8.

NON-VARICEAL GASTROINTESTINAL HEMORRHAGE

8.1 Introduction
Upper gastrointestinal hemorrhage is a common clinical problem, afflicting approximately one out of every
thousand people each year. In most cases, bleeding stops spontaneously. However a minority rebleeds or
continues to bleed despite attempts at hemostasis. This subpopulation accounts for most of the morbidity,
mortality and resource consumption associated with upper gastrointestinal hemorrhage. Risk stratification
allows targeted application of medical, 164 endoscopic and surgical therapy. Despite remarkable advances in
each of these domains, however, approximately 1 in 20 patients who present with upper gastrointestinal
bleeding will die over the course of their hospitalization.

8.2 Source of Hemorrhage

In most cases of upper gastrointestinal bleeding, a source is identified after careful clinical and endoscopic
evaluation. In approximately 15% of cases, bleeding originates from esophageal or gastric varices associated
with portal hypertension (discussed elsewhere). Among cases of non-variceal upper gastrointestinal
hemorrhage, over 50% are caused by peptic ulcers. Other common sources of bleeding include erosive
gastroduodenitis, esophagitis, Mallory-Weiss tears, angiodysplasia, Dieulafoy lesions and neoplasia.

8.3 Presentation and Risk Stratification

Bleeding from the upper gastrointestinal tract (proximal to the ligament of Treitz) manifests typically with
overt hematemesis or coffee ground emesis, or with passage of melena per rectum. Brisk hemorrhage with
rapid transit can present with maroon stool, hematochezia or features of hemodynamic instability. In all
cases, the priority at initial assessment is to ensure hemodynamic stability and initiate appropriate volume
resuscitation before conducting a detailed history and physical examination. Key features of the history
include: symptoms of hemodynamic instability (such as presyncope); prior upper gastrointestinal and liver
disease with or without hemorrhage; other blood loss suggestive of an underlying bleeding diathesis; use of
medications known to cause gastrointestinal injury (such as aspirin and NSAIDs); alcohol consumption; and
family history of gastrointestinal pathology. On physical examination, key features include serial assessment
of postural vital signs, thorough examination of the abdomen, careful inspection of the skin and mucus
membranes for telangiectasia, assessment for the stigmata of chronic liver disease, and digital rectal
examination. In all cases of overt hemorrhage, care must be taken to exclude respiratory or nasopharyngeal
sources of blood loss. Passage of a nasogastric tube for aspirate can be informative; a biliary aspirate
suggests a source of bleeding distal to the ampulla of Vater, while a bloody aspirate suggests a high-risk
lesion and increased risk of mortality. Upper gastrointestinal endoscopy (ideally within 24 hours of
presentation) is a key component of patient assessment, and is often essential to diagnosis, prognosis and
treatment. In most cases, an experienced endoscopist can localize the source of bleeding and estimate the
risk of rebleeding. Of note, the Forrest classification of peptic ulcer stigmata (first reported in 1974) has
withstood the test of time as a powerful predictor of the risk of rebleeding. By combining clinical and
endoscopic criteria, clinicians can estimate risk with even greater accuracy. The Rockall score combines five
domains (age, comorbidity, hemodynamic stability, bleeding source and Forrest classification) to predict
rebleeding and mortality. Patients at low risk can be discharged home from the emergency department for
outpatient follow-up.

8.4 Endoscopic Therapy

Endoscopic hemostatic therapy has been shown to reduce rebleeding, surgery and death among patients with
high-risk endoscopic stigmata (Forrest classification Ia, Ib or IIa). Both injection therapy (saline +/- 10,000
epinephrine) and thermal coagulation therapy to ablate the bleeding vessel are effective. The combination of
injection therapy plus thermal coagulation therapy is more effective than either intervention alone. In
patients with adherent clots (Forrest classification IIb), management is controversial. Aggressive irrigation to
dislodge the clot and treatment of the underlying lesion is generally accepted. Clinical trials from expert
centres have shown better outcomes when a cold snare is used to remove the clot, but many clinicians are
reluctant to use this technique for fear of precipitating a brisk bleed. The use of endoscopic clips for
hemostasis is a promising technique undergoing assessment in clinical trials. For patients who rebleed after
an initial attempt at endoscopic hemostasis, repeat endoscopy to reassess the lesion and apply further
 2
endoscopic treatment as needed is appropriate. However, routine second-look endoscopy in patients with no
evidence of recurrent bleeding is not advocated.

8.5 Medical Therapy

Acid suppression can improve clot stability and platelet aggregation. Accordingly, medical therapy of non-
variceal upper gastrointestinal hemorrhage is focused on achieving sustained and substantive elevation of
gastric pH. Clinical trials of intravenous histamine-2-receptor antagonists have been disappointing, in part
due to early induction of pharmacologic tolerance. However, an intravenous bolus of omeprazole followed
by an intravenous infusion for 72 hours has been shown in several well-designed clinical trials to reduce the
risk of rebleeding after endoscopy in patients with high-risk endoscopic lesions (Forrest classification Ia, Ib
and IIa). Meta-analyses pooling these trials have also shown intravenous proton pump inhibitors to be
associated with significant reductions in surgery and mortality. Several controversies persist in the medical
management of non-variceal upper gastrointestinal hemorrhage. First, the empiric use of proton pump
inhibitors in patients prior to endoscopy has intuitive appeal but has not been tested in clinical trials. High
doses of oral proton pump inhibitors may also be effective, but no rigorous head-to-head comparison with
intravenous dosing has assessed clinical outcomes. Intravenous infusion of somatostatin analogs such as
octreotide or vapreotide may also reduce rebleeding, and may be useful in patients with significant bleeding
facing delays to endoscopy. Other agents such as tranexamic acid and recombinant factor VII can be
considered in refractory patients, but have not been tested in clinical trials.

8.6 Surgery
Between 5% and 10% of patients who present with acute upper gastrointestinal bleeding will require surgery
because of continued or recurrent hemorrhage. Although this proportion is gradually declining, it remains
substantial as improvements in medical and endoscopic therapies are offset by the increasing age and
comorbidity of patients admitted with gastrointestinal bleeding. The decision to perform surgery must be
individualized, but consider factors such as patient comorbidity, transfusion requirements, the nature of the
bleeding lesion and the anticipated success of further endoscopic therapy. Surgery should be considered
early in patients at high risk of complications such as perforation (e.g., large, deep anterior duodenal ulcers).

8.7 Conclusions
Appropriate management of acute upper gastrointestinal hemorrhage entails early resuscitation and triage,
careful clinical assessment, early endoscopy, intravenous proton pump inhibitors infusion (if indicated) and
access to a skilled surgical team. Given the high prevalence of upper gastrointestinal bleeding, each acute
care hospital and health care system should develop institution-specific protocols for its management. These
protocols should address aspects of triage and multidisciplinary care including access to a therapeutic
endoscopist skilled in endoscopic hemostasis and trained support to assist with urgent endoscopy. Despite
remarkable advances in medical and endoscopic therapy, non-variceal upper gastrointestinal hemorrhage
continues to impose a significant disease burden.
 3

Upper Gastrointestinal Bleeding

Introduction

Background
The diagnosis and therapy for nonvariceal upper gastrointestinal bleeding (UGIB) has evolved over the
past 3 decades from passive diagnostic esophagogastroduodenoscopy with medical therapy until
surgical intervention was needed to active intervention with endoscopic techniques followed by
angiographic and surgical approaches if endoscopic therapy failed. Variceal hemorrhage is not
discussed in this article because the underlying mechanisms of bleeding are different and require
different therapies. The underlying mechanisms of nonvariceal bleeding involve either arterial
hemorrhage, such as in ulcer disease and mucosal deep tears, or low-pressure venous hemorrhage,
as in telangiectasias and angioectasias. In variceal hemorrhage, the underlying pathophysiology is due
to elevated portal pressure transmitted to esophageal and gastric varices and resulting in portal
gastropathy.

The patient presents with an ulcer that has bled or is actively bleeding. Approximately 80% of ulcers
stop bleeding. The overall mortality rate is approximately 10%. This patient population has become
progressively older, with significant comorbidities that increase mortality. Rebleeding or continued
bleeding is associated with increased mortality; therefore, differentiating the patient with a low
probability of rebleeding and little comorbidity from the patient at high risk for rebleeding with serious
comorbidities is imperative.

In the early history of endoscopy for UGIB, multiple published studies questioned the cost-
effectiveness of endoscopy in this setting because it was unclear whether the outcome was changed.
In a setting in which 80% of patients respond to conservative medical management, studies were
hampered by type 2 errors because of the large number of patients needed to demonstrate statistical
significance. Much debate also focused on the significance of the nonbleeding visible vessel (ie, color,
size, diagnostic characteristics, risk of rebleeding) in ulcer hemorrhage. These matters became
clarified after the characteristics and the significance of the visible vessel in the ulcer crater were
defined and the evidence for endoscopic therapy was established, demonstrating that patients
requiring therapy to control bleeding or rebleeding could be diagnosed and treated at the time of the
upper endoscopy.

In 1989, a National Institutes of Health consensus conference on UGIB concluded that effective
therapy was needed in the presence of active bleeding or a visible vessel. The consensus conference
affirmed that the treatment, when performed by an experienced endoscopist using 1 of 4 techniques
(ie, injection of epinephrine or sclerosants, heater-probe coagulation, bipolar electrode coagulation,
laser coagulation), was proven effective by the published evidence. Three other techniques have since
been developed: (1) endoscopic application of clips, (2) use of banding devices, and (3) argon plasma
coagulation.

Emergency surgery typically entails oversewing the bleeding vessel in the stomach or duodenum
(usually preoperatively identified by endoscopy), vagotomy with pyloroplasty, or partial gastrectomy.
Angiographic obliteration of the bleeding vessel is considered in patients with poor prognoses.

Other causes of gastrointestinal bleeding include mucosal tears in the esophagus or upper stomach
due to vomiting (Mallory-Weiss tears), venous blebs, or vascular ectasias. These lesions can be
treated with endoscopic coagulation. The bleeding from gastric cancers and ulcers in leiomyomas does
not usually respond to endoscopic therapy; surgical or radiologic intervention is needed.
Pathophysiology
Duodenal ulcer disease is strongly associated with Helicobacter pylori infection. The organism causes
disruption of the mucous barrier and has a direct inflammatory effect on gastric and duodenal mucosa.
 4
Eradication of H pylori has been demonstrated to reduce the risk of recurrent ulcers and, thus,
recurrent ulcer hemorrhage.

Nonsteroidal anti-inflammatory drugs (NSAIDs) are the second major etiology of ulcer hemorrhage
because of their effect on cyclooxygenase-1, which leads to impaired mucosal defense to acid. The
use of cyclooxygenase-2 inhibitors has been shown to reduce the risk of ulcer hemorrhage, although
only when not combined with aspirin therapy. Concerns have been raised about an increase in
myocardial infarction and stroke in patients taking selective cyclooxygenase-2 inhibitors. As
demonstrated in the study by al-Assi et al, the combination of H pylori infection and NSAID use may
increase the risk of ulcer hemorrhage; however, the treatment of H pylori in patients who are taking
NSAIDs remains controversial.1

As the ulcer burrows deeper into the gastroduodenal mucosa, the process causes weakening and
necrosis of the arterial wall, leading to the development of a pseudoaneurysm. The weakened wall
ruptures, producing hemorrhage. The flow through the vessel varies with the fourth power of the
radius; thus, small increases in vessel size can mean much larger amounts of blood flow and bleeding.
Visible vessels usually range from 0.3-1.8 mm.

Exsanguinating hemorrhage has been reported from larger vessels. The larger vessels are located
deeper in the gastric and duodenal submucosa and serosa. Larger branches of the left gastric artery
are found high on the lesser curvature, while the pancreatoduodenal artery and its major branches are
located posteroinferiorly in the duodenal bulb. The size of the vessel is important in the prognosis in
that larger vessels cause faster blood loss, with more severe hypotension and more complications,
especially in older patients.

During vomiting, the lower esophagus and upper stomach are forcibly inverted. Vomiting attributable to
any cause can lead to a mucosal tear of the lower esophagus or upper stomach. The depth of the tear
determines the severity of the bleeding. Rarely, vomiting can result in esophageal rupture (Boerhaave
syndrome), leading to bleeding, mediastinal air entry, left pleural effusion (salivary amylase can be
present) or left pulmonary infiltrate, and subcutaneous emphysema. Superficial (to submucosal layer)
tears cause Mallory-Weiss syndrome.

Gastric cancer is an uncommon cause of hemorrhage in the United States, but it remains a major
problem in non-Western countries; worldwide, it is the leading cause of digestive cancer deaths.
Patients with chronic liver disease and portal hypertension are at increased risk for variceal
hemorrhage and portal gastropathy in addition to ulcer hemorrhage.
Frequency
United States

UGIB is a common medical condition that results in high patient mortality and medical care costs.
Annually, approximately 100,000 patients are admitted to US hospitals for therapy for UGIB. Peptic
ulcer disease is the most common cause of UGIB. However, the proportion of cases caused by peptic
ulcer disease has declined.2 The decrease is believed to be due to the use of proton pump inhibitors
(PPIs) and H pylori therapy.

International

UGIB is a common occurrence throughout the world. In France, a report concludes that the mortality
from UGIB has decreased from about 11% to 7%; however, a similar report from Greece finds no
decrease in mortality. In a nationwide study from Spain, UGIB was 6 times more common than lower
GI bleeding.3
Mortality/Morbidity
Patients typically present with an ulcer that has bled or is actively bleeding, but approximately 80% of
ulcers stop bleeding. The overall mortality rate is approximately 10%. In a retrospective chart review by
Yavorski RT et al, 73.2% of deaths occurred in patients older than 60 years. 4 In patients with UGIB,
comorbid illness and not actual bleeding is the major cause of death. Comorbid illness was noted in
50.9% of patients, with similar occurrence in males (48.7%) and females (55.4%). One or more
comorbid illnesses were noted in 98.3% of patients who died, and, in 72.3% of patients, comorbid
 5
illnesses were the primary cause of death.  
4

According to the American Society for Gastrointestinal Endoscopy (ASGE), the following risk
factors are associated with increased mortality, recurrent bleeding, the need for endoscopic
hemostasis, or surgery: age older than 60 years, severe comorbidity, active bleeding (eg, witnessed
hematemesis, red blood per nasogastric tube, fresh blood per rectum), hypotension, red blood cell
transfusion greater than or equal to 6 units, inpatient at time of bleed, and severe coagulopathy. 5

An increasing amount of evidence in the literature states that therapy with high-dose PPIs (IV bolus
followed by continuous infusion) may decrease the rate of rebleeding after endoscopic therapy. By
increasing the gastric pH above 6, the clot is stabilized.

Sex
The incidence of UGIB is 2-fold greater in males than in females, in all age groups; however, the death
rate is similar in both sexes.4
Age
This patient population has become progressively older, with significant comorbidities that increase
mortality. As mentioned above, the mortality increases with older age (>60 y) in both males and
females.

Clinical

History
 The patient history findings include weakness, dizziness, syncope associated with hematemesis
(coffee ground vomitus), melena (black stools with a rotten odor), and hematochezia (red or
maroon stool).
 Patients may have a history of previous dyspepsia (especially nocturnal symptoms), ulcer
disease, early satiety, and nonsteroidal anti-inflammatory drug or aspirin use. Many patients
with UGIB who are taking nonsteroidal anti-inflammatory drugs present without dyspepsia but
with hematemesis or melena as their first symptom. Low-dose aspirin (81 mg) has been
associated with UGIB with or without the addition of NSAID therapy. Patients with a prior history
of ulcers are at an especially increased risk for UGIB when placed on aspirin or NSAID therapy
and should receive continuous acid suppression with a PPI.
 Because recurrence of ulcer disease is common, history findings are relevant.
o Patients may present in a more subacute phase with a history of dyspepsia and occult
intestinal bleeding manifesting as a positive fecal occult blood test result or as iron
deficiency anemia.
o A history of recent aspirin ingestion suggests that the patient may have nonsteroidal anti-
inflammatory drug gastropathy with an enhanced bleeding diathesis from poor platelet
adhesiveness.
o A history of chronic alcohol use of more than 50 g/d or chronic hepatitis (B or C)
increases the risk of variceal hemorrhage, gastric antral vascular ectasia (GAVE), or
portal gastropathy.
o The presence of postural hypotension indicates more rapid and severe blood loss.

Physical
 The goal of the patient's physical examination is to evaluate for shock and blood loss.
o Pulse and blood pressure should be checked with the patient in supine and upright
positions to note the effect of blood loss. Significant changes in vital signs with postural
changes indicate an acute blood loss of approximately 20% or more.
o Other signs of shock include cool extremities, oliguria, chest pain, presyncope,
confusion, and delirium.
o Hematemesis and melena should be noted. The redder the stool, the more rapid the
transit, which suggests a large upper tract hemorrhage.
 6
 Signs of chronic liver disease should be noted, including spider angiomata, gynecomastia,
increased luneals, splenomegaly, ascites, pedal edema, and asterixis.
 Signs of tumor are uncommon but portend a poor prognosis. Signs include a nodular liver,
abdominal mass, and enlarged and firm lymph nodes.
 The finding of subcutaneous emphysema with a history of vomiting is suggestive of Boerhaave
syndrome (esophageal perforation) and requires prompt consideration of surgical therapy.
 The finding of telangiectasias may indicate the rare case of Osler-Weber-Rendu syndrome.

Causes
 The major causes of UGIB are duodenal ulcer hemorrhage (25%), gastric ulcer hemorrhage
(20%), mucosal tears of the esophagus or fundus (Mallory-Weiss tear), esophageal varices,
erosive gastritis, erosive esophagitis, Dieulafoy lesion, gastric varices, gastric cancer, and
ulcerated gastric leiomyoma.
 Rare causes of UGIB include aortoenteric fistula, gastric antral vascular ectasia, angiectasias,
and Osler-Weber-Rendu syndrome.
 The proportion of UGIB cases caused by peptic ulcer disease has declined. 2 This decline is
believed to be due to the use of PPIs and H pylori therapy.
 Patients should be considered for upper endoscopy if blood loss from the upper gastrointestinal
tract is suspected. A high level of suspicion of UGIB should exist when the patient has a history
of intake of aspirin or NSAID, even if no history of hematemesis or melena exists. The color of
stool containing blood depends on the transit time; rapid transit from the upper gastrointestinal
tract can result in red or maroon stools. Melena results from more than 100 mL of blood with
moderate transit time. Slow transit of blood from the lower intestine can result in melena in the
presence of obstruction.
 Urgent endoscopy is indicated when patients present with hematemesis, melena, or postural
changes in blood pressure. Cooper et al have demonstrated a lower rate of rebleeding and
shorter length of stay when endoscopy is performed within 24 hours of admission. 6
 Primary surgical intervention should be considered in patients with a perforated viscus (eg, from
perforated duodenal ulcer, perforated gastric ulcer, or Boerhaave syndrome). In patients who
are poor operative candidates, conservative treatment with nasogastric suction and broad-
spectrum antibiotics can be instituted. Endoscopic clipping or sewing techniques have also
been used in such patients.

Workup

Laboratory Studies
 CBC with platelet count and differential: CBC is necessary to assess the level of blood loss.
Where possible, having the patient's previous results is useful to gauge the level of blood loss.
CBC should be checked frequently (q4-6h) during the first day.
 Basic metabolic profile (BMP): The BMP is useful to evaluate for renal comorbidity; however,
blood in the upper intestine can elevate the BUN level as well. Measurement of coagulation
parameters is necessary to assess for continued bleeding. Abnormalities should be corrected
rapidly.
 Prothrombin time/activated partial thromboplastin time
 Liver profile: The liver profile can identify hepatic comorbidity and suggest underlying liver
disease.
 Calcium level: A calcium level is useful to identify the patient with hyperparathyroidism as well
as to monitor calcium in patients receiving multiple transfusions of citrated blood.
 Gastrin level: A gastrin level can identify the rare patient with gastrinoma as the cause of UGIB
and multiple ulcers.

Imaging Studies
 Chest radiographs should be ordered to exclude aspiration pneumonia, effusion, and
esophageal perforation; abdominal scout and upright films should be ordered to exclude
perforated viscus and ileus.
 7
 Barium contrast studies are not usually helpful and can make endoscopic procedures more
difficult (ie, white barium obscuring the view) and dangerous (ie, risk of aspiration).
 CT scan and ultrasonography may be indicated to evaluate liver disease with cirrhosis,
cholecystitis with hemorrhage, pancreatitis with pseudocyst and hemorrhage, aortoenteric
fistula, and other unusual causes of upper gastrointestinal hemorrhage.
 Nuclear medicine scans may be useful to determine the area of active hemorrhage.
 Angiography may be useful if bleeding persists and endoscopy fails to identify a bleeding site.
As salvage therapy, embolization of the bleeding vessel can be as successful as emergent
surgery in patients who have failed a second attempt of endoscopic therapy.

Other Tests
 An ECG should be ordered to exclude arrhythmia and cardiac disease, especially acute
myocardial infarction due to hypotension.
 Esophagogastroduodenoscopy may increase the risk of arrhythmias.
 Performing a troponin test may be useful to identify patients with severe coronary ischemia or
atypical myocardial infarction.

Procedures
 Nasogastric lavage
o This procedure may confirm recent bleeding (coffee ground appearance), possible active
bleeding (red blood in the aspirate that does not clear), or a lack of blood in the stomach
(active bleeding less likely but does not exclude an upper gastrointestinal lesion).
o A nasogastric tube is an important diagnostic tool, and tube placement can reduce the
patient's need to vomit. Placement for diagnostic purposes is not contraindicated in
patients with possible esophageal varices.
o The characteristics of the nasogastric lavage fluid (eg, red, coffee grounds, clear) and
the stool (eg, red, black, brown) can indicate the severity of the hemorrhage. Red blood
with red stool is associated with an increased mortality rate from more active bleeding
compared with negative aspirate findings with brown stool.

Histologic Findings
The bleeding vessel lies in the deepest layer of the ulcer. Fibrinoid necrosis is observed at the site of
perforation of the vessel. Pseudoaneurysmal dilation of the vessel may be present at the site of
perforation. Biopsy samples should be taken from the edge of a gastric ulcer to rule out carcinoma.
The characteristic lesion of H pylori is chronic active gastritis with the organisms observed after routine
staining. The lesion of gastric antral vascular ectasia is capillary dilation with fibrin clots and
fibromuscular hyperplasia.

Treatment

Medical Care
 Stabilize the patient with intravenous fluids (usually normal saline, except in patients with severe
liver disease, ascites, or heart failure) and transfuse to maintain a hemoglobin level of 8-10 g.
Promptly correct any abnormalities in coagulation. Baradarian et al demonstrated that early,
aggressive resuscitation can reduce mortality in acute UGIB. 7
 The goal of medical therapy is to correct shock and coagulation abnormalities and to stabilize
the patient so that further evaluation and treatment can proceed. The primary problem is a
perforated arterial (high pressure) vessel with a potential to rebleed; therefore, pharmacologic
therapy should be used only after endoscopic therapy.
 The use of H2-receptor antagonists has not been shown to be effective in altering the course of
UGIB. A meta-analysis concluded that there was a possible minor benefit with intravenous H2
antagonists in bleeding gastric ulcers but no benefit in duodenal ulcers. 8
 The relative efficacy of the PPIs may be due to their superior ability to maintain a gastric pH at a
level above 6.0 and, thus, protect an ulcer clot from fibrinolysis. 9 Considering the available data,
the ideal pharmacologic therapy for patients with acute ulcer bleeding appears to be an
intravenous PPI.Pantoprazole, lansoprazole, and esomeprazole are the only PPIs available as
 8
an intravenous formulation in the United States; intravenous omeprazole is used in other
countries. The suggested dose of intravenous pantoprazole is 80-mg bolus followed by 8-mg/h
infusion. The infusion is continued for 48-72 hours. This therapy has been shown to be cost-
effective by Barkun et al.10 
 A meta-analysis of 24 randomized controlled trials that evaluated PPIs for bleeding ulcers (with
or without endoscopic therapy) found a significant reduction in the risk of rebleeding, the need
for repeat endoscopic hemostasis, and surgery. An improvement in mortality was also seen in
Asian trials and in patients with active bleeding or nonbleeding visible vessels.11
 High-dose oral PPIs may be used in patients who do not have active bleeding or other high-risk
stigmata for recurrent bleeding (eg, a visible vessel, adherent clots); in such patients, the risk of
recurrent bleeding is low. The goal of treatment in these patients (following resuscitation) should
be directed at healing the ulcers and at eliminating precipitating factors (eg, H pylori, NSAIDs).
A combined analysis of 5 studies that evaluated oral dosing (with or without endoscopic
therapy) found a significant reduction in the risk of rebleeding and surgery. 12

Surgical Care
 Specific characteristics at endoscopy can predict rebleeding.
o Rebleeding occurs in 55% of patients who have active bleeding (pulsatile, oozing), in
43% who have a nonbleeding visible vessel, in 22% who have an ulcer with an adherent
clot, and in 0-5% who have an ulcer with a clean base.
o At endoscopy, the prevalence rate for a clean base is 42%, for a flat spot is 20%, for an
adherent clot is 17%, for a visible vessel is 17%, and for active bleeding is 18%.
o Freeman et al have described a pale visible vessel that appears to have a very high risk
for rebleeding.13 This must be differentiated from the presence of a clean ulcer base.
Good visualization is important. The uncleared fundal pool may obscure an ulcer,
mucosal tear, gastric varices, portal gastropathy, or tumor (eg, leiomyoma,
adenocarcinoma, lymphoma). Endoscopic therapy is recommended for ulcers at
increased risk for rebleeding.
 Contraindications to upper endoscopy include an uncooperative or obtunded patient, severe
cardiac decompensation, acute myocardial infarction (unless active, life-threatening
hemorrhage is present), and perforated viscus (eg, esophagus, stomach, intestine).
 Contraindications to emergency surgery include impaired cardiopulmonary status and bleeding
diathesis.
 Esophagogastroduodenoscopy may be more difficult or impossible if the patient has had
previous oropharyngeal surgery or radiation therapy to the oropharynx.
o The presence of a Zenker diverticulum can make intubation of the esophagus more
difficult.
o Patients with Down syndrome are more sensitive to conscious sedation and should
receive much less sedation, or they should be monitored by an anesthesiologist and/or
intubated prophylactically prior to the procedure.
o Hypotension may be exacerbated by sedation; therefore, patients who are unstable
should be given less sedation.
o Patients with massive bleeding should be considered for intubation to reduce the
increased risk of aspiration. Such patients should be treated in an intensive care setting.
o Ideally, the patient should be stabilized prior to endoscopy and abnormalities in
coagulation should be corrected. When this is not possible, the judgment of an
experienced endoscopist is vital.
o The patient should undergo upper endoscopy prior to any operative intervention in order
to diagnose and localize the bleeding site. Most patients (85-90%) respond to
endoscopic therapy.
 During the endoscopy, the patient is monitored according to analgesia and sedation guidelines
formulated by the American Society of Anesthesiology. The characteristics of the bleeding
lesion are noted, and appropriate therapy is applied when necessary for high-risk lesions or
active bleeding.
 The techniques used are heater probe coaptive coagulation, bipolar probe coaptive coagulation,
sclerosant or epinephrine injection, laser therapy, and hemostatic clip placement.
 9
o A combination of therapies has become more common. Injection therapy is applied first
to better clarify the bleeding site, especially in the actively bleeding patient; then, heater
probe or bipolar (gold) probe coagulation is applied. Injection therapy can also be
performed prior to endoscopic placement of hemoclips. Injection therapy is useful prior to
laser therapy to reduce the heat sink effect of rapidly flowing blood prior to laser
coagulation.
o Reports have been published on the use of argon plasma coagulation for ulcer
hemorrhage; however, there are significant theoretical and experimental considerations
from animal models about the use of a noncontact monopolar cautery in this situation. In
the animal model, obliterative closure is effective only for vessels less than 1 mm in
diameter and causes deeper tissue damage.
o The heater probe consists of a resistor electrode enveloped by a titanium capsule and
covered by Teflon to reduce sticking to the mucosa. The probe temperature rises to
250°C (482°F).
o The bipolar probe consists of alternating bands of electrodes producing an electrical field
that heats the mucosa and the vessel. The electrodes are coated with gold to reduce
adhesiveness. The probes are stiff to allow adequate pressure to the vessel to appose
the walls and thus produce coaptive coagulation when the energy is transmitted (heat
energy by the heater probe, electrical-field energy by the bipolar probe). Careful
technique is required to heat-seal the perforated vessel.
o Alternatively, epinephrine (1:10,000, 1:20,000) can be injected in 0.5-mL aliquots around
the base of the vessel. This causes coagulation by compression, with perhaps some
additive effect of activation of platelet factor 3 or vasoconstriction. Other solutions that
have been used for injection are ethanol (more necrosis), hypertonic saline, sterile water,
and cyanoacrylate (not available in the United States). Epinephrine injection is often
used to reduce the volume of bleeding so that the lesion can be better localized and then
treated with a coaptive technique (ie, heater probe, gold probe). Such combination
therapy has become more frequent and has evolved into the standard technique.
o Laser therapy is rarely used. To perform laser coagulation, the area near the vessel is
first injected with epinephrine to reduce blood flow (reducing the heat-sink effect); then,
the laser is applied around the vessel (producing a wall of edema). Caution must be
observed to avoid drilling into the vessel with the laser, causing increased bleeding.
o Hemostatic clips are available in the United States.
 Modification of the delivery system has made clip placement much easier than
with the original model. With careful placement of the clip, closing the defect in
the vessel is possible. Usually, multiple clips are applied. They vary in the size
and the strength of the clip. Four models of hemoclips are available: QuickClip2,
which is rotatable; Resolution Clip, which can be reopened after closure; TriClip,
which has 3 prongs; and InScope, which is a multiclip applier with 4 endoclips.
 Considering the available data, the hemoclips have similar efficacy as thermal
coagulation methods. One report of 113 patients with major stigmata of ulcer
hemorrhage found no difference in hemostasis, 30-day mortality, or the need for
emergency surgery.14 Patients randomized to the endoclip group had significantly
lower rebleeding rates (2% vs 21%). However, only 60% of active bleeders were
successfully treated with the heater probe, a rate much lower than in previous
reports.
 Another study of 80 patients found a higher rate of control of initial bleeding with
the heater probe compared with the Olympus endoclip (100% vs 85%). 15
Rebleeding rates were not significantly different. No significant differences in
procedure duration, initial hemostasis, or rebleeding rates in a study of 47
patients comparing combination therapy with epinephrine injection plus
monopolar electrocoagulation versus hemoclips.16 
 There are some clinical settings in which endoclips may be preferred to other
hemostatic methods: (1) treatment of ulcers in patients who are coagulopathic or
who require ongoing anticoagulation; in such patients, electrocoagulation will
increase the size, depth, and healing time of treated lesions; and (2) retreatment
of lesions that rebleed after initial thermal hemostasis.
 10
 Animal studies support reduced injury after ulcer hemostasis with hemoclips
versus thermal coagulation.17 However, ulcers on the lesser curvature, the
posterior duodenum, or the cardia increase the difficulty of deployment and clip
failure rates. Larger endoclips have advantages over smaller hemoclips for
hemostasis of chronic ulcers and closure of larger lesions.
o Argon plasma coagulation is a technique in which a stream of electrons flows along a
stream of argon gas. The coagulation is similar to monopolar cautery with the current
flow going from a point of high current density, the point of contact of the gas with the
mucosa, to an area of low current density, the conductive pad on the patient's body. The
current flows through the body in an erratic path to the pad. This monopolar cautery
technique is similar to the laser technique in that energy is delivered to the vessel for
coagulation with apposition of the vessel walls. This technique was not effective for
visible vessels larger than 1 mm. No animal models have been used for ulcer
hemorrhage to validate this technique.
 The choice of treatment modality is influenced by the size of the vessel. Animal studies have
demonstrated that the heater probe and bipolar probe are effective for vessels as large as 2 mm
in diameter. Other techniques (eg, clips, band ligation) or a combination of techniques is needed
for larger vessels or vessels not approachable by the heater probe or bipolar probe. Surgical
intervention should be considered when dealing with vessels larger than 2 mm in diameter
(discounting the enlargement due to the development of pseudoaneurysm). Angiographic
embolization is an option in the patient at high risk for surgical intervention.
 In the patient who has an ulcer with an overlying clot, attempting to remove the clot by target
washing is critical. Endoscopic removal of the clot by washing or cold snare has been
demonstrated to be effective in reducing the recurrence of bleeding. 18
o The findings under the clot (eg, bleeding vessel, visible vessel, flat spot, clean base)
help determine the therapy needed and improve efficacy by allowing treatment to be
applied directly to the vessel.
o If the clot cannot be removed by washing, then cutting away the clot using a cold snare
can be considered by experienced endoscopists.
o Vigorous washing of the clot formed after therapy is useful to determine the adequacy of
coagulation. A combination of injection with heater probe or bipolar coaptive coagulation
is often used and has been shown to be more effective in patients with active bleeding.
 In an effort to improve the teaching of these techniques, an in vitro pig stomach model has been
developed by Hochberger.19 This model has been used with first-year fellows in
gastroenterology in New York and Germany.19
 The patient is monitored under the protocol for conscious sedation, also called analgesia and
sedation (ie, per American Society of Anesthesiologists and American Society for
Gastrointestinal Endoscopy guidelines).
 Rebleeding occurs in 10-30% of endoscopically treated patients. A second attempt at
endoscopic control is warranted. Some authorities have concerns about the perils of a second
esophagogastroduodenoscopy, which may result in delayed surgery, perforation, and increased
morbidity and mortality rates. This approach has been validated in a large, randomized
controlled trial that showed decreased morbidity and mortality rates.
 Using a combination of techniques is prudent when re-treating the ulcer site because the first
therapy produced necrosis and weakening of the intestinal wall. Ulcers on the anterior surface
of the stomach and duodenum are at increased risk for perforation. Using injection as the first
step increases the thickness of the submucosal layer, thus providing an extra margin of safety.
 Even operative techniques can have a significant rebleeding rate with significant mortality, as
noted in the study of Poxon et al in which the rebleeding rate was 10% (80% mortality for
rebleeders) in patients who underwent a conservative surgical technique in which the ulcer base
was undersewn.20 This more conservative approach was compared with the standard surgical
technique (ie, vagotomy and pyloroplasty or partial gastrectomy). The comparison of the
conservative approach with a standard gastrectomy resulted in similar mortality rates, ie, 26%
versus 19%, respectively, with no rebleeding after partial gastrectomy.
 Blind total gastrectomy has been studied by Primrose et al as a last resort when the UGIB site
cannot be determined in a patient who continues to bleed. 21
 11
 Bouillot et al report the use of a combined endoscopic and laparoscopic approach in a patient
with a bleeding Dieulafoy lesion unresponsive to endoscopic therapy. 22 Intraoperative
endoscopy located the small bleeding lesion, which was removed by a limited resection.
 Postoperatively, the patient is monitored for recovery from conscious sedation after endoscopy
and from general anesthesia after abdominal surgery. Monitor the patient's mental status, vital
signs, chest, cardiac, and abdominal findings to ascertain that the patient's clinical status has
stabilized and that no complications (eg, aspiration, perforation, recurrent bleeding, myocardial
infarction due to hypotension) have occurred. Monitor the hemoglobin level.

Consultations
Consultation with a surgeon should be considered for all patients with gastrointestinal hemorrhage.