274
Conferences and Reviews Upper Gastrointestinal Tract Hemorrhage Discussant
LOREN LAINE, MD
This discussion was selectedfrom the weekly staff conferences in the Department of Medicine, University of California, San Francisco. Taken from a transcription, it has been edited by Nathan M. Bass, MD, PhD, Associate Professor ofMedicine, under the direction ofLloyd H. Smith Jr, MD, Professor of Medicine and Associate Dean in the School ofMedicine.
FLOYD C. RECrOR, MD*: Acute upper gastrointestinal tract hemorrhage is one of the commonest medical emergencies encountered in clinical practice. Although considerable progress has been made in endoscopic diagnoses and treatment, hemodynamic assessment and support remain the mainstays of successful therapy. As an investigator active in the development and evaluation oftherapeutic approaches to the problem of upper gastrointestinal tract hemorrhage, Loren Laine, MD, of the University of Southern California (USC) School ofMedicine, brings a wealth ofexperience and insight to his review of this important topic.
than one who comes in with a hematocrit of 0. 15 and normal vital signs.
LOREN LAINE, MDt: Upper gastrointestinal (GI) tract hemorrhage remains a major health problem in the United States today, necessitating more than 300,000 hospital admissions annually. In this review, I provide an update on the approach to the management of upper GI tract hemorrhage from the initial assessment to diagnosis and treatment.
* Patients presenting with upper GI tract bleeding today are older than they were 50 years ago. Only a small percentage ofthe patients presenting with bleeding 40 to 50 years ago were older than 60 years; now perhaps half of these patients are older than 60.2 With this increase in age comes an increase in the number of concurrent illnesses and, as a result, a higher mortality. Patients with upper GI tract hemorrhage usually do not die of exsanguination but rather of worsening of their underlying illness during or after a bleed-
Initial Assessment The most important initial step in a patient presenting with upper GI tract hemorrhage is assessing the hemodynamic state-blood pressure, heart rate, 'and postural changes. Palpating the distal extremities can be likened to a "poor person's Swan-Ganz catheter"-the extremities of patients with major upper GI tract hemorrhage will be cool because of the peripheral vasoconstriction that occurs with hypovolemia. Too often physicians make the mistake of relying primarily on a patient's hematocrit value while paying little attention to the vital signs. The hematocrit is a poor indicator of the severity or rapidity of an acute bleeding episode. People bleed whole blood. Thus, if a patient's hematocrit is 0.45 (45%), and a third of the blood volume is lost in five minutes, the hematocrit will still be about 0.45. Studies in which healthy volunteers had 15% to 20% of their blood volume removed in about 10 minutes show that the hematocrit takes as long as 72 hours to equilibrate. ' Patients with an acute, severe bleeding episode typically get to a hospital quickly, and their hematocrit often has not yet fallen greatly. A patient presenting with a systolic blood pressure of 80 mm of mercury and a hematocrit of 0.40 is of far greater concern *Professor and Chair, Department of Medicine, University of California, San Francisco, School of Medicine.
tAssociate Professor, Department of Medicine, Division of Gastrointestinal and Liver Diseases, University of Southern California School of Medicine, Los Angeles.
Prognosis About 80% of patients with upper GI tract hemorrhage stop bleeding spontaneously and generally have an uneventful recovery with relatively low morbidity and mortality. Mortality in patients with this disorder is said to be about 10%. Remarkably, this has remained relatively constant over the past 30 to 50 years. Following are some of the reasons the outcome in these patients has not improved:
ing episode. * Most patients with upper GI tract hemorrhage do well without specific therapy or intervention. Therefore, showing a significant improvement is difficult. Successful treatment in high-risk patients is the key to improving outcome, but any improvement in this small group is diluted when all patients with upper GI tract bleeding are considered together. * Most important, no effective nonsurgical treatment has been available in the past. Without such therapy, the outcome in these patients was unlikely to improve. Various clinical factors provide prognostic information about the course of a patient presenting with upper GI tract hemorrhage.3 Patients whose bleeding begins while they are already in hospital for another medical problem-cardiac disease or cancer, for instance-have about a fourfold increase in mortality.3 4 As mentioned earlier, the combination of upper GI tract hemorrhage and other serious illness is associated with an increase in mortality. Patients presenting with both hematemesis of red blood and hematochezia and patients with hypotension have a mortality higher than those with other clinical presentations of bleeding. Patients with persistent or recurrent bleeding and those with ongoing transfusion requirements-an indicator of persistent or
(Laine L: Upper gastrointestinal tract hemorrhage. West J Med 1991 Sep; 155:274-279) Reprint requests to Loren Laine, MD, Department of Medicine, Division of Gastrointestinal and Liver Diseases, USC School of Medicine, 2025 Zonal Ave, Los Angeles, CA 90033.
THE WESTERN JOURNAL OF MEDICINE THl
SEPTEMBER 1991
e
o
155
9
ABBREVIATIONS USED IN TEXT GI = gastrointestinal NIH = National Institutes of Health NSAID = nonsteroidal anti-inflammatory drug USC = University of Southern California
recurrent
bleeding-also have
a
higher morbidity and
mortality.
Upper Versus Lower Gastrointestinal Tract Hemorrhage We must first determine if the source of gastrointestinal bleeding is in the upper or lower GI tract. Vomiting of red blood or coffee-ground emesis (hematemesis) clearly indicates an upper GI tract source. Melena (black, tarry stool) usually suggests an upper GI tract hemorrhage, and hematochezia (bright red blood through the rectum) is usually from a lower GI tract site. The color does not relate directly to the location of the bleeding site, however; rather, it relates to the length of time the blood remains in the GI tract. Blood from rapidly bleeding lesions in the esophagus or stomach may move through the gut quickly and still be bright red on egress from the rectum. Studies in which blood was instilled at various levels ofthe GI tracts of volunteers show that hematochezia may result when blood has remained in the gut for eight hours or less. 5 On the other hand, blood must remain in the GI tract for more than 14 hours before melena will develop.5 Thus, although unusual, hemorrhage from the proximal colon can present as melena ifthe transit time is slow. If a patient has hematochezia and hemodynamic instability, do not automatically assume that the patient is bleeding from the lower GI tract. Patients with hematochezia and hemodynamic evidence of major bleeding should have an upper endoscopy before the more arduous evaluation of the lower GI tract is undertaken. Aspiration through a nasogastric tube also has frequently unrecognized pitfalls in establishing the site of gastrointestinal bleeding. Assuming there is no contamination from outside the GI tract such as epistaxis or hemoptysis, the nasogastric aspirate is clearly diagnostic when gross blood or "coffee grounds" are present. Testing the aspirate for occult blood is not of value in a patient with acute bleeding. A normal nasogastric aspirate, however, does not rule out an upper GI tract source of bleeding. As many as 15% of patients with active upper GI tract bleeding have a normal nasogastric aspirate.6 This probably is seen most commonly with duodenal ulcers. Physicians often suggest that a nasogastric aspirate which is not bloody and appears bilious absolutely rules out a source of bleeding proximal to the second portion of the duodenum. Unfortunately, a physician's impression about the presence of bile in the nasogastric aspirate will be correct in only about 50% of cases.' Auscultating the bowel sounds also may help distinguish between an upper and lower GI tract hemorrhage. Gut motility is increased by intraluminal blood, and this leads to an increase in bowel sounds in patients with bleeding from the GI tract. Blood urea nitrogen levels are classically elevated in patients with upper GI tract bleeding. This increase may reflect prerenal azotemia due to hypovolemia or it may relate to the absorption of a large amount of protein from blood in the small intestine. Instilling blood into the upper small intestine upper
3
275
of volunteers can lead to a rise in blood urea nitrogen levels, but no rise is seen when blood is placed in the colon.8
Empiric Medical Therapy A variety of therapies are given to patients with upper GI tract hemorrhage, but none have proved to be effective in stopping bleeding or preventing recurrent bleeding. Older textbooks report that lavage stops bleeding in 75% of cases. But bleeding stops spontaneously 75% to 80% ofthe time, and no good evidence exists that lavage plays a role in stopping upper GI tract hemorrhage. Iced saline solution is usually recommended for lavage. Studies in dogs with experimentally induced ulcers indicate, however, that results with lavage are no better and may even be worse at temperatures of 0°C to 4°C.9 Nasogastric lavage does have a role in these patients: it can document an upper GI tract source of bleeding, and it allows better visualization of the upper GI tract at endoscopy by removing blood and clots. Patients with bleeding from the upper GI tract frequently receive histamine H2-receptor antagonists, antacids, or both. Again, numerous studies have failed to document a clear benefit in stopping bleeding or preventing recurrent bleeding. A meta-analysis of studies evaluating the use of cimetidine in patients with this disorder did suggest a possible slight benefit of using H2-receptor antagonists in the subset of patients with bleeding gastric ulcers. A clinical study aiming to document this improvement would require 10,000 or more patients. 10 Vasopressin should not be used empirically in patients with gastrointestinal hemorrhage. Its use in those with variceal bleeding will be discussed later. The use of intravenous somatostatin-which is not available in the United States has shown mixed results.11 Promising data have been generated with the use of tranexamic acid, an antifibrinolytic agent. 12
Diagnosis The two primary methods of diagnosis for upper GI tract hemorrhage are radiography (upper GI series) and endoscopy. The diagnostic yield of upper endoscopy is superior to that of barium studies; endoscopy should identify the source of bleeding in at least 90% of cases. We like to think that identifying the cause of a problem leads to improved management and outcome. Seven controlled trials, however, of early endoscopy in patients with upper GI tract bleeding, all done more than a decade ago, failed to show a benefit of urgent endoscopy in important outcome variables such as surgical intervention, length of hospital stay, and mortality."3 Thus, the excellent diagnostic yield of endoscopy was not translated into a clinical benefit for the patients. The reason for this may be that these studies were done before gastroenterologists took advantage of the prognostic and therapeutic usefulness of endoscopy. We are unlikely to alter the outcome unless the diagnostic information obtained is used when making patient-management decisions (for example, when to discharge the patient or when to administer endoscopic hemostatic therapy). Contrast radiographic studies of the upper GI tract have several disadvantages compared with endoscopy. The upper GI series is less accurate than endoscopy; lesions missed include gastric ulcers, Mallory-Weiss tears, erosive and hemorrhagic "gastritis," and varices. In addition, contrast studies do not show active bleeding or stigmata of hemorrhage,
276
UPPER GASTROINTESTINAL HEMORRHAGE
TABLE 1.-Sites of Bleding in 445 Consecutive Patients Pr nting Wth Ma or Wer Gastintstinal Tat Hemorhge at Los Angeles ICuntyUniversity of Southern Caifonia Medical Cnter Patients,
(1O
Site Of BleedingN
Duodenal ulcer0 (7...................._.,I.,0...)
Gastric ulcer. Varices
.............................
Mallory-Weiss Erosive
or
tear
...........
57
(13)
137
(31)
........(. t 13)
hemorrhagic gastritis. "
14
Argon laser Nd:YAG laser Electrocoagulation Monopolar
Sclerosing agents, such as polidocanol Epinephrine .: uto
Bipolar or multipolar
Saline solution
Heater probe Nd:YAG = neodymium yttrium aluminum garnet
(
8 2): esophagitis Vascular malformations.5 (rmaltormatios 1)
infusion through the left gastric artery also can be effective. Surgical therapy is rarely needed to oversew a bleeding Mallory-Weiss tear.
Other
Peptic Ulcers Peptic ulcers are the most common cause of serious upper GI tract hemorrhage. The appearance of the ulcer at endoscopy provides important prognostic information. Ulcers may have a clean base, they may have stigmata of recent hemorrhage (flat pigmented spot, adherent clot, or nonbleeding visible vessel), or they may be actively bleeding. Ulcers with a clean base have a 1% incidence of recurrent bleeding, similar to that of a nonbleeding Mallory-Weiss tear. Flat pigmented spots portend rebleeding in 7% to 8% of cases, and patients whose ulcers have adherent clots rebleed about 20% of the time. Bleeding recurs in 40% of patients with nonbleeding visible vessels, and about a third will require urgent surgical therapy for bleeding if not treated endoscopically. Patients with active bleeding identified at endoscopy have the worst outcome-more than half may require urgent intervention. Patient management can therefore be guided by the features of the ulcer identified at endoscopy. I think that patients with nonbleeding Mallory-Weiss tears or clean-based ulcers can be discharged home after being stabilized, thereby decreasing hospital costs and justifying the expense of early endoscopy. Patients with flat pigmented spots or adherent clots probably should be observed in the hospital, but no specific intervention is required. On the other hand, patients with nonbleeding visible vessels or active bleeding are at high risk for further bleeding and often require surgical intervention. This is the subset of patients in whom effective alternatives to emergency surgical treatment are needed to decrease the morbidity and mortality associated with upper GI tract hemorrhage. The recent introduction of endoscopic hemostatic therapy appears to have provided an effective nonsurgical therapy for patients with active bleeding or nonbleeding visible vessels. Endoscopic Hemostatic Therapy Those diseases which medicines do not cure the knife cures Those which the knife cannot cure
Erosive
(.
St
Prolapse gastropathy 3 1) Gastric cancer.3 (1) ..........
Not identified
.........
.
.
5( 1) 33
7)
:Sytoic blood pressure 90:mm of mercury or less, heart rate 1 10 per minute or more, or orthostatic change in systolic blood pressure of 20 mm of mercury or more or in heart rate of 20 pe minute; transfusion of 2 units in 12 hours; or drop in hematocrit of 6'b in 12 hours.
so
TABLE 2.-Efndosopic Hemostatk Therapy ThermalM Injection Therapy l e:0 ULser photocoagulation Absolute ehanol
lesions identified on upper GI series can only be assumed
to be the source of bleeding. Finally, the use of barium contrast is contraindicated in patients with upper GI tract hemor-
rhage because it may interfere with subsequent studies such angiography or endoscopy. The major disadvantage of upper endoscopy is its cost. Endoscopy is generally three to four times more expensive than an upper GI series. If endoscopy does not improve the outcome in patients with upper GI tract bleeding, then we are not justified in doing endoscopy in them. Recent studies indicate, however, that early endoscopy can improve patient care and decrease costs. Endoscopic findings can be used prognostically to help decide on the level of patient care and time of discharge, and endoscopic hemostatic therapy is beneficial in patients at high risk for persistent or recurrent bleeding. Causes and Treatment of Upper Gastrointestinal Tract Hemorrhage Table 1 lists the sources of bleeding in 445 consecutive patients presenting to Los Angeles County-USC Medical Center with clinically severe upper GI tract hemorrhage. About 40% of patients bled from ulcers (with duodenal ulcers twice as common as gastric ulcers), and 30% bled from esophagogastric varices. Series from hospitals with fewer patients with alcoholic liver disease generally show a lower prevalence of variceal bleeding (about 10%) and a greater number of bleeding ulcers (usually 50% or more). Although many series list erosive and hemorrhagic "gastritis" as a common cause of bleeding, in our experience, this is an uncommon cause of clinically severe upper GI tract bleeding. as
Mallory-Weiss Tears Mallory-Weiss tears, seen most commonly in patients with alcoholism, are shallow lacerations that are usually on the gastric side of the gastroesophageal junction. Most Mallory-Weiss tears stop bleeding spontaneously, and hemorrhage rarely recurs. Generally no intervention is required, and supportive therapy is sufficient. A patient with persistent bleeding from a Mallory-Weiss tear can be treated successfully with endoscopic therapy. If this is unsuccessful or unavailable, angiographic therapy with embolization or
vasopressin
cautery cures
And those which cautery cannot cure are reckoned to be wholly incurable. HIPPOCRATES [5TH CENTURY BC]
Thermal methods. Various thermal methods for endoscopic hemostasis are available (Table 2). Laser photocoagulation was among the first methods studied in patients with upper GI tract bleeding. Although effective in some clinical trials, laser therapy is no longer generally used to treat this
THE WESTERN JOURNAL OF MEDICINE
*
SEPTEMBER 1991
155
*
*
TAB 3.Controlled Tial of AMuPolarEleto* ulation in Patients With Aciv Upper Gatnteti Vanabl
Shm Coauatn T--herpy, n=23
n21
H.mostasis.% . 0 13 T,-,rfusions units.-24 5 Emerenc inevention, qb 7 14 4.4 HOsp*itasd .............. 7.2 3.420 Hospital ost $.7.550 0 13 Mortality, % ..
t t t t
41 3.0 Emergency operation. %.30 6.2 Hospital stay d.
18 1.6 8 4.3
Hospital ct. $.5,730
3,790
Mortality, % ............... ¶rom Lane18
0
3
tP
Conferences and Reviews Upper Gastrointestinal Tract Hemorrhage Discussant
LOREN LAINE, MD
This discussion was selectedfrom the weekly staff conferences in the Department of Medicine, University of California, San Francisco. Taken from a transcription, it has been edited by Nathan M. Bass, MD, PhD, Associate Professor ofMedicine, under the direction ofLloyd H. Smith Jr, MD, Professor of Medicine and Associate Dean in the School ofMedicine.
FLOYD C. RECrOR, MD*: Acute upper gastrointestinal tract hemorrhage is one of the commonest medical emergencies encountered in clinical practice. Although considerable progress has been made in endoscopic diagnoses and treatment, hemodynamic assessment and support remain the mainstays of successful therapy. As an investigator active in the development and evaluation oftherapeutic approaches to the problem of upper gastrointestinal tract hemorrhage, Loren Laine, MD, of the University of Southern California (USC) School ofMedicine, brings a wealth ofexperience and insight to his review of this important topic.
than one who comes in with a hematocrit of 0. 15 and normal vital signs.
LOREN LAINE, MDt: Upper gastrointestinal (GI) tract hemorrhage remains a major health problem in the United States today, necessitating more than 300,000 hospital admissions annually. In this review, I provide an update on the approach to the management of upper GI tract hemorrhage from the initial assessment to diagnosis and treatment.
* Patients presenting with upper GI tract bleeding today are older than they were 50 years ago. Only a small percentage ofthe patients presenting with bleeding 40 to 50 years ago were older than 60 years; now perhaps half of these patients are older than 60.2 With this increase in age comes an increase in the number of concurrent illnesses and, as a result, a higher mortality. Patients with upper GI tract hemorrhage usually do not die of exsanguination but rather of worsening of their underlying illness during or after a bleed-
Initial Assessment The most important initial step in a patient presenting with upper GI tract hemorrhage is assessing the hemodynamic state-blood pressure, heart rate, 'and postural changes. Palpating the distal extremities can be likened to a "poor person's Swan-Ganz catheter"-the extremities of patients with major upper GI tract hemorrhage will be cool because of the peripheral vasoconstriction that occurs with hypovolemia. Too often physicians make the mistake of relying primarily on a patient's hematocrit value while paying little attention to the vital signs. The hematocrit is a poor indicator of the severity or rapidity of an acute bleeding episode. People bleed whole blood. Thus, if a patient's hematocrit is 0.45 (45%), and a third of the blood volume is lost in five minutes, the hematocrit will still be about 0.45. Studies in which healthy volunteers had 15% to 20% of their blood volume removed in about 10 minutes show that the hematocrit takes as long as 72 hours to equilibrate. ' Patients with an acute, severe bleeding episode typically get to a hospital quickly, and their hematocrit often has not yet fallen greatly. A patient presenting with a systolic blood pressure of 80 mm of mercury and a hematocrit of 0.40 is of far greater concern *Professor and Chair, Department of Medicine, University of California, San Francisco, School of Medicine.
tAssociate Professor, Department of Medicine, Division of Gastrointestinal and Liver Diseases, University of Southern California School of Medicine, Los Angeles.
Prognosis About 80% of patients with upper GI tract hemorrhage stop bleeding spontaneously and generally have an uneventful recovery with relatively low morbidity and mortality. Mortality in patients with this disorder is said to be about 10%. Remarkably, this has remained relatively constant over the past 30 to 50 years. Following are some of the reasons the outcome in these patients has not improved:
ing episode. * Most patients with upper GI tract hemorrhage do well without specific therapy or intervention. Therefore, showing a significant improvement is difficult. Successful treatment in high-risk patients is the key to improving outcome, but any improvement in this small group is diluted when all patients with upper GI tract bleeding are considered together. * Most important, no effective nonsurgical treatment has been available in the past. Without such therapy, the outcome in these patients was unlikely to improve. Various clinical factors provide prognostic information about the course of a patient presenting with upper GI tract hemorrhage.3 Patients whose bleeding begins while they are already in hospital for another medical problem-cardiac disease or cancer, for instance-have about a fourfold increase in mortality.3 4 As mentioned earlier, the combination of upper GI tract hemorrhage and other serious illness is associated with an increase in mortality. Patients presenting with both hematemesis of red blood and hematochezia and patients with hypotension have a mortality higher than those with other clinical presentations of bleeding. Patients with persistent or recurrent bleeding and those with ongoing transfusion requirements-an indicator of persistent or
(Laine L: Upper gastrointestinal tract hemorrhage. West J Med 1991 Sep; 155:274-279) Reprint requests to Loren Laine, MD, Department of Medicine, Division of Gastrointestinal and Liver Diseases, USC School of Medicine, 2025 Zonal Ave, Los Angeles, CA 90033.
THE WESTERN JOURNAL OF MEDICINE THl
SEPTEMBER 1991
e
o
155
9
ABBREVIATIONS USED IN TEXT GI = gastrointestinal NIH = National Institutes of Health NSAID = nonsteroidal anti-inflammatory drug USC = University of Southern California
recurrent
bleeding-also have
a
higher morbidity and
mortality.
Upper Versus Lower Gastrointestinal Tract Hemorrhage We must first determine if the source of gastrointestinal bleeding is in the upper or lower GI tract. Vomiting of red blood or coffee-ground emesis (hematemesis) clearly indicates an upper GI tract source. Melena (black, tarry stool) usually suggests an upper GI tract hemorrhage, and hematochezia (bright red blood through the rectum) is usually from a lower GI tract site. The color does not relate directly to the location of the bleeding site, however; rather, it relates to the length of time the blood remains in the GI tract. Blood from rapidly bleeding lesions in the esophagus or stomach may move through the gut quickly and still be bright red on egress from the rectum. Studies in which blood was instilled at various levels ofthe GI tracts of volunteers show that hematochezia may result when blood has remained in the gut for eight hours or less. 5 On the other hand, blood must remain in the GI tract for more than 14 hours before melena will develop.5 Thus, although unusual, hemorrhage from the proximal colon can present as melena ifthe transit time is slow. If a patient has hematochezia and hemodynamic instability, do not automatically assume that the patient is bleeding from the lower GI tract. Patients with hematochezia and hemodynamic evidence of major bleeding should have an upper endoscopy before the more arduous evaluation of the lower GI tract is undertaken. Aspiration through a nasogastric tube also has frequently unrecognized pitfalls in establishing the site of gastrointestinal bleeding. Assuming there is no contamination from outside the GI tract such as epistaxis or hemoptysis, the nasogastric aspirate is clearly diagnostic when gross blood or "coffee grounds" are present. Testing the aspirate for occult blood is not of value in a patient with acute bleeding. A normal nasogastric aspirate, however, does not rule out an upper GI tract source of bleeding. As many as 15% of patients with active upper GI tract bleeding have a normal nasogastric aspirate.6 This probably is seen most commonly with duodenal ulcers. Physicians often suggest that a nasogastric aspirate which is not bloody and appears bilious absolutely rules out a source of bleeding proximal to the second portion of the duodenum. Unfortunately, a physician's impression about the presence of bile in the nasogastric aspirate will be correct in only about 50% of cases.' Auscultating the bowel sounds also may help distinguish between an upper and lower GI tract hemorrhage. Gut motility is increased by intraluminal blood, and this leads to an increase in bowel sounds in patients with bleeding from the GI tract. Blood urea nitrogen levels are classically elevated in patients with upper GI tract bleeding. This increase may reflect prerenal azotemia due to hypovolemia or it may relate to the absorption of a large amount of protein from blood in the small intestine. Instilling blood into the upper small intestine upper
3
275
of volunteers can lead to a rise in blood urea nitrogen levels, but no rise is seen when blood is placed in the colon.8
Empiric Medical Therapy A variety of therapies are given to patients with upper GI tract hemorrhage, but none have proved to be effective in stopping bleeding or preventing recurrent bleeding. Older textbooks report that lavage stops bleeding in 75% of cases. But bleeding stops spontaneously 75% to 80% ofthe time, and no good evidence exists that lavage plays a role in stopping upper GI tract hemorrhage. Iced saline solution is usually recommended for lavage. Studies in dogs with experimentally induced ulcers indicate, however, that results with lavage are no better and may even be worse at temperatures of 0°C to 4°C.9 Nasogastric lavage does have a role in these patients: it can document an upper GI tract source of bleeding, and it allows better visualization of the upper GI tract at endoscopy by removing blood and clots. Patients with bleeding from the upper GI tract frequently receive histamine H2-receptor antagonists, antacids, or both. Again, numerous studies have failed to document a clear benefit in stopping bleeding or preventing recurrent bleeding. A meta-analysis of studies evaluating the use of cimetidine in patients with this disorder did suggest a possible slight benefit of using H2-receptor antagonists in the subset of patients with bleeding gastric ulcers. A clinical study aiming to document this improvement would require 10,000 or more patients. 10 Vasopressin should not be used empirically in patients with gastrointestinal hemorrhage. Its use in those with variceal bleeding will be discussed later. The use of intravenous somatostatin-which is not available in the United States has shown mixed results.11 Promising data have been generated with the use of tranexamic acid, an antifibrinolytic agent. 12
Diagnosis The two primary methods of diagnosis for upper GI tract hemorrhage are radiography (upper GI series) and endoscopy. The diagnostic yield of upper endoscopy is superior to that of barium studies; endoscopy should identify the source of bleeding in at least 90% of cases. We like to think that identifying the cause of a problem leads to improved management and outcome. Seven controlled trials, however, of early endoscopy in patients with upper GI tract bleeding, all done more than a decade ago, failed to show a benefit of urgent endoscopy in important outcome variables such as surgical intervention, length of hospital stay, and mortality."3 Thus, the excellent diagnostic yield of endoscopy was not translated into a clinical benefit for the patients. The reason for this may be that these studies were done before gastroenterologists took advantage of the prognostic and therapeutic usefulness of endoscopy. We are unlikely to alter the outcome unless the diagnostic information obtained is used when making patient-management decisions (for example, when to discharge the patient or when to administer endoscopic hemostatic therapy). Contrast radiographic studies of the upper GI tract have several disadvantages compared with endoscopy. The upper GI series is less accurate than endoscopy; lesions missed include gastric ulcers, Mallory-Weiss tears, erosive and hemorrhagic "gastritis," and varices. In addition, contrast studies do not show active bleeding or stigmata of hemorrhage,
276
UPPER GASTROINTESTINAL HEMORRHAGE
TABLE 1.-Sites of Bleding in 445 Consecutive Patients Pr nting Wth Ma or Wer Gastintstinal Tat Hemorhge at Los Angeles ICuntyUniversity of Southern Caifonia Medical Cnter Patients,
(1O
Site Of BleedingN
Duodenal ulcer0 (7...................._.,I.,0...)
Gastric ulcer. Varices
.............................
Mallory-Weiss Erosive
or
tear
...........
57
(13)
137
(31)
........(. t 13)
hemorrhagic gastritis. "
14
Argon laser Nd:YAG laser Electrocoagulation Monopolar
Sclerosing agents, such as polidocanol Epinephrine .: uto
Bipolar or multipolar
Saline solution
Heater probe Nd:YAG = neodymium yttrium aluminum garnet
(
8 2): esophagitis Vascular malformations.5 (rmaltormatios 1)
infusion through the left gastric artery also can be effective. Surgical therapy is rarely needed to oversew a bleeding Mallory-Weiss tear.
Other
Peptic Ulcers Peptic ulcers are the most common cause of serious upper GI tract hemorrhage. The appearance of the ulcer at endoscopy provides important prognostic information. Ulcers may have a clean base, they may have stigmata of recent hemorrhage (flat pigmented spot, adherent clot, or nonbleeding visible vessel), or they may be actively bleeding. Ulcers with a clean base have a 1% incidence of recurrent bleeding, similar to that of a nonbleeding Mallory-Weiss tear. Flat pigmented spots portend rebleeding in 7% to 8% of cases, and patients whose ulcers have adherent clots rebleed about 20% of the time. Bleeding recurs in 40% of patients with nonbleeding visible vessels, and about a third will require urgent surgical therapy for bleeding if not treated endoscopically. Patients with active bleeding identified at endoscopy have the worst outcome-more than half may require urgent intervention. Patient management can therefore be guided by the features of the ulcer identified at endoscopy. I think that patients with nonbleeding Mallory-Weiss tears or clean-based ulcers can be discharged home after being stabilized, thereby decreasing hospital costs and justifying the expense of early endoscopy. Patients with flat pigmented spots or adherent clots probably should be observed in the hospital, but no specific intervention is required. On the other hand, patients with nonbleeding visible vessels or active bleeding are at high risk for further bleeding and often require surgical intervention. This is the subset of patients in whom effective alternatives to emergency surgical treatment are needed to decrease the morbidity and mortality associated with upper GI tract hemorrhage. The recent introduction of endoscopic hemostatic therapy appears to have provided an effective nonsurgical therapy for patients with active bleeding or nonbleeding visible vessels. Endoscopic Hemostatic Therapy Those diseases which medicines do not cure the knife cures Those which the knife cannot cure
Erosive
(.
St
Prolapse gastropathy 3 1) Gastric cancer.3 (1) ..........
Not identified
.........
.
.
5( 1) 33
7)
:Sytoic blood pressure 90:mm of mercury or less, heart rate 1 10 per minute or more, or orthostatic change in systolic blood pressure of 20 mm of mercury or more or in heart rate of 20 pe minute; transfusion of 2 units in 12 hours; or drop in hematocrit of 6'b in 12 hours.
so
TABLE 2.-Efndosopic Hemostatk Therapy ThermalM Injection Therapy l e:0 ULser photocoagulation Absolute ehanol
lesions identified on upper GI series can only be assumed
to be the source of bleeding. Finally, the use of barium contrast is contraindicated in patients with upper GI tract hemor-
rhage because it may interfere with subsequent studies such angiography or endoscopy. The major disadvantage of upper endoscopy is its cost. Endoscopy is generally three to four times more expensive than an upper GI series. If endoscopy does not improve the outcome in patients with upper GI tract bleeding, then we are not justified in doing endoscopy in them. Recent studies indicate, however, that early endoscopy can improve patient care and decrease costs. Endoscopic findings can be used prognostically to help decide on the level of patient care and time of discharge, and endoscopic hemostatic therapy is beneficial in patients at high risk for persistent or recurrent bleeding. Causes and Treatment of Upper Gastrointestinal Tract Hemorrhage Table 1 lists the sources of bleeding in 445 consecutive patients presenting to Los Angeles County-USC Medical Center with clinically severe upper GI tract hemorrhage. About 40% of patients bled from ulcers (with duodenal ulcers twice as common as gastric ulcers), and 30% bled from esophagogastric varices. Series from hospitals with fewer patients with alcoholic liver disease generally show a lower prevalence of variceal bleeding (about 10%) and a greater number of bleeding ulcers (usually 50% or more). Although many series list erosive and hemorrhagic "gastritis" as a common cause of bleeding, in our experience, this is an uncommon cause of clinically severe upper GI tract bleeding. as
Mallory-Weiss Tears Mallory-Weiss tears, seen most commonly in patients with alcoholism, are shallow lacerations that are usually on the gastric side of the gastroesophageal junction. Most Mallory-Weiss tears stop bleeding spontaneously, and hemorrhage rarely recurs. Generally no intervention is required, and supportive therapy is sufficient. A patient with persistent bleeding from a Mallory-Weiss tear can be treated successfully with endoscopic therapy. If this is unsuccessful or unavailable, angiographic therapy with embolization or
vasopressin
cautery cures
And those which cautery cannot cure are reckoned to be wholly incurable. HIPPOCRATES [5TH CENTURY BC]
Thermal methods. Various thermal methods for endoscopic hemostasis are available (Table 2). Laser photocoagulation was among the first methods studied in patients with upper GI tract bleeding. Although effective in some clinical trials, laser therapy is no longer generally used to treat this
THE WESTERN JOURNAL OF MEDICINE
*
SEPTEMBER 1991
155
*
*
TAB 3.Controlled Tial of AMuPolarEleto* ulation in Patients With Aciv Upper Gatnteti Vanabl
Shm Coauatn T--herpy, n=23
n21
H.mostasis.% . 0 13 T,-,rfusions units.-24 5 Emerenc inevention, qb 7 14 4.4 HOsp*itasd .............. 7.2 3.420 Hospital ost $.7.550 0 13 Mortality, % ..
t t t t
41 3.0 Emergency operation. %.30 6.2 Hospital stay d.
18 1.6 8 4.3
Hospital ct. $.5,730
3,790
Mortality, % ............... ¶rom Lane18
0
3
tP
