Acute gastrointestinal (GI) bleeding is a common problem causing significant morbidity and mortality. The source of GI bleeding can be anywhere in the GI tract, from the esophagus to the rectum. GI bleeding is classified into upper or lower bleeding based on the site of bleeding relative to the ligament of Treitz. Upper GI hemorrhage occurs from sites proximal to the ligament of Treitz and accounts for more than 80% of acute bleeding.¹ Lower GI bleeding originates distal to the ligament of Treitz, most commonly from the colon. The small intestine is the site of bleeding in less than 5% of patients.¹ Hemorrhage persisting or recurring after negative endoscopy is termed obscure bleeding. Occasionally patients present with occult bleeding, where there are no signs of overt bleeding but only symptoms of chronic blood loss anemia. In all cases, thorough investigation to localize the source of bleeding allows rapid and often definitive management.

Incidence and Economic Impact of GI Bleeding

Acute GI hemorrhage is one of the most common problems prompting outpatient, emergency room (ER), and inpatient visits. In 2012, nearly 800,000 patients seen in the emergency department (ED) were discharged with a diagnosis of GI hemorrhage, or 254 visits per 100,000 adults.² Of those patients, 54.6% were admitted, and over 500,000 patients had GI bleeding as their principal diagnosis for admission in 2012, with an estimated cost of nearly $5 billion US. The incidence of acute upper GI bleeding is estimated at 170 cases per 100,000 adults, and increases with age, affecting 1% of those older than age 85, and is more frequent than lower GI bleeding.^3-6 There are geographical variations in the GI bleeding incidence, with reported rates varying from 45 per 100,000 in the Netherlands to 172 per 100,000 in Scotland. This difference is likely related to differences in population demographics and prevalence of various etiological factors between the countries.^7-11

Morbidity and Mortality

Despite advances in therapy, mortality from GI bleeding remains high. In hospital death from GI bleeding in 2012 occurred in 2.2% of cases, and is particularly high in the elderly.² Mortality from GI bleeding occurs frequently on presentation in the ED or early in hospitalization. Evidence suggests that upper GI bleed mortality rates have declined in the last 3 decades, with mortality rates based on the NHDS trending from 4.8% in 1979 to 1989 to 3.1% from 2000 to 2009, primarily due to reductions in early hospital mortality. ^{5-6, 9, 12-13}

A structured approach is recommended in the initial evaluation and management of the patient with acute GI bleeding (Fig. 17-1). Early resuscitation with the aim of restoring hemodynamic stability is the initial priority, followed by a careful history and physical examination to help identify the etiology and source of bleeding. Particular attention should be paid to comorbidities and the drug history as this may further complicate management. Diagnostic tests are subsequently performed to confirm the site of bleeding, and therapeutic interventions commenced to control active bleeding and prevent future recurrent hemorrhage.

Initial Assessment

Management of resuscitation should follow the principles of A (airway), B (breathing), and C (circulation). Once airway and breathing have been managed, adequate hemodynamic resuscitation is of the highest priority. In particular, the clinician needs to assess the amount of blood lost and the extent of ongoing bleeding. Initial evaluation should focus on rapid assessment of the magnitude of both the pre-existing deficits and of ongoing hemorrhage. This can be determined by history and examination of the presenting symptoms. In the majority of cases, a wealth of information can be obtained from simple clinical parameters such as consciousness level, blood pressure, and heart rate (Table 17-1), and further facilitated by measurement of urine output as a marker of end-organ perfusion. Not all patients will demonstrate a tachycardic response to bleeding, particularly in the elderly or those on β-blockers. Occasionally severe blood loss may cause vagal-mediated bradycardia. Depending on the hemodynamic status of the patient and existing comorbidities, more invasive forms of monitoring such as central venous monitoring may be required.

While initial blood tests including a complete blood count and a type and cross are important, a normal hematocrit in the early stages of bleeding may be falsely reassuring, as the hematocrit will only decrease following dilution of the blood volume with resuscitation.

Resuscitation

The importance of adequate resuscitation cannot be overemphasized. The most important contributor to morbidity and mortality in acute GI bleeding is fulminant multiorgan failure from inadequate resuscitation. The critical care team should be involved early in the resuscitation process, as early intubation and ventilation will reduce the complications of any respiratory compromise. Large-bore venous access is crucial, particularly in the hemodynamically unstable. Fluid resuscitation should be commenced with an isotonic crystalloid solution such as lactated Ringer’s. Adequacy of resuscitation should be continuously assessed using clinical parameters such as heart rate, blood pressure, and urine output. A central venous catheter will facilitate assessment of preload in those with cardiac, pulmonary, or renal comorbidities and thereby facilitate more sensitive assessment of fluid balance. Basic laboratory tests including a complete blood count, basic metabolic panel, liver function test, coagulation profile, and type and cross should be obtained.

Transfusion

Several factors need to be considered when deciding whether a blood transfusion is required. Of these, the most important are the presence and extent of ongoing bleeding and the response of the patient to fluid resuscitation. Other factors include the age of the patient and the presence of cardiopulmonary comorbidities that might compromise tissue perfusion. The suspected likelihood of rebleeding should also be taken into account; for instance, a transfusion is more likely to be required for esophageal varices, which have a high propensity for profuse rebleeding.

Packed red cells are the usual form of transfusion but are defective in clotting factors, calcium, and platelets. In patients with significant bleeding requiring massive transfusion (more than six units of blood), supplementation of red blood cell (RBC) transfusion with fresh frozen plasma, platelets, and calcium is important. In the trauma literature, use of a 1:1:1 ratio of RBC, plasma, and platelets to approximate whole blood in cases of massive blood loss has been associated with improved outcomes and may be applicable to similar situations due to massive GI hemorrhage.¹⁴

In cases of slower bleeding in the hemodynamically stable patient, hematocrit transfusion triggers have changed over time. There is literature from cardiac surgery and orthopedic surgery patients to suggest that a restrictive transfusion strategy may be linked to improved outcomes, primarily due to decreased infectious complications. A hematocrit less than 21 in the young and healthy patient is a threshold where transfusion should be considered, but in older patients with cardiac morbidity, transfusion is recommended at higher values. However, the decision to transfuse should ultimately be guided by the individual patient, taking into account factors such as the degree of ongoing bleeding, potential for recurrent bleeding, and assessment of tissue perfusion.

Risk Stratification

The development of risk stratification scores has facilitated prediction of mortality, risk of rebleeding, and triage for admission and timing of investigations. These scores help differentiate patients stable for outpatient investigation and those requiring admission and urgent endoscopy. The BLEED study identified ongoing bleeding, low blood pressure (systolic blood pressure less than 100 mm Hg), elevated prothrombin time (greater than 1.2 times control), erratic mental status, and unstable comorbid disease as risk factors for significantly higher rates of surgery, increased recurrent bleeding, and higher mortality.¹⁵ Other studies have identified hepatic cirrhosis, high Acute Physiologic and Chronic Health Evaluation II (APACHE II) scores, active GI bleeding, hypotension, and end-organ dysfunction as independent predictors for the above outcomes.¹⁶ These studies highlight the importance of comorbidities in determining the outcome of GI bleeding. For example, one study found a mortality rate of nearly 30% in patients with significant renal disease and 65% in patients with acute renal failure.¹⁷

A thorough history and examination can assist in diagnosing the cause of bleeding and identify comorbidities likely to influence outcome.

Important Characteristics of GI Bleeding

Time of onset, volume, and frequency of bleeding are key aspects of the history in determining amount of blood loss. The character of bleeding is extremely important. Hematemesis is defined as the vomiting of blood, and usually represents upper GI bleeding (rarely bleeding from the nasopharynx or oropharynx). Hematemesis may be bright red when fresh, but older blood will resemble coffee grounds. Melena is defined as the passage of offensive, black, tarry stool, again usually due to upper GI bleeding. The appearance of the stool is a result of gastric acid degradation (which converts hemoglobin to hematin), as well as the effects of intestinal enzymes and bacteria. Rarely, in cases of slow intestinal transit, blood loss from distal small bowel or the right colon may also present as melena. A guaiac test will allow differentiation of the tarry black stool of melena from the dark green stool of patients on iron supplementation (melena will test positive). Bright red blood per rectum is called hematochezia—this may represent blood on the tissue paper, blood around the stool, or blood mixed in with the stool—all important features to elicit on history-taking. Hematochezia usually results from bleeding from the distal colon, usually sigmoid colon or rectum, but may also occur from massive upper GI bleeds with rapid intestinal transit.

Other Essential Features in the History

Other useful features to elicit in the history include antecedent vomiting (suggesting a Mallory–Weiss tear), recent weight loss or loss of appetite (suggesting malignancy), recent epigastric pain (possibility of peptic ulceration), and alcohol intake or liver disease (likelihood of variceal bleeding). Demographic data will assist in narrowing down the cause of bleeding—diverticulitis, angiodysplasias, malignancy, and ischemic colitis are likely culprits in the elderly. Younger patients are more likely to bleed from peptic ulceration, Meckel diverticula, hemorrhoids, or esophageal varices. In patients presenting with occult bleeding, the history may reveal syncope, angina, or myocardial infarction related to anemia. Previous abdominal surgery is relevant—previous aortic surgery in particular should raise suspicion of aorto-enteric fistula, and patients with gastrojejunal anastomosis, such as after gastric bypass, are susceptible to marginal ulceration. Drug history is particularly relevant in upper GI bleeding. Nonsteroidal anti-inflammatory drugs (NSAIDs) are a common cause of peptic ulceration, and similarly, salicylates and selective serotonin-reuptake inhibitors (SSRIs) are associated with upper GI bleeding.^18,19 Use of anticoagulants may require reversal with blood products.

Bleeding from the nasopharynx and oropharynx may occasionally present as GI bleeding, so these sites should be routinely examined. Pigmented lesions in the oral mucosa suggest Peutz–Jegher disease—a rare cause of GI bleeding. The abdomen should be examined to identify any masses or hepatosplenomegaly. A tender epigastrium may suggest peptic ulcer disease. The neck and groins should be examined for lymphadenopathy suggestive of malignancy. The examination should include inspection for stigmata of liver disease. The jaundiced patient with ascites, caput medusae, and palmar erythema presenting with GI bleeding should raise suspicion for variceal hemorrhage. Rectal examination and anoscopy are other essential aspects of the examination to exclude rectal cancer or, more commonly, hemorrhoids.

Performing a nasogastric (NG) tube lavage is an important diagnostic maneuver to help localize GI bleeding. An NG aspirate positive for fresh or old blood (either fresh blood or coffee grounds) confirms upper GI bleeding, aids in assessing the rate of bleeding, and allows removal of blood to facilitate endoscopic evaluation of the gastric mucosa during esophagogastroduodenoscopy (EGD).

A nonbilious, non-bloody aspiration of the stomach does not rule out bleeding from the duodenum, as a competent pylorus will prevent reflux of bile or blood into the stomach. A bilious aspirate without blood does suggest a lower GI source for the bleeding. However, a recent study showed that 20% of patients had a blood-free aspirate from the duodenum despite a diagnosis of upper GI bleeding.²⁰

Endoscopy in Upper GI Bleeding

EGD remains the gold standard investigation for the diagnosis and management of upper GI bleeding. EGD facilitates identification of the source of bleeding, determining the underlying etiology, achieving hemostasis, and providing prognostic information for risk stratification.²¹ The timing of endoscopic assessment in patients with GI bleeding remains controversial. Although there is little doubt that early endoscopy in hemodynamically unstable patients is necessary, the ideal timing for endoscopic intervention in stable patients remains less clear. A review of studies examining the utility of early endoscopic intervention in upper GI bleeding concluded that while endoscopy within 24 hours of presentation was of benefit in terms of aiding risk assessment and reduced length of hospital stay, earlier endoscopies (within 12 hours) offered no additional benefit. Indeed, endoscopy within 12 hours of presentation was associated with unnecessarily increased use of therapeutic endoscopy without any benefit in terms of rate of rebleeding or survival. Overall, these studies suggested that endoscopy should be performed within 24 hours of presentation, and in hospitals without a 24-hour endoscopy service, this should be offered to patients the following day.²²

Other issues should be considered regarding the use of EGD in acute GI bleeding. First, the sensitivity of EGD may be reduced in the presence of active bleeding, as mucosal visibility is impaired. Also, endoscopic complications such as perforation and aspiration increase in the emergency setting. Similarly, sedative medications administered during endoscopy can exacerbate hypotension and hypoxemia. Resuscitative measures should not be delayed or paused for the endoscopic procedure. All patients undergoing urgent endoscopy should be continuously monitored, and consideration given to early anesthesia consultation.

Endoscopy in Lower GI Bleeding

FLEXIBLE SIGMOIDOSCOPY AND COLONOSCOPY

Colonoscopy is recommended over flexible sigmoidoscopy in lower GI bleeding with few exceptions. Colonoscopy has been deemed the most appropriate investigation in patients over 50 years of age with hematochezia or iron deficiency anemia. In younger patients, colonoscopy can be omitted if a convincing benign source of bleeding has been demonstrated on flexible sigmoidoscopy but should be pursued in cases of repeated bleeding.²³

Colonoscopy has a diagnostic yield of 89% to 97% in the setting of acute GI bleeding.^24,25 Bowel preparation using polyethylene glycol with a prokinetic such as metoclopramide has been recommended to improve endoscopic visualization and thus diagnostic yield.^23,26 This step may have to be omitted in patients with severe ongoing GI bleeding, where there is insufficient time for a formal bowel preparation routine.

Capsule Endoscopy and Deep Enteroscopy

Endoscopic access to the small bowel is difficult, secondary to the length of the small bowel, intraperitoneal location, and contractility. Capsule endoscopy has emerged as a suitable option for small bowel imaging, and is now the third diagnostic test in patients with obscure bleeding following EGD and colonoscopy.²⁷ Capsules are swallowed that contain a camera that visualizes mucosal surface as it travels the intestine, and wirelessly transmits images for later review. A device is available for patients with dysphagia, dysmotility disorders, and children to deliver the capsule directly to the duodenum. Capsule endoscopy has better yield than push enteroscopy or small bowel series, and an equivalent yield to intraoperative enteroscopy without the morbidity and mortality of the operative procedure.^28,29 Capsule endoscopy is unsuitable for imaging of the proximal duodenum due to poor visualization of the periampullary region, and should not be performed in those with bowel obstruction or strictures. Yield of capsule endoscopy is dependent on the experience of the reader.^30,31

Techniques have been developed to allow endoscopy of the small bowel, including single balloon, double balloon, and spiral enteroscopy. The first two techniques use balloons to grip the intestinal wall and facilitate advancement of the endoscope through the intestine. Spiral enteroscopy uses a special overtube with helices at the distal end to pleat the small bowel onto the overtube, again allowing advancement of the endoscope through the intestine. The advantage of deep, or “push,” enteroscopy is the ability to perform biopsies, treat bleeding, and perform other therapeutic maneuvers.³² The most significant disadvantage of deep enteroscopy over capsule endoscopy is the risk of perforation (0.3% to 3.4%), particularly in patients with inflammatory bowel disease, malignancy, and bowel anastomosis.^31,33-36

Angiography

Visceral angiography is a relatively insensitive investigation, able to detect bleeding only at a rate of 0.5 to 1 mL/min.^37,38 Although the specificity approaches 100%, the sensitivity varies from 47% with acute lower GI bleeding to 30% with recurrent bleeding.³⁹ Angiography has a role in patients with massive lower GI bleeding precluding endoscopic visualization or in patients with negative endoscopies. Like endoscopy, angiography offers the advantage of potential simultaneous therapeutic intervention.

Red Cell Labeling (Nuclear Scintigraphy)

Red cell labeling has been found to play a limited role in the diagnosis of GI bleeding and may be useful after other methods have failed. While sensitive (this method can detect GI bleeding at a rate of 0.1 mL/min), the site of bleeding is localized to an area of the abdomen rather than a portion of the GI tract. Intestinal motility can shift intraluminal blood away from the site of bleeding, resulting in incorrect localization. Specificity is improved when scans are positive within 2 hours after injection of labeled erythrocytes, as less transit through the bowel will have occurred, with accurate localization in 95% to 100% of cases. Correct localization falls to 57% to 67% when scans are positive more than 2 hours after injection.⁴⁰ Red cell scans are therefore more often used to identify a potential role for subsequent angiography. In patients with negative red cell scans or scans positive only after 2 hours, angiography is unlikely to be positive.

CT Angiography

CT angiography (CTA) is a promising new technique that offers advantages compared to tagged RBC scanning. With the dissemination of high-resolution CT scanners, a CTA is available widely and at all times of the day and night. While nuclear RBC scanning takes hours to perform, a CTA can be obtained in minutes. A recent non-randomized retrospective study of patients who received either a tagged RBC scan or a CTA prior to visceral angiography showed that CTA was better at localizing the site of GI bleeding.⁴¹ Despite increased use of IV contrast in CTA patients, no adverse impact on renal function was demonstrated. CTA has been reported to detect blood loss as low as 0.4 mL/min, which would make it less sensitive than nuclear RBC scanning. Further studies are still required to determine the role of CTA in assessing GI bleeding.

Pharmacologic Management

Pharmacologic management is unlikely to halt active bleeding but is aimed at preventing recurrent bleeding. Proton pump inhibitors, but not H2 blockers, have been shown to reduce recurrent bleeding from gastric ulcers, as clot formation is stabilized in the absence of gastric acid. Octreotide is useful in variceal bleeding and may have an adjunctive role in other upper GI bleeds (see variceal bleeding below). Triple therapy treatment for Helicobacter pylori can prevent recurrent ulcers and bleeds.

Endoscopic Treatment

Endoscopy remains the mainstay of investigation and therapy for most causes of upper and lower GI bleeding. Techniques used for control of hemorrhage include thermal coagulation, injection therapy, and mechanical devices such as metallic clips and band ligation. Thermal coagulation probes include bipolar, monopolar, and heat probes, with an overall perforation rate of up to 2.5%, particularly frequent in the thin-walled right colon.⁴² Argon plasma coagulation is a means of non-contact coagulation with an almost nonexistent risk of perforation in the colon.⁴³ Laser-mediated coagulation (such as with the Nd:YAG laser) uses high-energy laser light to vaporize the tissue, producing deeper penetration than argon plasma coagulation but with a higher perforation rate.

Injection of a 1:10,000 dilution of epinephrine is an effective and inexpensive method of endoscopic treatment, causing vasoconstriction and physical compression of the vessel. Metallic clips, in both reusable and disposable forms, are used to control hemorrhage endoscopically. Rubber band ligation is frequently employed in lower GI bleeding due to hemorrhoids or rectal varices.

Interventional Angiography

While initial attempts of embolization led to high rates of bowel infarction due to the use of large-bore catheters for cannulation, current approaches with microcatheters have produced success rates of 70% to 90% without significant complications, and recurrent hemorrhage rates of only 15%.⁴⁴ Embolization materials include microcoils, gelfoam, and polyvinyl alcohol particles. Selective angiographic embolization has been shown to arrest life-threatening bleeding from gastroduodenal ulcers, with a low rate of early rebleeding and no late rebleeding, obviating the need for emergency surgery in high-operative-risk patients.⁴⁵

Early bleeding recurrence is associated with coagulation disorders, longer time to angiography, higher preprocedural blood transfusion volume, two or more comorbidities, and the use of coils as the only embolic agent.⁴⁶ Embolization has been shown to have 85% success rate in patients with diverticular lower GI bleeding, with higher efficacy in the left compared to right colon. Angiography is less efficacious in non-diverticular lower GI bleeding (eg, arteriovenous dysplastic lesions) with a greater than 40% rate of rebleeding.⁴⁷

Angiography may also be coupled with selective infusion of a vasoconstrictor such as vasopressin, or the longer-acting analogue terlipressin. However, this strategy is associated with a 50% rate of rebleeding after cessation of the infusion.⁴⁸ The side effects of vasopressin and terlipressin, including abdominal pain and cardiac complications, have meant that this technique is now only rarely used.

Surgery

Surgery is usually reserved for therapy when less invasive therapeutic modalities have failed and bleeding has been clearly localized. Surgery does remain the treatment of choice in patients with malignant or benign tumors, and may also be used as a last resort in patients with recurrent bleeding without a defined bleeding point or in fulminant hemorrhage. Further discussion of surgical options is detailed in relevant sections below.

Causes of Upper GI Hemorrhage

Causes of upper GI hemorrhage can be divided into variceal and non-variceal bleeding, which accounts for 80% to 90% of acute upper GI bleeding (Table 17-2).^6-13 Even in patients with portal hypertension, non-variceal bleeding is still more common. However, due to its morbidity and mortality, if suspected, variceal bleeding should be excluded first.

NON-VARICEAL BLEEDING

Peptic Ulcer Disease and Bleeding

Gastroduodenal peptic ulceration accounts for 40% of all non-variceal upper GI bleeding.²⁰ The introduction of H pylori eradication therapy and proton pump inhibitors has reduced the incidence of peptic ulcer disease (PUD), leading to reduced rates of operation and mortality from PUD. However, the overall incidence of peptic ulcer bleeding remains high, with significant associated mortality and cost.^49,50

A large proportion of PUD bleeding is linked to use of aspirin and NSAIDs, and the majority of cases occur in the elderly (68% of patients are over 60 years of age and 27% over 80 years of age).⁵¹ Ten percent to 15% of ulcers will bleed at some point during the course of the disease. Patients with bleeding ulcers commonly present with hematemesis and/or melena. History, examination, and investigations should proceed as outlined earlier (Fig. 17-2). Duodenal ulcers are more common, but gastric ulcers usually bleed more profusely. Ulcers involving an artery such as branches of the gastroduodenal or left gastric arteries can bleed significantly.

Several risk stratification scores have been developed to assist in identification of patients who require close monitoring and are at risk of rebleeding. The two most commonly used tools are the Rockall score and the Blatchford score (Table 17-3). The Rockall score utilizes clinical as well as endoscopic findings to risk-stratify patients. The score ranges from 0 to 11, with a higher score indicating a greater risk of rebleeding or death.⁵² The Blatchford score incorporates clinical and laboratory values to produce a maximum score of 23, with higher scores again associated with a greater likelihood of rebleeding or death.⁵³

The endoscopic appearance of a bleeding ulcer alone can also be used to stratify the risk of rebleeding using the Forrest criteria⁵⁴ (Table 17-4). High-grade lesions are those which are actively spurting or oozing blood, or have a non-bleeding visible vessel or adherent clot.

All ulcerogenic medication such as salicylates, NSAIDs, and SSRIs should be stopped and non-ulcerogenic alternatives prescribed. COX-2 inhibitors which initially showed promise as a gastroprotective alternative to NSAIDs have recently been shown to demonstrate cardiotoxicity without significant benefit on gastric mucosal protection and are therefore infrequently used.⁵⁵

The association of bleeding with H pylori infection is not as strong as the association reported for perforated ulcers, with H pylori infection reported in only 60% to 70% of bleeding ulcers. However, recent data show that treating patients positive for H pylori with eradication therapy reduces the risk of rebleeding and obviates the need for long-term acid suppression, hence eradication with triple therapy is recommended in all bleeders infected with H pylori.⁵⁶

Gastric acid has been shown to impair clot formation, promote platelet disaggregation, and increase fibrinolysis. In keeping with this, proton pump inhibitors have been shown to significantly reduce the risk of ulcer rebleeding, the need for urgent surgery and, in patients with high-risk stigmata who have undergone endoscopic therapy, mortality.^57,58

Patients with high-risk stigmata on endoscopy (active bleeding or non-bleeding visible vessel) require hemostatic intervention, such as injection or thermal or mechanical therapy, such as clips (Fig. 17-3). Addition of any one of these to adrenaline injection further reduces rebleeding rates, the need for surgery, and mortality.^59–61

Several factors are predictors of failure of endoscopic therapy for peptic ulcer bleeding, including previous ulcer bleeding, shock on presentation, active bleeding during endoscopy, ulcers >2 cm in diameter, a large underlying bleeding vessel ≥2 mm in diameter, and ulcers on the lesser curve of the stomach or the posterior or superior duodenal bulb.⁶² Recent studies suggest that second-look endoscopy (within 24 hours of the initial endoscopic therapy) provided only a small reduction in the rate of rebleeding, is not cost-effective in the presence of acid-suppressing medication, and is overall not recommended.^21,63,64 Repeat endoscopy should only be considered in cases of recurrent hemorrhage or unsuccessful first treatment.