding an- ic analyses can be useful for developing institution-
other medication, or switching to a different agent. (Strength specific guidelines for prophylaxis. Two cost-effective-
of evidence = D) ness analyses have been completed concerning prophy-
laxis for stress-induced bleeding. In the first investiga-
Prophylaxis in non-ICU settings
tion, by Schumock et al.,272 an antacid regimen of 30 mL
It is possible that patients in non-ICU settings could given every 4 hours was compared with sucralfate 1 g
have coagulopathy or other conditions that have been given four times daily and three H2-receptor antago-
identified as risk factors for bleeding (e.g., transplanta- nists (ranitidine 50 mg i.v. every 8 hours, cimetidine
tion). Only one RCT has addressed stress ulcer prophy- 300 mg i.v. every 6 hours, and famotidine 20 mg i.v.
laxis in non-ICU settings.54 The study involved 100 every 12 hours), all of which were based on a seven-day
adults and was conducted on a general medical and course of therapy. The study was conducted from the
surgical hospital ward in Spain. Bleeding was signifi- viewpoint of the payer, with hospital charges measured
cantly (p < 0.001) less frequent with antacid prophylax- in 1994 dollars. The authors assumed that the frequen-
is than with placebo. However, bleeding was defined cy of bleeding would be approximately 70%, 80%, and
broadly to include patients who had a decrease in 36% lower with the use of sucralfate, antacids, and H2-
hematocrit, regardless of whether they required trans- receptor antagonists, respectively, and that it cost
fusion. Three patients required transfusion in the place- $6128 to treat an episode of bleeding. The frequency of
bo group and none in the antacid group. Upon institu- pneumonia was assumed to be unchanged with sucral-
tion of antacid and H2-receptor antagonist therapy, the fate but increased by approximately 7% with antacids
bleeding stopped in the three patients in the placebo and 13% with H2-receptor antagonists. The net saving
group without subsequent complications. The frequen- associated with prevention of one case of acute upper
cy of all types of bleeding (e.g., overt, clinically impor- GI bleeding was $7373 for sucralfate and $4321 for
tant) was significantly lower with prophylaxis only if antacids. In contrast, therapy with an H2-receptor an-
two or more risk factors for bleeding were present before tagonist would cost between $6655 and $7986 to pre-
prophylaxis was initiated. vent one episode of acute upper GI bleeding. The
difference in costs between sucralfate or antacids and
Discontinuation of prophylaxis H2-receptor antagonists was related to the assumed
An increase in the number of risk factors has been higher rates of bleeding and nosocomial pneumonia
associated with a higher risk of bleeding in multiple with H2-receptor antagonists. One-way sensitivity anal-
studies conducted in the ICU setting and one study ysis showed that sucralfate remained the preferred
conducted on a hospital ward. The risk of clinically agent regardless of the cost of prophylaxis, the cost of
important bleeding would probably diminish as the treatment for pneumonia or bleeding, or the presumed
number of risk factors is reduced. In most clinical trials, frequency of bleeding or nosocomial pneumonia.
prophylaxis was discontinued without evidence of clin- The second cost-effectiveness study, by Ben-Men-
achem et al.,64 was based on the perspective of the reflect laboratory or consultation estimates that would
health care provider. Cimetidine (300 mg i.v. every six be incurred even if the problem were eventually found
hours) was compared with sucralfate (1 g every six to be unrelated to the medication. One hundred pa-
hours), both given for seven days. One notable differ- tients were assumed to be in each treatment group in
ence from the previous analysis was the assumption our decision model because of the small probabilities of
that it would cost $595 to treat a GI bleed. It was some of the events. It was assumed that the number of
presumed that both therapies would decrease the fre- events would not vary with the number of patients.
quency of bleeding by 50%; in the sensitivity analysis, Figure 1 is the decision tree used for the economic
the percent reduction was varied over the range of 10 analysis associated with these guidelines. The overall
90%. Despite different assumptions from the investiga- results are consistent with those of other published
tion by Schumock et al., the results were similar in that cost-effectiveness analyses that concluded that sucral-
sucralfate (cost of $1144 per bleeding event averted) fate is the most cost-effective agent for prophylaxis,
was more cost-effective than cimetidine ($7538 per with a saving of $4913 for each episode of bleeding
bleeding event averted), even when rates of pneumonia averted. The cost-effectiveness ratio for ranitidine was
were presumed to be equal. lower, with a cost saving of $1766. Antacids were nearly
In one study involving a small number of patients, cost neutral. The differences are particularly notable if
the costs of enteral and i.v. ranitidine administration the pH-altering medications are assumed to cause a
were compared, although only enteral administration higher rate of nosocomial pneumonia (0.8% for acid
was used.273 Ten patients were assigned to receive raniti- suppressants and 0 for sucralfate); this is true regardless
dine 150 mg every 12 hours and eight patients to receive of the route of administration for the H2-receptor antag-
300 mg every 12 hours; both dosages were given by onists. If it is assumed that none of the medications
nasogastric tube. Both therapies reduced stimulated cause adverse effects (including pneumonia) and that
acid secretion by at least 50%. Drug costs were calculat- they are similar in efficacy, sucralfate and orally admin-
ed to be at least 48% less with the enteral administra- istered H2-receptor antagonists would have similar
tion than with i.v. administration. costs.
A formal economic analysis from an institutional The following is a template for evaluating the cost-
perspective was conducted as part of the guideline effectiveness of stress ulcer prophylaxis regimens on
development process. The general steps of acceptable the basis of institution-specific data.
cost-effectiveness analysis were followed: identification Step 1. Estimate the risk of clinically important bleed-
of clinical choices, determination of costs and benefits, ing events without prophylaxis (baseline rate stated as a
statement of time frame, determination of cost-effective- percentage).
ness ratio, and performance of a sensitivity analysis.274 Step 2. Estimate the reduction in clinically important
Readers are encouraged to make institution-specific deci- bleeding events associated with prophylaxis.
sions and to use the template to construct their own Step 3. Estimate the cost per patient of treating GI
bleeding at the institution.
institution-specific economic analysis.
Step 4. Estimate the cost of using sucralfate prophylaxis
The medications were assumed to have equal pro- in 100 patients.
phylactic efficacy on the basis of the most recent meta- Step 5. Estimate the cost of adverse events per 100
analysis.5 Although both the meta-analysis and the patients if all 100 are given prophylaxis.
recent large RCT met our criteria for a grade of A, the Step 6. Estimate the cost of treating bleeding per 100
meta-analysis had a level of evidence of I and the RCT
a level of evidence of I (see Table 1 for definitions).
Drug and administration costs9,64 and adverse event Table 9.
costs9,185,275,276 and frequency61,64,176,194-199,277,278 were ex- Cost Assumptions 9,64,185,275,276
trapolated from the literature (Table 9). Ranitidine was Cost Assumptions per
chosen as the representative H2-receptor antagonist; Episode or Course
i.v. administration was assumed. Because of the cost of Variable of Therapy ($)a
commercial sucralfate suspension, the cost analysis was Pneumonia 10,062
Bleeding 7,000
done under the assumption that nursing staff would Hepatitis 500
administer sucralfate tablets as a slurry through a naso- CNS toxicity 500
gastric tube. The medication cost figures were based on Cytopenia 500
a five-day course of prophylaxis. Costs and benefits Cardiovascular toxicity 500
Clogged tubes 160
were assumed to occur within the same period, making Antacids every two hours 78.00
discounting unnecessary. Any estimates not based on Histamine H2-receptor antagonist
the literature were mentioned. The cost estimates for three times dailyb 61.50
Sucralfate four times daily 33.80
many of the adverse effects were small (e.g., $500),
a
A course of therapy was assumed to be five days. Drug acquisition
because it was assumed that the effects would be readily costs are average wholesale costs.
reversible or found to have a different cause. The costs b
Ranitidine used as the example.
Figure 1. Decision tree for stress ulcer prophylaxis. H2-RAs = histamine H2-receptor antagonists, CV = cardiovascular.
No bleeding 94%
0 TOTAL
No (no prophylaxis)
n = 100 Bleeding 6% $7000 per episode $42,000 per 100
42,000 patients
patients treated [(risk of bleeding without prophylaxis) Step 9. Determine the cost-effectiveness ratio [(margin-
(estimated cost of treating GI bleed per patient)]. al cost of prophylaxis in 100 patients)/(risk of
Step 7. Estimate the savings with stress ulcer prophy- bleeding)(risk reduction with prophylaxis expressed as a
laxis per 100 patients treated {[(risk of bleeding without decimal)]. A negative value indicates a saving.
prophylaxis) (risk reduction with prophylaxis expressed Table 10 provides an example of how the cost-
as a decimal)] (cost of treating GI bleed per patient)}.
Step 8. Determine the marginal cost of prophylaxis per
effectiveness calculation for sucralfate was performed.
100 patients treated {[(cost of medication per 100 pa- The cost-effectiveness ratios for the various therapies
tients) + (cost of adverse drug events per 100 patients)] can be compared in order to choose the appropriate
(savings with stress ulcer prophylaxis in 100 patients)}. agent. As with any guidelines, patient-specific informa-
Note that positive numbers represent a cost liability, tion should always be used for determining prophylaxis
negative numbers a cost saving. for the individual patient. To perform a sensitivity
analysis for various rates of bleeding and reductions in data. The strength of the evidence for prophylaxis varies
bleeding associated with prophylaxis, simply vary the substantially from one risk factor to the next (Table 3).
values in step 9. Table 11 shows the results of the ICU patients with a history of recent GI pathology are
sensitivity analysis for sucralfate. deemed appropriate candidates for prophylaxis, but the
Figure 2 is an algorithm for stress ulcer prophylaxis recommendation must be graded D because patients
in adult patients that is based on the recommendations with recent histories of GI pathology were often excluded
contained in this document. The lack of pediatric RCTs from the published investigations, and thus it is un-
involving medications and the age stratification in known whether these patients are at increased risk for
those studies that did include pediatric patients pre- clinically important bleeding. The recommendation for
clude development of a separate pediatric algorithm. use of oral H2-receptor antagonists in patients with func-
Because the number (2) and type (coagulopathy, respi- tioning GI tracts (Figure 2) is based on presumptive
ratory failure, and Pediatric Risk of Mortality Score of efficacy, because no RCTs are available that have com-
10) of risk factors for clinically important bleeding pared i.v. with oral H2-receptor antagonists administra-
have been elucidated,66 the reader is referred to the tion for preventing clinically important bleeding.
discussion in the section titled Adverse effects of pro- Recommendation (adults): Given some of the unresolved
phylactic agents for help in deciding which medica- issues regarding stress ulcer prophylaxis, it is recommended
tion, if any, should be used for prophylaxis. that institution-specific guidelines be developed on the
The adult algorithm is intended to serve as a template basis of economic models such as the one included in this
that can be modified according to institution-specific document. For institutions willing to accept the assumption
that H2-receptor antagonists and sucralfate have equal
efficacy, the accompanying economic analysis found su-
cralfate to be the most cost-effective agent. Exceptions to
Table 10. the use of sucralfate include lack of oral or other gastric
Cost-effectiveness Calculations for Sucralfate access for administration and, potentially, documented
Variable Value failure of prophylaxis with sucralfate. If ranitidine is as-
Risk of bleeding without
sumed to be more effective than sucralfate without increas-
6.0 (0.139)61,64
prophylaxis (%) ing the risk of pneumonia, ranitidine (and presumably
Risk reduction with 50 (1090)64 other H2-receptor antagonists) would be the drug of choice
prophylaxis (%) and would also result in a cost saving, although less than
Cost of treating episode of GI 7000 that achieved with sucralfate.
bleeding per patient ($) Recommendation (pediatrics): There are also unresolved
Cost of sucralfate per 100 3380
patients ($)
issues for pediatric patients. Institution-specific guidelines
Cost of adverse drug events 2880
can be developed by using the economic models included in
per 100 patients treated ($) this document, but the percentages and costs need to be
Cost of treating GI bleeding in 42,000 tailored to this population on the basis of institution-
100 patients ($) specific data. There is supporting evidence in the literature
[(Six episodes of bleeding) that the risk of stress-induced bleeding is greater in pediatric
($7000 cost of treatment)] patients with respiratory failure, coagulopathy, a Pediatric
Savings with stress ulcer 21,000 Risk of Mortality Score of 10, and thermal injuries. Whether
prophylaxis in 100 patients ($) prophylaxis will reduce the risk of bleeding is yet to be
[(Three episodes of
bleeding) ($7000 cost of
proven in this population, so there is no clear agent of choice
treatment)] at this time.
Marginal cost of prophylaxis 14,740
in 100 patients ($) Conclusion
[($3380 + $2880) ($21,000)]
Cost-effectiveness ratio 4913 for each Although medications such as antacids have been
bleeding episode used for at least two decades to prevent stress-induced
averted bleeding, many issues remain unresolved. While pro-
Table 11.
Cost-effectiveness Ratio for Sucralfate Prophylaxis: Results of Sensitivity Analysisa
Yes No
Yes No
Yes No
58. Martin LF, Booth FVM, Karlstadt RG et al. Continuous intra- 80. Chan K, Mann KS. Failure of cimetidine prophylaxis in neu-
venous cimeditine decreases stress-related upper gastrointes- rosurgery. Aust N Z J Surg. 1989; 59:133-6.
tinal hemorrhage without promoting pneumonia. Crit Care 81. Reusser P, Gyr K, Scheidegger D et al. Prospective endoscopic
Med. 1993; 21:19-30. study of stress erosions and ulcers in critically ill neurosurgi-
59. Ben-Menachem T, Fogel R, Patel RV et al. Prophylaxis for cal patients: current incidence and effect of acid-reducing
stress-related gastric hemorrhage in the medical intensive prophylaxis. Crit Care Med. 1990; 18:270-4.
care unit: a randomized, controlled, single-blind study. Ann 82. Metz CA, Livingston DH, Smith JS et al. Impact of multiple
Intern Med. 1994; 121:568-75. risk factors and ranitidine prophylaxis on the development
60. Eddleston JM, Pearson RC, Holland J et al. Prospective endo- of stress-related upper gastrointestinal bleeding: a prospec-
scopic study of stress erosions and ulcers in critically ill adult tive, multicenter, double-blind, randomized trial. Crit Care
patients treated with either sucralfate or placebo. Crit Care Med. 1993; 21:1844-9.
Med. 1994; 22:1949-54. 83. Chan K, Lai ECS, Tuen H et al. Prospective double-blind
61. Cook DJ, Fuller HD, Guyatt GH et al. Risk factors for gas- placebo-controlled randomized trial on the use of ranitidine
trointestinal bleeding in critically ill patients. N Engl J Med. in preventing postoperative gastroduodenal complications
1994; 330:377-81. in high-risk neurosurgical patients. J Neurosurg. 1995;
62. Zandstra DF, Stoutenbeek CP. The virtual absence of stress- 82:413-7.
ulceration related bleeding in ICU patients receiving pro- 84. Burgess P, Larson GM, Davidson P et al. Effect of ranitidine
longed mechanical ventilation without any prophylaxis. In- on intragastric pH and stress-related upper gastrointestinal
tensive Care Med. 1994; 20:335-40. bleeding in patients with severe head injury. Dig Dis Sci.
63. Kuusela AL, Ruuska T, Karikoski R et al. A randomized, con- 1995; 40:645-50.
trolled study of prophylactic ranitidine in preventing stress- 85. ONeill JA, Bash MC, Pruitt BA et al. Studies related to the
induced gastric mucosal lesions in neonatal intensive care pathogenesis of Curlings ulcer. J Trauma. 1967; 7:275-87.
unit patients. Crit Care Med. 1997; 25:346-51. 86. Stone HH. Stress ulcers in patients with major burns. Am
64. Ben-Menachem T, McCarthy BD, Fogel R et al. Prophylaxis Surg. 1972; 38:107-10.
for stress-related gastrointestinal hemorrhage: a cost-effec- 87. Pruitt BA, Foley FD, Moncrief JA. Curlings ulcer: a clinical-
tiveness analysis. Crit Care Med. 1996; 24:338-45. pathology study of 323 cases. Ann Surg. 1970; 172:523-39.
65. Lopez-Herce J, Dorao P, Elola P et al., for the Gastrointestinal 88. Bruck HM, Pruitt BA. Curlings ulcer in children: a 12-year
Hemorrhage Study Group. Frequency and prophylaxis of review of 63 cases. J Trauma. 1972; 12:490-6.
upper gastrointestinal hemorrhage in critically ill children: a 89. Czaja AJ, McAlhany JC, Pruitt BA. Acute gastroduodenal
prospective study comparing the efficacy of almagate, raniti- disease after thermal injury: an endoscopic evaluation of
dine, and sucralfate. Crit Care Med. 1992; 20:1082-9. incidence and natural history. N Engl J Med. 1974; 291:925-9.
66. Chaibou M, Tucci M, Dugas MA et al. Clinically significant 90. McAlhany JC, Czaja AJ, Pruitt BA. Antacid control of compli-
upper gastrointestinal bleeding acquired in a pediatric inten- cations from acute gastroduodenal disease after burns. J
sive care unit. Pediatrics. 1998; 102:933-8. Abstract. Trauma. 1976; 16:645-9.
67. Wiedeck H, Stanescu A, Reinelt H et al. Evaluation of stress 91. Solem LD, Strate RG, Fischer RP. Antacid therapy and nutri-
ulcer prophylaxis in high risk intensive care patients. Inten- tional supplementation in the prevention of Curlings ulcer.
sive Care Med. 1997; 23(suppl 1):S139. Surg Gynecol Obstet. 1979; 148:367-70.
68. Lacroix J, Infante-Rivarde C, Jenicek M et al. Prophylaxis of 92. Jones RH, Rudge CJ, Bewick M et al. Cimetidine: prophylaxis
upper gastrointestinal bleeding in intensive care units: a against upper gastrointestinal haemorrhage after renal trans-
meta-analysis. Crit Care Med. 1989; 17:862-9. plantation. Br Med J. 1978; 1 (Feb 18):398-400.
69. Tryba M. Prophylaxis of stress ulcer bleeding: a meta-analy- 93. Schiessel R, Starlinger M, Wolf A et al. Failure of cimetidine
sis. J Clin Gastroenterol. 1991; 13(suppl 2):S44-55. to prevent gastroduodenal ulceration and bleeding after re-
70. Cook DJ, Witt LG, Cook RJ et al. Stress ulcer prophylaxis in nal transplantation. Surgery. 1981; 90:456-8.
the critically ill: a meta-analysis. Am J Med. 1991; 91:519-27. 94. Walter S, Andersen JT, Christensen U et al. Effect of cimeti-
71. Tryba M. Sucralfate versus antacids or H2-antagonists for dine on upper gastrointestinal bleeding after renal transplan-
stress ulcer prophylaxis: a meta-analysis on efficacy and tation: a prospective study. Br Med J. 1984; 289:1175-6.
pneumonia rate. Crit Care Med. 1991; 19:942-9. 95. MacDougall BRD, Bailey RJ, Williams R. H2-receptor antago-
72. Layon AJ, Florete O, Day AL et al. The effect of duodenojeju- nists and antacids in the prevention of acute gastrointestinal
nal alimentation on gastric pH and hormones in intensive hemorrhage in fulminant hepatic failure: two controlled
care unit patients. Chest. 1991; 99:695-702. trials. Lancet. 1977; 1:617-9.
73. Choctaw WT, Fujita C, Zawacki BE. Prevention of upper 96. Nagasue N, Yukaya H, Ogawa Y et al. Prophylaxis of upper
gastrointestinal bleeding in burn patients. Arch Surg. 1980; gastrointestinal bleeding with cimetidine in patients under-
115:1073-6. going partial hepatectomy. Ann Chir Gynaecol. 1984; 73:6-10.
74. Pingleton SK, Hadzima SK. Enteral alimentation and gas- 97. Klompmaker IJ, Slooff MJH, de Bruijn KM et al. Prophylaxis
trointestinal bleeding in mechanically ventilated patients. with ranitidine against peptic ulcer disease after liver trans-
Crit Care Med. 1983; 11:13-6. plantation. Transpl Int. 1988; 1:209-12.
75. Gurman G, Samri M, Sarov B et al. The rate of gastrointesti- 98. Fusamoto H, Hagiwara H, Meren H et al. A clinical study of
nal bleeding in a general ICU population: a retrospective acute gastrointestinal hemorrhage associated with various
study. Intensive Care Med. 1990; 16:44-9. shock states. Am J Gastroenterol. 1991; 86:429-33.
76. Cochard JF, Leger A, Pinaquy C et al. Gastrointestinal bleed- 99. Rosen HR, Vlahakes GJ, Rattner DW. Fulminant peptic ulcer
ing in trauma ICU patients: incidence and risk factors. Inten- disease in cardiac surgical patients: pathogenesis, preven-
sive Care Med. 1997; 23(suppl 1):S140. Abstract. tion, and management. Crit Care Med. 1992; 20:354-9.
77. Kamada T, Fusamoto H, Kawano S et al. Gastrointestinal 100. Apte NM, Karnad PR, Medhekar TP et al. Gastric colonization
bleeding following head injury: a clinical study of 433 cases. and pneumonia in intubated critically ill patients receiving
J Trauma. 1977; 17:44-7. stress ulcer prophylaxis: a randomized, controlled trial. Crit
78. Halloran LG, Zlass AM, Gayle WE et al. Prevention of acute Care Med. 1992; 20:590-3.
gastrointestinal complications after severe head injury: a 101. Simons RK, Hoyt DB, Winchell RJ et al. A risk analysis of
controlled trial of cimetidine prophylaxis. Am J Surg. 1980; stress ulceration after trauma. J Trauma Inj Infect Crit Care.
139:44-8. 1995; 39:289-94.
79. Mouawad E, Deloof T, Genette F et al. Open trial of cimeti- 102. Faigel DO, Metz DC. Prevalence, etiology, and prognostic
dine in the prevention of upper gastro-intestinal haemor- significance of upper gastrointestinal hemorrhage in diabetic
rhage in patients with severe intracranial injury. Acta Neuro- ketoacidosis. Dig Dis Sci. 1996; 41:1-8.
chir. 1983; 67:239-44. 103. Schepp W. Stress ulcer prophylaxis: still a valid option in the
1990s? Digestion. 1993; 54:189-99. 126. Eddleston JM, Vohra A, Scott P et al. A comparison of the
104. Rixen D, Livingston DH, Loder P et al. Ranitidine improves frequency of stress ulceration and secondary pneumonia in
lymphocyte function after severe head injury: results of a sucralfate- or ranitidine-treated intensive care unit patients.
randomized, double-blind study. Crit Care Med. 1996; Crit Care Med. 1991; 19:1491-6.
24:1787-92. 127. Sirvent JM, Verdaguer R, Ferrer MJ et al. Nosocomial pneu-
105. Friedland JS, Porter JC, Daryanani S et al. Plasma proinflam- monia and stress ulcer prophylaxis in intubated patients:
matory cytokine concentrations, Acute Physiology and comparative study of antacids and ranitidine versus sucral-
Chronic Health Evaluation (APACHE) III scores and survival fate. Intensive Care Med. 1992; 18:S186. Abstract.
in patients in an intensive care unit. Crit Care Med. 1996; 128. Ryan P, Dawson J, Teres D et al. Nosocomial pneumonia
24:1775-81. during stress ulcer prophylaxis with cimetidine and sucral-
106. Walt RP. Misoprostol for the treatment of peptic ulcer and fate. Arch Surg. 1993; 128:1353-7.
antiinflammatory-drug-induced gastroduodenal ulceration. 129. Grau JM, Casademont J, Fernandez-Sola J et al. Prophylaxis
N Engl J Med. 1992; 327:1575-80. of gastrointestinal tract bleeding in patients admitted to a
107. Guth PH. Mucosal coating agents and other nonantisecreto- general hospital ward. Scand J Gastroenterol. 1993; 28:244-8.
ry agents: are they cytoprotective? Dig Dis Sci. 1987; 32:647- 130. Levy MM, Carson R, Miyasaki A et al. Randomized trial: con-
54. tinuous infusion H2 blockers, bolus H2 blockers or Carafate bid
108. Walt RP, Langman MJS. Antacids and ulcer healing: a review for the prophylaxis of stress ulcers in the ICU. Crit Care Med.
of the evidence. Drugs. 1991; 42:205-12. 1993; 21(suppl):S181.
109. McCarthy DM. Sucralfate. N Engl J Med. 1991; 325:1017-25. 131. Fabian TC, Boucher BA, Croce MA et al. Pneumonia and
110. Tagamet product information. Philadelphia, PA. SmithKline stress ulceration in severely injured patients. Arch Surg. 1993;
Beecham; 1998. 128:185-92.
111. Rollins CJ. Adult enteral nutrition. In: Young LY, Koda- 132. Pickworth KK, Falcone RE, Hoogeboom JE et al. Occurrence
Kimble MA. Applied therapeutics. Vancouver, WA: Applied of nosocomial pneumonia in mechanically ventilated trau-
Therapeutics; 1995:1-28. ma patients: a comparison of sucralfate and ranitidine. Crit
112. Sharma VK, Heinzelmann EJ, Steinberg EN et al. Nonencap- Care Med. 1993; 21:1856-62.
sulated, intact omeprazole granules effectively suppress in- 133. Prodhom G, Leuenberger P, Koerfer J et al. Nosocomial
tragastric acidity when administered via a gastrostomy. Am J pneumonia in mechanically ventilated patients receiving
Gastroenterol. 1997; 92:848-51. antacid, ranitidine, or sucralfate as prophylaxis for stress
113. Balaban DH, Duckworth CW, Peura DA. Nasogastric omepra- ulcer: a randomized controlled trial. Ann Intern Med. 1994;
zole: effects on gastric pH in critically ill patients. Crit Care 120:653-62.
Med. 1997; 92:79-83. 134. Maier RV, Mitchell D, Gentilello L. Optimal therapy for
114. Phillips JO, Metzler MH, Palmieri TL et al. A prospective stress gastritis. Ann Surg. 1994; 220:353-63.
study of simplified omeprazole suspension for the prophy- 135. Cioffi WG, McMannus AT, Rue LW et al. Comparison of acid
laxis of stress-related mucosal damage. Crit Care Med. 1996; neutralizing and non-acid neutralizing stress ulcer prophy-
24:1793-800. laxis in thermally injured patients. J Trauma. 1994; 36:541-7.
115. Lasky MR, Metzler MH, Phillips JO. A prospective study of 136. Thomason M, Payseur ES, Hakanewerth AM et al. Nosocomi-
omeprazole suspension to prevent clinically significant gas- al pneumonia in ventilated patients during stress ulcer pro-
trointestinal bleeding from stress ulcers in mechanically ven- phylaxis with sucralfate, antacid and ranitidine. J Trauma Inj
tilated trauma patients. J Trauma Inj Infect Crit Care. 1998; Infect Crit Care. 1996; 4:503-8.
44:527-33. 137. Borrero E, Margolis IB, Bank S. Antacid versus sucralfate in
116. Prevacid product information. North Chicago, IL. Abbott; preventing acute gastrointestinal bleeding. Am J Surg. 1984;
1998. 148:809-12.
117. Tryba M, Zevounou F, Torok M et al. Prevention of acute 138. Borrero E, Bank S, Margolis I et al. Comparison of antacid
stress bleeding with sucralfate, antacid or cimetidine. Am J and sucralfate in the prevention of gastrointestinal bleeding
Med. 1985; 79(suppl 2C):55-61. in patients who are critically ill. Am J Med. 1985; 79(suppl
118. Cannon LA, Heiselman D, Gardner W et al. Prophylaxis of 2C):62-4.
upper gastrointestinal tract bleeding in mechanically venti- 139. Borerro E, Ciervo J, Chang JB. Antacid versus sucralfate in
lated patients. Arch Intern Med. 1987; 147:2101-6. preventing acute gastrointestinal tract bleeding in abdomi-
119. Driks MR, Craven DE, Bartolome RC et al. Nosocomial pneu- nal-aortic surgery. Arch Surg. 1986; 121:810-2.
monia in intubated patients given sucralfate as compared 140. Bresalier RS, Grendell JH, Cello JP et al. Sucralfate suspension
with antacids or histamine type 2 blockers. N Engl J Med. versus titrated antacid for the prevention of acute stress-
1987; 317:1376-82. related gastrointestinal hemorrhage in critically ill patients.
120. Rath T, Walzer LR, Meissl G. Preventative measures for stress Am J Med. 1987; 83(suppl 3B):110-6.
ulcers in burn patients. Burns. 1988; 14:504-7. 141. Tryba M. Risk of acute stress bleeding and nosocomial pneu-
121. Laggner AN, Lenz K, Graninger W et al. Stress-blutungspro- monia in ventilated intensive care unit patients: sucralfate
phylaxe auf einer internen intensivstation: sucralfat versus versus antacids. Am J Med. 1987; 83(suppl 3B):117-24.
ranitidin. Anaesthesist. 1988; 37:704-10. 142. Israsena S, Kladcharoen N, Limthongkul S et al. Sucralfate
122. Laggner AN, Lenz K, Base W et al. Prevention of upper versus antacid in the prevention of stress ulcer bleeding in
gastrointestinal bleeding in long-term ventilated patients. patients on mechanical ventilation. J Med Assoc Thai. 1987;
Am J Med. 1989; 86(suppl 6A):81-4. 70:678-82.
123. Colardyn F, Vogelaers D, Verschraegen G et al. Cimetidine 143. Mundinger GH, Hays A, Allo M et al. Effect of meciadanol
versus sucralfate for stress ulcer prophylaxis in medical in- and sucralfate as compared with antacid titration regimen
tensive care unit patients. Intensive Care Med. 1990; 16:S18. on prevention of acute gastrointestinal hemorrhage in post-
Abstract. operative intensive care patients. Surg Forum. 1985; 36:121-3.
124. Kappstein I, Schulgen G, Friedrich T et al. Incidence of pneu- 144. Kitler ME, Hay A, Enterline JP et al. Preventing postoperative
monia in mechanically ventilated patients treated with su- acute bleeding of the upper part of the gastrointestinal tract.
cralfate or cimetidine as prophylaxis for stress bleeding: bac- Surg Gynecol Obstet. 1990; 171:366-72.
terial colonization of the stomach. Am J Med. 1991; 91(suppl 145. Bonten MM, Gaillard CA, van der Geest S et al. The role of
2A):125S-31S. intragastric acidity and stress ulcer prophylaxis on coloniza-
125. Simms HH, DeMaria E, McDonald L et al. Role of gastric tion and infection in mechanically ventilated ICU patients.
colonization in the development of pneumonia in critically Am J Respir Crit Care Med. 1995; 152:1825-34.
ill trauma patients: results of a prospective randomized trial. 146. McElwee HP, Sirinek KR, Levine BA. Cimetidine affords pro-
J Trauma. 1991; 31:531-7. tection equal to antacids in prevention of stress ulceration
following thermal injury. Surgery. 1979; 86:620-6. lesions of the upper gastrointestinal tract in pediatric pa-
147. Martin LF, Max MH, Polk HC. Failure of gastric pH control by tients after cardiac surgery: effects of prophylaxis. Br Heart J.
antacids or cimetidine in the critically ill: a valid sign of 1994; 72:186-9.
sepsis. Surgery. 1980; 88:59-68. 169. Mustafa NA, Akturk G, Ozen I et al. Acute stress bleeding
148. Priebe HJ, Skillman JJ, Bushnell LS et al. Antacid versus prophylaxis with sucralfate versus ranitidine and incidence
cimetidine in preventing acute gastrointestinal bleeding. N of secondary pneumonia in intensive care unit patients.
Engl J Med. 1980; 302:426-30. Intensive Care Med. 1995; 21:287. Letter.
149. Stothert JC, Siminowitz DA, Dellinger EP et al. Randomized 170. Reusser P, Zimmerli W, Scheidegger D et al. Role of gastric
prospective evaluation of cimetidine and antacid control of colonization in nosocomial infections and endotoxemia: a
gastric pH in the critically ill. Ann Surg. 1981; 192:169-74. prospective study in neurosurgical patients on mechanical
150. Kahn F, Parekh A, Patel S et al. Results of gastric neutraliza- ventilation. J Infect Dis. 1989; 160:414-21.
tion with hourly antacids and cimetidine in 320 intubated 171. Mahul P, Auboyer C, Jospe R et al. Prevention of nosocomial
patients with respiratory failure. Chest. 1981; 79:409-12. pneumonia in intubated patients: respective role of mechan-
151. Weigelt LA, Aurbakken CM, Geweritz BL et al. Cimetidine ical subglottic secretions drainage and stress ulcer prophy-
versus antacid in prophylaxis for stress ulceration. Arch Surg. laxis. Intensive Care Med. 1992; 18:20-5.
1981; 115:597-601. 172. Thomasen M, Reeves R, Payseur E et al. Nosocomial pneumo-
152. Luk GD, Summer WR, Messersmith JF et al. Cimetidine and nia in ventilated trauma patients during stress ulcer prophy-
antacid in prophylaxis of acute gastrointestinal bleeding: a laxis with sucralfate, ranitidine and antacid. Crit Care Med.
randomized, double-blind, controlled study. Gastroenterolo- 1995; 23:A93. Abstract.
gy. 1982; 82:1121. Abstract. 173. Fiddian-Green RG, McGough E, Pittenger G et al. Predictive
153. Zach GA, Gyr KE, von Alvensleben E et al. A double-blind value of intramural pH and other risk factors for massive
randomized, controlled study to investigate the efficacy of bleeding from stress ulceration. Gastroenterology. 1983;
cimetidine given in addition to conventional therapy in the 85:613-20.
prevention of stress ulceration and haemorrhage in patients 174. Levy MJ, Seelig CB, Robinson NJ et al. Comparison of ome-
with acute spinal injury. Digestion. 1984; 29:214-22. prazole and ranitidine for stress ulcer prophylaxis. Dig Dis
154. Kingsley AN. Prophylaxis for acute stress ulcers: antacids or Sci. 1997; 42:1255-9.
cimetidine. Am Surg. 1985; 51:545-7. 175. Phillips JO, Metzler MH, Huckfeldt RE et al. A multicenter,
155. Moscona R, Kaufman T, Jacobs R et al. Prevention of gas- prospective, randomized clinical trial of continuous infusion
trointestinal bleeding in burns: the effects of cimetidine or IV ranitidine versus omeprazole suspension in the prophy-
antacids combined with early enteral feeding. Burns. 1985; laxis of stress ulcers. Crit Care Med. 1998; 26(suppl):A101.
12:65-7. Abstract.
156. Polesky MH, Spanier AH. Cimetidine versus antacids in the 176. Cook D, Guyatt G, Marshall J et al. A comparison of sucral-
prevention of stress erosions in critically ill patients. Am J fate and ranitidine for the prevention of upper gastrointesti-
Gastroenterol. 1986; 81:107-11. nal bleeding in patients requiring mechanical ventilation. N
157. Noseworthy TW, Shustack A, Johnston RG et al. A random- Engl J Med. 1998; 338:791-7.
ized clinical trial comparing ranitidine and antacids in criti- 177. Cook DJ, Reeve BK, Scholes LC. Histamine-2-receptor antag-
cally ill patients. Crit Care Med. 1987; 15:817-9. onists and antacids in the critically ill population: stress
158. LaMothe PH, Rao E, Serra AJ et al. Comparative efficacy of ulceration versus nosocomial pneumonia. Infect Control Hosp
cimetidine, famotidine, ranitidine, and Mylanta in postoper- Epidemiol. 1994; 15:437-42.
ative stress ulcers: gastric pH control and ulcer prevention in 178. McEvoy GK, ed. AHFS drug information 98. Bethesda, MD:
patients undergoing coronary artery bypass graft surgery. American Society of Health-System Pharmacists; 1998.
Gastroenterology. 1991; 100:1515-20. 179. Moore JG, Coburn JW, Sanders MC et al. Effects of sucralfate
159. Martin LF. Stress ulcers are common after aortic surgery: and ranitidine on aluminum concentrations in elderly vol-
endoscopic evaluation of prophylactic therapy. Am Surg. unteers. Pharmacotherapy. 1995; 15:742-6.
1994; 60:169-74. 180. Burgess E, Muruve D, Audette R. Aluminum absorption and
160. Cloud ML, Offen W, and the Nizatidine Intensive Care Unit excretion following sucralfate therapy in chronic renal insuf-
Study Group. Continuous infusions of nizatidine are safe ficiency. Am J Med. 1992; 92:471-5.
and effective in the treatment of intensive care unit patients 181. Tryba M, Kurz-Muller K, Donner B. Plasma aluminum con-
at risk for stress gastritis. Scand J Gastroenterol. 1994; 29(suppl centrations in long-term mechanically ventilated patients
206):29-34. receiving stress ulcer prophylaxis with sucralfate. Crit Care
161. Skillman JJ, Lisbon A, Long PC et al. 15R-15-methyl prosta- Med. 1994; 22:1769-73.
glandin E2 does not prevent gastrointestinal bleeding in 182. Baines PB, Ratcliffe J, Petros A. Aluminum concentrations in
seriously ill patients. Am J Surg. 1984; 147:451-5. critically ill children with renal impairment. Crit Care Med.
162. More DG, Raper RF, Munro IA et al. Randomized, prospective 1995; 23:1931.
trial of cimetidine and ranitidine for control of intragastric 183. Leung ACT, Henderson IS, Halls DJ et al. Aluminum hydrox-
pH in the critically ill. Surgery. 1985; 87:215-24. ide versus sucralfate as a phosphate binder in uraemia. Br
163. Jmelnitzky AC, Moday MC, Gallardo EA et al. Prophylaxis of Med J. 1983; 286:1379-81.
acute gastric hemorrhage from stress-induced lesions: com- 184. Brogden RN, Heel RC, Speight TM et al. Sucralfate: a review
parison of pirenzepine and cimetidine in intensive care of its pharmacodynamic properties and therapeutic use in
units. Dig Dis Sci. 1986; 31:206. Abstract. peptic ulcer disease. Drugs. 1984; 27:194-209.
164. Tryba M. Prevention of stress bleeding with ranitidine or 185. Marcuarad SP, Perkins AM. Clogging of feeding tubes. J
pirenzepine and the risk of pneumonia. J Clin Anesth. 1988; Parenter Enteral Nutr. 1988; 12:403-5.
1:12-20. 186. Beckwith MC, Barton RG, Graves C. A guide to drug therapy
165. Zinner MJ, Rypins EB, Martin LK et al. Misoprostol versus in patients with enteral feeding tubes: dosage form selection
antacid titration for preventing stress ulcers in postoperative and administration methods. Hosp Pharm. 1997; 32:57-64.
surgical ICU patients. Ann Surg. 1989; 210:590-5. 187. Algozzine GJ, Hill G, Scoggins WG et al. Sucralfate bezoar. N
166. Salim AS. Protection against stress-induced acute gastric mu- Engl J Med. 1983; 309:1387. Letter.
cosal injury by free radical scavengers. Intensive Care Med. 188. Anderson W, Weatherstone G. Esophageal medication be-
1991; 17:455-60. zoar in a patient receiving enteral feedings and sucralfate.
167. Martin LF, Booth FVM, Reines HD et al. Stress ulcers and Am J Gastroenterol. 1989; 84:205-6.
organ failure in intubated patients in surgical intensive care 189. Quigley MA, Flicker M, Caldwell EG. Sucralfate bezoar
units. Ann Surg. 1992; 215:332-7. theory or fact? Am J Gastroenterol. 1986; 81:724-5. Letter.
168. Behrens R, Hofbeck M, Singer H et al. Frequency of stress 190. Strozik KS, Walele AH, Hoffman H. Bezoar in a preterm baby
associated with sucralfate. Clin Pediatr. 1996; 35:423-4. 220. Tryba M. Side effects of stress bleeding prophylaxis. Am J
191. Tang TS, Yeung CY. Intestinal obstruction and perforation Med. 1989; 86(suppl 6A):85-93.
following sucralfate administration in a very low birth 221. Daschner F, Kappstein I, Engels I et al. Stress ulcer prophylax-
weight infant. J Perinat Med. 1992; 20:317-21. is and ventilation pneumonia: prevention by antibacterial
192. Portuzguez-Malavasi A, Aranda JV. Antacid bezoar in a new- cytoprotective agents? Infect Control Hosp Epidemiol. 1988;
born. Pediatrics. 1979; 63:679-80. 9:59-65.
193. Brand JM, Greer FR. Hypermagnesemia and intestinal perfo- 222. Bonten MJ, Bergmans DJ, Ambergen AW et al. Risk factors
ration following antacid administration in a premature in- for pneumonia, and colonization of respiratory tract and
fant. Pediatrics. 1990; 85:121-4. stomach in mechanically ventilated ICU patients. Am J Respir
194. Dobbs JH, Muir JG, Smith RN. H2-antagonists and hepatitis. Crit Care Med. 1996; 154:1339-46.
Ann Intern Med. 1986; 105:803. Letter. 223. Ephgrave KS, Kleiman-Wexler R, Pfaller M et al. Effects of
195. Lewis JH. Hepatic effects of drugs used in the treatment of sucralfate vs antacids on gastric pathogens: results of a dou-
peptic ulcer disease. Am J Gastroenterol. 1987; 82:987-1003. ble-blind clinical trial. Arch Surg. 1998; 133:251-7.
196. Davis TG, Pickett DL, Schlosser JH. Evaluation of a world- 224. Craven DE, Steger KA. Epidemiology of nosocomial pneumo-
wide spontaneous reporting system with cimetidine. JAMA. nia: new perspectives on an old disease. Chest. 1995;
1980; 243:1912-4. 108(suppl):1S-16S.
197. Aymard J-P, Aymard B, Netter P et al. Haematological adverse 225. The Hospital Infection Control Practices Advisory Commit-
effects of histamine H2-receptor antagonists. Med Toxicol. tee. Part II: recommendations for prevention of nosocomial
1988; 3:430-48. pneumonia. MMWR. 1997; 46:45-79.
198. Hinrichsen H, Halabi A, Kirch W. Clinical aspects of cardio- 226. Toung TK, Rosenfeld BA, Yoshiki A et al. Sucralfate does not
vascular effects of H2-receptor antagonists. Eur J Clin Invest. reduce the risk of acid aspiration pneumonitis. Crit Care Med.
1995; 25(suppl 1):47-56. 1993; 21:1359-64.
199. Cantu TG, Korek JS. Central nervous system reactions to 227. Cook DJ, Laine LA, Guyatt GH et al. Nosocomial pneumonia
histamine-2 receptor blockers. Ann Intern Med. 1991; 114: and the role of gastric pH. Chest. 1991; 100:7-13.
1027-34. 228. Bonten MJM, Guillard CA, de Leeuw PW et al. Role of coloni-
200. Smallwood RA, Berlin RG, Castagnoli N et al. Safety of acid- zation of the upper intestinal tract in the pathogenesis of
suppressing drugs. Dig Dis Sci. 1995; 40(suppl):63S-80S. ventilator-associated pneumonia. Clin Infect Dis. 1997;
201. Puff MR, Carey WD. The effect of cimetidine on cyclosporine 24:309-19.
A levels in liver transplant recipients: a preliminary report. 229. Hardman JG, Limbird LE, Molinoff PB et al., eds. Goodman
Am J Gastroenterol. 1992; 87:287-91. and Gilmans the pharmacological basis of therapeutics. 9th
202. Berardi RR, Kirking DM, Townsend KA et al. Identifying ed. New York: McGraw-Hill Health Professions Division;
potential interactions of sucralfate with other drugs in hospi- 1996.
talized patients. Am J Hosp Pharm. 1992; 49:1488-90. 230. Watson LC, Abston S. Prevention of upper gastrointestinal
203. Guay DRP. The role of fluoroquinolones. Pharmacotherapy. hemorrhage in 582 burned children. Am J Surg. 1976;
1992; 12:71S-85S. 132:790-3.
204. Just PM. Overview of fluoroquinolone antibiotics. Pharmaco- 231. Vandenplas Y, Sacre L. The use of cimetidine in newborns.
therapy. 1993; 13:4S-17S. Am J Perinatol. 1987; 4:131-3.
205. Holman JS, Shwed JA. Influence of sucralfate on the detec- 232. Lopez-Herce J, Velasco LA, Codoceo R et al. Ranitidine pro-
tion of occult blood in simulated gastric fluid by two screen- phylaxis in acute gastric mucosal damage in critically ill
ing tests. Clin Pharm. 1992; 11:625-7. patients. Crit Care Med. 1988; 16:591-3.
206. Dal Negro R. Pharmacokinetic drug interactions with anti- 233. Harrison AM, Lugo RA, Vernon DD. Gastric pH control in
ulcer drugs. Clin Pharmacokinet. 1998; 35:135-50. critically ill children receiving intravenous ranitidine. Crit
207. Blum RA. Lansoprazole and omeprazole in the treatment of Care Med. 1998; 26:1433-6.
acid peptic disorders. Am J Health-Syst Pharm. 1996; 53:1401- 234. Taketomo CK, Hodding JH, Kraus DM. Pediatric dosage
15. handbook. 4th ed. Hudson, OH: Lexi-Comp; 199798.
208. Hausken T, Stene-Larsen G, Aronsen O et al. Misoprostol 235. Osteyee JL, Banner W. Effects of two dosing regimens of
treatment exacerbates abdominal discomfort in patients intravenous ranitidine on gastric pH in critically ill children.
with non-ulcer dyspepsia and erosive prepyloric changes. Am J Crit Care. 1994; 3:267-72.
Scand J Gastroenterol. 1990; 25:1028-33. 236. Martyn JAJ, Greenblatt DJ, Hagen J et al. Alteration by burn
209. Prilosec product information. Wayne, PA: Astra Merck; 1997. injury of the pharmacokinetics and pharmacodynamics of
210. Pepcid product information. West Point, PA: Merck; 1996. cimetidine in children. Eur J Clin Pharmacol. 1989; 36:361-7.
211. Zantac product information. Research Triangle Park, NC: 237. Lloyd CW, Martin WJ, Taylor BD et al. Pharmacokinetics and
Glaxo Wellcome; 1996. pharmacodynamics of cimetidine and metabolites in criti-
212. Axid product information. Indianapolis, IN: Lilly; 1997. cally ill children. J Pediatr. 1985; 107:295-300.
213. Carafate product information. Kansas City, MO: Hoechst 238. Blumer JL, Rothstein FC, Kaplan BS et al. Pharmacokinetic
Marion Roussel; 1996. determination of ranitidine pharmacodynamics in pediatric
214. Garris RE, Kirkwood CF. Misoprostol: a prostaglandin E1 ulcer disease. J Pediatr. 1985; 107:301-6.
analogue. Clin Pharm. 1989; 8:627-44. 239. Sarna MJ, Saili A, Dutta AK et al. Stress associated gastric
215. Pittet D. Predicting nosocomial pneumonia in the ICU . . . an bleeding in newborns: role of ranitidine. Indian Pediatr. 1991;
ongoing challenge. Intensive Care Med. 1996; 22:1139-40. 28:1305-8.
216. Official statement of the American Thoracic Society. Hospi- 240. Treem WR, Davis PM, Hyams JS. Suppression of gastric acid
tal-acquired pneumonia in adults: diagnosis, assessment of secretion by intravenous administration of famotidine in
severity, initial antimicrobial therapy, and preventative children. J Pediatr. 1991; 118:812-6.
strategies. Am J Respir Crit Care Med. 1995; 153:1711-25. 241. Nagita A, Manago M, Aoki S et al. Pharmacokinetics and
217. Du Moulin GC, Paterson DG, Hedley-Whyte J et al. Aspira- pharmacodynamics of famotidine in children with gas-
tion of gastric bacteria in antacid-treated patients: a frequent troduodenal ulcers. Ther Drug Monit. 1994; 16:444-9.
cause of postoperative colonization of the airway. Lancet. 242. Bush A, Busst CM, Knight WB et al. Cardiovascular effects of
1982; 1:242-5. tolazoline and ranitidine. Arch Dis Child. 1987; 62:241-6.
218. Garvey BM, McCambley JA, Tuxen DV. Effects of gastric 243. Feldman M, Burton ME. Histamine2-receptor antagonists. N
alkalinization on bacterial colonization in critically ill pa- Engl J Med. 1990; 323:1672-80.
tients. Crit Care Med. 1989; 17:211-6. 244. Kuusela AL. Long term gastric pH monitoring for determin-
219. Garrard CS, ACourt CD. The diagnosis of pneumonia in the ing optimal dose of ranitidine for critically ill preterm and
critically ill. Chest. 1995; 108(suppl):17S-25S. term neonates. Arch Dis Child Fetal Neonatal Ed. 1998;
78:F151-3. 260. Vincent JL. Prevention and therapy of multiple organ failure.
245. Kelly EJ, Chatfield SL, Brownlee KG et al. The effect of intra- World J Surg. 1996; 20:465-70.
venous ranitidine on the intragastric pH of preterm infants 261. Chernov MS, Hale HW, Wood M. Prevention of stress ulcers.
receiving dexamethasone. Arch Dis Child. 1993; 69:37-9. Am J Surg. 1971; 122:674-7.
246. Smythe MA, Zarowitz BJ. Changing perspectives of stress 262. Spilker CA, Hinthorn DR, Pingleton SK. Intermittent enteral
gastritis prophylaxis. Ann Pharmacother. 1994; 28:1073-85. feeding in mechanically ventilated patients: the effect on
247. Dobkin ED, Valcour A, McCloskey CR et al. Does pH paper gastric pH and gastric cultures. Chest. 1996; 110:243-8.
accurately reflect gastric pH? Crit Care Med. 1990; 18:985-8. 263. Bengmark S, Gianotti L. Nutritional support to prevent and
248. Ostro MJ, Russell JA, Soldin SJ et al. Control of gastric pH treat multiple organ failure. World J Surg. 1996; 20:474-81.
with cimetidine: boluses versus primed infusions. Gastroen- 264. Baue AE, Durham RM, Mazuski JE. Clinical trials of new and
terology. 1985; 89:532-7. novel therapeutic agents. World J Surg. 1996; 20:493-8.
249. Heiselman DE, Hulisz DT, Fricker R et al. Randomized com- 265. Matsumoto T, Kaibara N, Sugimachi K et al. Pathophysiology
parison of gastric pH control with intermittent and continu- and management of acute gastric mucosal hemorrhage. Jpn J
ous intravenous infusion of famotidine in ICU patients. Am J Surg. 1978; 8:261-9.
Gastroenterol. 1995; 90:277-9. 266. Salim AS. Scavenging free radicals to prevent stress-induced
250. Jones LA, Gonzalez ER, Reines HD et al. Assessment of 24- gastric mucosal injury. Lancet. 1989; 2:1390.
hour gastric pH measurements in trauma patients receiving 267. Curtis LE, Simonian S, Buerk CA et al. Evaluation of the
intravenous famotidine by intermittent bolus versus contin- effectiveness of controlled pH in management of massive
uous infusion administration. Ann Pharmacother. 1994; upper gastrointestinal bleeding. Am J Surg. 1973; 125:474-6.
28:841-4. 268. Taylor PC, Loop FD, Hermann RE. Management of acute
251. Baghaie AA, Mojtahedzadeh M, Levine RL et al. Comparison stress ulcer after cardiac surgery. Ann Surg. 1973; 178:1-5.
of the effect of intermittent administration and continuous 269. MacDonald AS, Pyne DA, Freeman ANG et al. Upper gas-
infusion of famotidine on gastric pH in critically ill patients: trointestinal bleeding in the intensive care unit. Can J Surg.
results of a prospective, randomized, crossover study. Crit 1978; 21:81-4.
Care Med. 1995; 23:687-91. 270. Wadibia EC, Lucas BD, Hilleman DE et al. Economic impact
252. Ballesteros MA, Hogan DL, Koss MA et al. Bolus or intrave- of inappropriate histamine2-receptor antagonist use. Clin
nous infusion of ranitidine: effects on gastric pH and acid Ther. 1997; 19:1085-91.
secretion. Ann Intern Med. 1990; 112:334-9. 271. Erstad BL, Camamo JM, Miller MJ et al. Impacting cost and
253. Gedeit RG, Weigle CGM, Havens PL et al. Control and vari- appropriateness of stress ulcer prophylaxis at a university
ability of gastric pH in critically ill children. Crit Care Med. medical center. Crit Care Med. 1997; 25:1678-84.
1993; 21:1850-5. 272. Schumock GT, Lam NP, Winkler SR et al. Pharmacoeconom-
254. Tryba M. Stress ulcer prophylaxis and gastric alkalinization ic analysis of stress ulcer prophylaxis for critically ill patients.
death of a myth? Intensive Care Med. 1992; 18:1-3. PharmacoEconomics. 1996; 5:455-65.
255. Moore JG, Clemmer TP, Taylor S et al. Twenty-four-hour 273. Pemberton LB, Schaefer N, Goehring L et al. Oral ranitidine
intragastric pH patterns in ICU patients on ranitidine. Dig as prophylaxis for gastric stress ulcers in intensive care unit
Dis Sci. 1992; 37:1802-9. patients: serum concentrations and cost comparisons. Crit
256. Mela GS, Savarino V. Inaccuracy of hourly sampled pH mea- Care Med. 1992; 21:339-42.
surements in describing the effect of antisecretory drugs on 274. Chalfin DB, Cohen IL, Lambrinos J. The economics and cost-
circadian gastric acidity. J Clin Pharmacol. 1990; 30:45-9. effectiveness of critical care medicine. Intensive Care Med.
257. Wilder-Smith CH, Halter F, Hackl W et al. pH feedback 1995; 21:952-6.
controlled infusions of ranitidine are no more effective than 275. Jiranek GC, Kozarek RA. A cost-effective approach to the
fixed-dose infusions in reducing gastric acidity and variabili- patient with peptic ulcer bleeding. Surg Clin North Am. 1996;
ty in antisecretory responses. Br J Clin Pharmacol. 1992; 76:83-103.
33:487-93. 276. Roberts MS. Gastrointestinal bleeding in critically ill pa-
258. Doglio GR, Pusajo JF, Egurrola MA et al. Gastric mucosal pH tients. N Engl J Med. 1994; 331:52. Letter.
as a prognostic index of mortality in critically ill patients. 277. Hofstetter J, Allen LV Jr. Causes of non-medication-induced
Crit Care Med. 1991; 19:1037-40. nasogastric tube occlusion. Am J Hosp Pharm. 1992; 49:603-7.
259. Levine JH, Durham RM, Moran J et al. Multiple organ failure: 278. Nicholson LJ. Declogging small-bore feeding tubes. JPEN.
is it disappearing? World J Surg. 1996; 20:471-3. 1987; 11:594-7.