Editor: Joseph Sung
4. Upper GI bleeding (non-variceal)
The incidence of acute bleeding from the upper digestive tract has lessened somewhat in recent years, owing mainly to the
reduction in ulcer disease. However, ulcers still account for approximately half of all cases of upper gastrointestinal bleeding
requiring hospital treatment. Varices are the second commonest cause of serious bleeding. Bleeding can occur also from erosions,
tears, tumors and vascular malformations. This contribution concentrates on ulcer bleeding; varices are addressed in another
Fortunately, ulcer bleeding stops spontaneously in 7580% of patients, with a relatively benign outcome. However, despite the numerous advances in endoscopic and medical therapy,
intensive care and surgery, the mortality rate can be high as 810%. Mortality occurs in patients with massive and recurrent bleeding, especially in elderly patients with major comorbidities.
Bleeding which commences in hospitalized patients is especially serious. Approximately a third of the patients with significant
ulcer bleeding will develop recurrent bleeding on long-term follow-up, in the absence of active intervention.
Much progress has been made in methods for endoscopic diagnosis and treatment of patients with acute bleeding over recent
years. This contribution reviews methods for hemostasis, the care of patients after treatment, and the prediction and prevention
of rebleeding. It concludes with some hopes for future progress in this field.
Although the clinical history and physical examination may provide some clues about the cause of bleeding (e.g. aspirin-induced
peptic ulcer), they are too non-specific to be useful clinically, as is the result of nasogastric lavage.
Skillful modern endoscopy is > 95% accurate in determining the cause of bleeding in patients presenting with upper gastrointestinal
hemorrhage, especially when performed within 48 h of the onset. Barium study should be avoided in acute bleeding, since it
interferes with endoscopic visualization necessary for hemostasis. Angiography is only helpful in patients with active bleeding.
Routine early endoscopy was avoided in the past, because one of the first randomized studies comparing it to routine endoscopy
did not show any improvement in morbidity or mortality unsurprisingly, since endoscopy was used only to diagnose rather than to treat the cause of the bleeding. In addition, mortality
is not always the best end point to use in clinical bleeding studies, because its low occurrence requires a large sample size.
However meta-analyses have shown that endoscopic hemostasis reduces mortality [2,3]. Finally over-emphasis on mortality tends to ignore other important outcomes such as rebleeding, transfusion, hospital stay,
cost, and need for surgery, all of which are significantly improved by endoscopic therapy.
Early endoscopy is helpful in a different way for the 5060% of patients with peptic ulcer bleeding who don't need hemostasis, because it can precisely predict the clinical outcome.
This information can then be used to triage patients with good outcome to less intensive and thus expensive care as outpatients
Endoscopy, when and where?
Patients who continue in hemodynamic unstability require urgent endoscopy, preferably in the intensive care unit. All others
can undergo endoscopy as soon as is practical, hopefully within 2448 h. Patients with clean-based ulcers who have negligible risk of rebleeding could then be discharged after the procedure (Fig. 1). One practical approach is to perform the endoscopy before admitting the patient from the emergency department, provided
that the endoscopy unit is open. Patients presenting after hours could be held in the emergency department to undergo the
procedure during the next available session. Otherwise, admission to the intensive care unit should be reserved for patients
with continued hemodynamic instability after fluid resuscitation (~ 2 L) or with a comorbid condition requiring the intensive care unit, such as myocardial ischemia.
Lavage before endoscopy?
Lavage is usually unnecessary before endoscopy, because placing the patient in the left lateral position pools the blood in
the fundus and the greater curve of the stomach, and usually exposes most of the bleeding lesions, which tend to be in the
oesophagus, the lesser curve, the antrum, or the duodenal bulb. If the bleeding lesion is not seen then the patient can be
turned, the head of the bed elevated, blood suctioned (connect the suction tubing directly to the biopsy channel of the endoscope),
or endoscopy repeated 12 hours after administering erythromycin or metoclopramide. The chance of missing a second bleeding lesion under a blood pool
is low for patients with non-variceal bleeding, but is not obviously zero, and thus all such patients should undergo repeat
endoscopy in 2448 hours .
Endoscopy will show one of three possible findings in patients with ulcer bleeding. Approximately 4045% have a clean-based ulcer; 40% have stigmata of recent hemorrhage, and 1520% have either oozing or spurting .
An adherent clot (i.e. one resistant to washing) is found in up to 15% of cases (Fig. 2). An adherent clot should be removed to expose any underlying stigmata for treatment, as the risk of rebleeding is almost
as high for clots as for stigmata of recent hemorrhage . Probably the safest way to remove an adherent clot is to inject liberally 1 : 10 000 epinephrine through the clot into the ulcer base then transect it using a cold snare.
Many endoscopic methods for treatment have been developed and evaluated. Injection, 10 Fr heat probe, 10 Fr multipolar electrocoagulation
(MPEC), and laser have been shown in randomized controlled studies to be effective at stopping active bleeding and preventing
rebleeding in patients with stigmata of recent hemorrhage, significantly reducing transfusion, hospital stay, cost of care,
and urgent surgery. Although no single study has been or will be likely to have the power needed to demonstrate significant
reduction in mortality, meta-analysis has shown mortality reduction in the range of 45% . The results of endoscopic hemostasis appear comparable regardless of the method used for patients with non-bleeding stigmata
of recent hemorrhage and oozing ulcers, but combination therapy using epinephrine injection and thermocoagulation is significantly
more effective compared to injection alone for spurting ulcer bleeding (hemostasis of 93% vs. 70%) . On the basis of these data, a large-channel therapeutic endoscope should be used in every patient with upper gastrointestinal
bleeding, since it cannot be determined in advance which patient will have spurting hemorrhage and require combination therapy
using a 10 Fr thermal probe. Finally, clipping appears to be as effective if not better than thermocoagulation (rebleeding
rate of 2% vs. 21%) for ulcer bleeding, and can be deployed using a diagnostic endoscope .
Effective agents for injection therapy include epinephrine, saline, glucose, alcohol, and sclerosant for active bleeding and
alcohol and sclerosant for non-bleeding stigmata of recent haemorrhage. Epinephrine is injected in 1 mL aliquots around and into the bleeding point until bleeding stops and mucosal blanching occurs; typically this requires
1020 mL. Epinephrine injection is safe and can be repeated with minimal risk. Alcohol is injected in 0.10.2 mL aliquots for a total dose of 0.51.0 mL. Repeat therapy using alcohol or a sclerosant increases the risk of perforation.
Thermocoagulation using a 10 Fr probe, but not a 7 Fr probe has been shown to improve significantly the outcome of patients
with peptic ulcer bleeding, partly because only the larger probe provides enough rigidity to tamponade the vessel effectively.
Thermal devices can coagulate an artery up to 2 mm in diameter under experimental conditions, which fortunately is larger than the average diameter of a bleeding ulcer artery
(0.7 mm range). The heat probe should be set at 30 J and applied for 46 pulses before changing position, and the MPEC probe set at 15 watts and applied for 1520 s before changing position (Fig. 3). Thermocoagulation is usually performed at four quadrants surrounding the lesion, followed by application to the bleeding
point. The end-point of hemostasis is cessation of bleeding and obliteration of the previously raised stigmata, i.e. the scorched
earth appearance. Although thermocoagulation is very safe, repeat therapy increases the risk of perforation.
Combination therapy using injection of 1 : 10 : 000 epinephrine (1 mL aliquots for total of ~10 mL) followed by thermocoagulation should be used in patients with spurting ulcers . In addition to providing superior hemostasis, injection slows the bleeding and facilitates the accurate placement of the
Care after bleeding
Resuming a regular diet after endoscopy does not harm the patient. Alternatively a clear liquid diet could be considered in
patients with questionable hemostasis or if repeat endoscopy is needed to clear the stomach.
All patients should receive high-dose intravenous proton pump inhibitor therapy (e.g. omeprazole or pantoprazole 80 mg bolus followed by 8 mg/ h continuous infusion) for three days, followed by full-dose oral therapy for ulcer healing 
All patients should be tested for H. pylori, preferably using at least two tests and with eradication confirmed in infected patients.
Aspirin and NSAIDs
All patients should be screened for NSAID and/or aspirin use and their use discontinued whenever possible.
Patients with stigmata of recent hemorrhage could be discharged 2448 hours after resuscitation and hemostasis and patients with spurting hemorrhage 4872 hours later, as the risk of rebleeding decreases rapidly over the initial 72 hours.
Recurrent bleeding may occur early, within hours to days of the index bleed, or late, after months to years. While there is
no formal definition as to what constitutes early rebleeding, it seems reasonable to define it as occurring during the hospitalization
or within 57 days of the onset of bleeding. Late rebleeding is usually defined as recurrent hemorrhage requiring hospitalization and
It can be difficult at times to diagnose early rebleeding, since many patients with presumed rebleeding do not undergo repeat
endoscopy. The diagnosis can be ambiguous even in patients undergoing repeat endoscopy, since the ulcer may have already stopped
bleeding by the time the procedure is performed. For these reasons, most studies do not require endoscopic confirmation of
active bleeding, but rely on some combination of hemodynamic parameter, hematemesis, ongoing melena, blood transfusion requirement,
and falling blood count to diagnose rebleeding.
Predictors of rebleeding
Early rebleeding after hemostasis has been often (but not always) associated older age, comorbid illness, posterior interior
duodenal bulb ulcer, large ulcer size, active bleeding, and hypotension. The best characterized predictor of early rebleeding
is the endoscopic appearance of the ulcer.
A clean-based ulcer means that the bleeding point has already healed. Stigmata of a recent hemorrhage (also called a pigmented
protuberance or a visible vessel) usually take the form of a clot overlying a ruptured artery or a pseudoaneurysm, and are
associated with recent bleeding . Finally, active bleeding can be seen as oozing or spurting. Untreated patients with clean-based ulcers will rebleed at a
rate of 05%, those with flat spots at 510%, and those with pigmented protuberances at 4045%. A patient with arterial spurting will continue to bleed or rebleed in > 75% of cases (Fig. 4). If we accept that a clean-based ulcer does not have a bleeding lesion, then the rare patient who rebleeds with a clean-based
ulcer must have had stigmata of recent hemorrhage that were missed. The high interobserver variability for diagnosing stigmata
of recent hemorrhage supports this concept (Fig. 5) [12,13].
Endoscopic hemostasis is effective at stopping the bleeding in 8598% of patients initially, but treated patients may rebleed in up to 20% of casesunsurprisingly, since the endoscopist cannot determine the path of the bleeding vessel or whether the therapy obliterated
the bleeding point. Even the 'objective' method of checking Doppler is prone to both false positive (from other vessels) and negative (from missing the bleeding artery)
Prevention of rebleeding
Raising the intragastric pH improves coagulation by enhancing platelet aggregation and plasma coagulation factors, inhibiting
pepsin, and starting ulcer healing. While conceptually simple, raising and maintaining the intragastric pH > 4 is difficult clinically, short of using high-dose continuous intravenous proton pump inhibitor therapy. For example 80 mg of intravenous omeprazole bolus followed by continuous infusion at 8 mg/ h for 3 days significantly reduced the risk of rebleeding after endoscopic hemostasis from 22.5% to 6.7% in one study . The end-point of the study was rebleeding within 30 days; but most reduction in rebleeding occurred during the initial three
days of therapy. Although intravenous omeprazole is not available in the United States, intravenous pantoprazole, which has
a similar pharmacokinetic profile, could be substituted in the same doses. No convincing evidence supports the use of other
medical therapies such as somatostatin analogue, vasopressin, or tranexamic acid to reduce rebleeding after hemostasis. Similarly,
repeating endoscopy in 24 h and re-treating the bleeding lesion does not seem to reduce the rebleeding risk significantly
Treatment of rebleeding
Three choices for patients with rebleeding are  medical management (i.e. blood transfusion and waiting) , repeat endoscopic therapy, and  surgery. Although medical management is a reasonable alternative initially, most patients with ongoing bleeding require some
sort of treatment or risk dying. A randomized comparative trial showed that repeat endoscopic therapy was comparable to surgery,
but resulted in significantly fewer complications (15% vs. 36%) . Repeat epinephrine injection and heat probe therapy achieved permanent hemostasis in 73%, resulted in perforation in 4%
and failed in 23% of cases. Six out of seven patients in the repeat endoscopy group who developed a complication had salvage
surgery. These data clearly support repeating endoscopic hemostasis before operating in patients with rebleeding.
Approximately a third of the patients with major ulcer hemorrhage will experience recurrent ulcer bleeding within 3 years.
Three issues are key to this problem- H. Pylori, acid suppression [17,18], and NSAIDs.
It is now well established that eradication of H. pylori will virtually eliminate the risk of recurrent ulcer disease and bleeding in a patient with a 'pure'H. pylori ulcer . Therefore patients with H. pylori-caused bleeding peptic ulcers who are confirmed to have had the infection eradicated and are not using aspirin (including
low dose) or NSAIDs do not require maintenance acid suppression for prevention of delayed rebleeding.
NSAIDs and aspirin
The risk of delayed rebleeding of an ulcer from aspirin or NSAID treatment can be similarly lowered by stopping the offending
drug, whenever possible. Discontinuation or alternative therapy is especially important in patients with increased risk of
bleeding, such as previous bleeding of a complicated ulcer, older age, and use of multiple and higher doses of NSAIDs. Alternatively,
the dosage of the NSAID used can be reduced and/or prophylaxis administered in patients who cannot discontinue it. Newer agents appear to be safer in this regard.
Two large trials comparing celecoxib or rofecoxib to conventional NSAID therapy showed reduced ulcer complications, including
bleeding, only for patients not taking concomitant aspirin cardioprophylaxis [23,24]. What is unclear from these studies of primary prevention is whether proton pump inhibitor prophylaxis or COX-2 inhibitor
therapy will be similarly effective in reducing the risk of clinically significant rebleeding in patients with a previous
history of NSAID-induced ulcer bleeding, i.e. secondary prevention.
One of the few studies to address specifically the risk of delayed rebleeding in patients continuing to use NSAIDs randomized
patients with H. pylori infection and aspirin or naproxen use to antibiotic therapy or omeprazole 20 mg QD maintenance . H. pylori eradication before starting 80 mg of aspirin daily reduced the risk of delayed rebleeding similarly to omeprazole (probability of rebleeding of 1.9% vs.
0.9%, respectively), but was significantly less effective than omeprazole for patients taking naproxen 500 mg BID (probability of rebleeding of 18.8% vs. 4.4%). Thus long-term proton pump inhibitor prophylaxis appears to be indicated
only in patients infected with H. pylori who are taking full-dose NSAID therapy.
Two large studies showed that maintenance therapy using misoprostol 200 µg QID or omeprazole 20 mg QD was significantly better than placebo in reducing the risk of recurrent gastric ulceration (recurrence of 10% vs. 13%
vs. 32%, respectively), but only omeprazole was superior to placebo in reducing the recurrence of duodenal ulcers (recurrence
of 3% for omeprazole, 12% for control, and 10% for misoprostol) . Another study showed that omeprazole 20 mg QD was superior to ranitidine 150 mg QD for preventing gastric and duodenal ulcers (5.2% vs. 16.3%% and 0.5% vs. 5.7%, respectively) . Although it might be tempting to assume that a reduction in endoscopically diagnosed ulcers as shown in these studies would
also reduce clinical symptoms such as bleeding, no data exist to corroborate this notion. Thus at this time it remains unproven
whether long-term proton pump inhibitor therapy actually reduces the risk of delayed ulcer rebleeding.
Misoprostol 200 µg QID has been shown to reduce significantly ulcer complications, including bleeding . Unfortunately, 25% of the patients dropped out because of side-effects of the misoprostol. Moreover, the clinical relevance
of the 'statistically significant' (odds ratio, 0.598; p = 0.049) result remains to be determined, as misoprostol prophylaxis only prevented 17 more complications (out of 4404 randomized
patients) compared to placebo.
Variceal bleeding is discussed elsewhere. Bleeding from erosions is rarely a serious problem, except in the case of Dieulafoy
lesions. These essentially are small erosions with an exposed vessel. These lesions are sometimes missed by inexperienced
endoscopists, because they are small, and often high in the stomach. They are often best seen and treated in retroversion.
Whilst all types of endoscopic hemostasis can be used, clipping seems to be particularly effective in these cases.
Patients with MalloryWeiss tears often give a history of repeated retching or vomiting. Although the original description was of massive hemorrhage,
this is very unusual. When treatment is necessary, endoscopists should be mindful of the risk of perforating the damaged esophagus.
Clipping is now favored over injection or thermal methods.
Issues and future trends
There are several areas where further developments and study would be beneficial.
New suturing devices, clips, and bands
Two different suturing devices have been approved by the FDA for use in anti-reflux therapy. Their use for bleeding therapy
is limited by the inaccuracy of suture placement, superficial depth of suture, poor visibility and high costs. Banding is
ineffective against most ulcer bleeding, because the fibrotic base cannot be sucked up for banding. Two commercially available
clipping devices are better at stopping ulcer bleeding but still cumbersome to use, and have variable depth of penetration.
Several devices which could be used to perform full-thickness suturing should be available in the near future.
Visualization (scope size)
Which endoscope to use in this context remains a dilemma. Large channels facilitate lavage, but the necessary increased scope
size may be problematic. Scopes with greater than the standard channel size (10 FG) have been proposed.
Propofol and MAC (monitored anesthesia care) will gain wider acceptance for endoscopy in the near future, which should make
procedures easier for patients and physicians. Airway protection will continue to be needed for patients at risk for aspiration;
the development of better methods for clearing the stomach could decrease the need for intubation.
Increasing use of models and computers for training should improve endoscopic expertise. Implementation of a feedback system
(for example Doppler after hemostasis) could be an effective training aid.
1 Peterson, WL, Barnett, CC & Smith, HJ et al. Routine early endoscopy in upper-gastrointestinal-tract bleeding: a randomized, controlled trial. N Engl J Med 1981; 304: 9259. PubMed
2 Sacks, HS, Chalmers, TC & Blum, AL et al. Endoscopic hemostasis. An effective therapy for bleeding peptic ulcers. JAMA 1990; 264: 4949. PubMed
3 Cook, DJ, Guyatt, GH & Salena, BJ et al. Endoscopic therapy for acute non variceal upper gastrointestinal hemorrhage: a meta-analysis. Gastroenterology 1992; 102: 13948. PubMed
4 Lee, JG, Turnipseed, S & Romano, PS et al. Endoscopy-based triage significantly reduces hospitalization rates and costs of treating upper GI bleeding: a randomized controlled
trial. Gastrointest Endosc 1999; 50: 75561. PubMed
5 Stollman, NH, Putcha, RV & Neustater, BR et al. The uncleared fundal pool in acute upper gastrointestinal bleeding: implications and outcomes. Gastrointest Endosc 1997; 46: 3247. PubMed
6 Laine, L & Peterson, WL. Bleeding peptic ulcer. N Engl J Med 1994; 331: 71727. PubMed
7 Jensen, DM. Spots and clots leave them or treat them? Why and how to treat. Can J Gastroenterol 1999; 13: 4135. PubMed
8 Chung, SC, Lau, JY & Sung, JJ et al. Randomised comparison between adrenaline injection alone and adrenaline injection plus heat probe treatment for actively bleeding
ulcers. BMJ 1997; 314: 130711. PubMed
9 Cipolletta, L, Bianco, MA & Marmo, R et al. Endoclips versus heater probe in preventing early recurrent bleeding from peptic ulcer: a prospective and randomized trial. Gastrointest Endosc 2001; 53: 14751. PubMed
10 Lau, JY, Sung, JJ & Lee, KK et al. Effect of intravenous omeprazole on recurrent bleeding after endoscopic treatment of bleeding peptic ulcers. N Engl J Med 2000; 343: 3106. PubMed
11 Swain, CP, Storey, DW & Bown, SG et al. Nature of the bleeding vessel in recurrently bleeding gastric ulcers. Gastroenterology 1986; 90: 595608. PubMed
12 Lau, JY, Sung, JJ & Chan, AC et al. Stigmata of hemorrhage in bleeding peptic ulcers: an interobserver agreement study among international experts. Gastrointest Endosc 1997; 46: 336. PubMed
13 Laine, L, Freeman, M & Cohen, H. Lack of uniformity in evaluation of endoscopic prognostic features of bleeding ulcers. Gastrointest Endosc 1994; 40: 4117. PubMed
14 Wong, RC, Chak, A & Kobayashi, K et al. Role of Doppler US in acute peptic ulcer hemorrhage: can it predict failure of endoscopic therapy? Gastrointest Endosc 2000; 52: 31521. PubMed
15 Messmann, H, Schaller, P & Andus, T et al. Effect of programmed endoscopic follow-up examinations on the rebleeding rate of gastric or duodenal peptic ulcers treated
by injection therapy: a prospective, randomized controlled trial. Endoscopy 1998; 30: 5839. PubMed
16 Lau, JY, Sung, JJ & Lam, YH et al. Endoscopic retreatment compared with surgery in patients with recurrent bleeding after initial endoscopic control of bleeding
ulcers. N Engl J Med 1999; 340: 7516. PubMed
17 Jensen, DM, Cheng, S & Kovacs, TO et al. A controlled study of ranitidine for the prevention of recurrent hemorrhage from duodenal ulcer. N Engl J Med 1994; 330: 3826. PubMed
18 Garcia Rodriguez, LA & Ruigomez, A. Secondary prevention of upper gastrointestinal bleeding associated with maintenance acid-suppressing treatment in patients
with peptic ulcer bleed. Epidemiology 1999; 10: 22832. PubMed
19 Macri, G, Milani, S & Surrenti, E et al. Eradication of Helicobacter pylori reduces the rate of duodenal ulcer rebleeding: a long-term follow-up study. Am J Gastroenterol 1998; 93: 9257. PubMed
20 Hawkey, CJ, Karrasch, JA & Szczepanski, L et al. Omeprazole compared with misoprostol for ulcers associated with nonsteroidal antiinflammatory drugs. Omeprazole versus Misoprostol
for NSAID-induced Ulcer Management (OMNIUM) Study Group. N Engl J Med 1998; 338: 72734. PubMed
21 Yeomans, ND, Tulassay, Z & Juhasz, L et al. A comparison of omeprazole with ranitidine for ulcers associated with nonsteroidal antiinflammatory drugs. Acid Suppression
Trial: Ranitidine versus Omeprazole for NSAID-associated Ulcer Treatment (ASTRONAUT) Study Group. N Engl J Med 1998; 338: 71926. PubMed
22 Silverstein, FE, Graham, DY & Senior, JR et al. Misoprostol reduces serious gastrointestinal complications in patients with rheumatoid arthritis receiving nonsteroidal anti-inflammatory
drugs. Ann Intern Med 1995; 123: 2419. PubMed
23 Bombardier, C, Laine, L & Reicin, A et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. VIGOR Study
Group. N Engl J Med 2000; 343: 15208. PubMed
24 Silverstein, FE, Faich, G & Goldstein, JL et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis:
the CLASS study: a randomized controlled trial. Celecoxib Long-Term Arthritis Safety Study JAMA 2000; 284: 124755. PubMed
25 Chan, FK, Chung, SC & Suen, BY et al. Preventing recurrent upper gastrointestinal bleeding in patients with Helicobacter pylori infection who are taking low-dose aspirin or naproxen. N Engl J Med 2001; 344: 96773. PubMed
Copyright © Blackwell Publishing, 2003