Gastrointestinal (GI) Hemorrhage

1. Description of the problem

Gastrointestinal (GI) hemorrhage occurs anywhere throughout the GI tract. Classically it was differentiated into upper and lower GI bleeding, with the point of division being proximal or distal to the ligament of Treitz respectively. However, new insights into the pathology of bleeding and therapeutic strategies have led to a proposed new categorization. Bleeding originating in the small bowel is now viewed as a separate entity, reserving the term of lower GI hemorrhage only for those occurring in the colon and rectum. This chapter will follow this classification.

2. Emergency Management

Initial assessment and management

Although most bleeding episodes are of low magnitude and self-limited, the potential for life-threatening hemorrhage necessitates a systematic evaluation of the patient. This process begins by assessing vital signs and looking for signs of hemodynamic instability. If clinical signs of hypovolemia are present, they should be corrected without delay after obtaining adequate vascular access.

Continue Reading

Preferably, a large-bore intravenous catheter (at least 16 gauge) should be inserted in the antecubital fossa of each arm. If quick and easy access is needed due to patient instability, a central venous catheter can be inserted. However, the most commonly used central venous catheters have a higher flow resistance, resulting in slower fluid administration rates. Initial resuscitation begins with crystalloid solutions followed by packed red blood cell transfusions if necessary.

Initial laboratory examination should include a complete blood count, electrolytes and coagulation profile. Every effort should be made to correct a coagulopathy. Cirrhotic patients and those taking warfarin may require fresh frozen plasma (FFP), while patients taking aspirin or clopidogrel may require platelet transfusions. Renal failure patients with uremia or those with von Willebrand disease may respond to the administration of 0.3 mcg/kg of intravenous desmopressin (1-deamino-8-D-arginin vasopressin or DDAVP).

Routine endotracheal intubation is not recommended. However, the threshold for intubation should be low for patients who may be at high risk for aspiration due to severe vomiting or mental status changes secondary to conditions like shock or hepatic encephalopathy.

Unstable patients should always be admitted to an ICU. For stable patients, the decision is based on clinical judgment. Besides hemodynamic instability, additional recommended criteria for ICU admission are: two or more comorbidities, age above 60 years, and/or the need for hemodynamic monitoring or mechanical ventilation.

The source of bleeding needs to be identified to direct treatment. Active vomiting of blood is indicative of an upper GI source of bleeding.

For patients who are passing blood per rectum, a nasogastric tube should be inserted first. If the nasogastic aspirate contains bile without blood, it is safe to assume that the hemorrhage originates distally to the ligament of Treitz.

A rectal examination should always be performed to rule out anorectal pathology and evaluate the stool color.

Although frequently described as essential in aiding diagnosis, a complete history is generally not too useful and may be difficult to obtain from patients who are intubated or in shock. It can also obscure the diagnosis. For example, a history of cirrhosis secondary to alcohol abuse or hepatitis C may lead the clinician to suspect a variceal bleed when the actual bleeding source is from a duodenal ulcer or a gastric tumor. Consequently, all potential etiologies for bleeding need to be considered.

Specific algorithms

After the initial evaluation and resuscitation, specific protocols should be followed to confirm the diagnosis and to treat the source of bleeding. These algorithms will be further elaborated in the specific sections below, but as a general rule, this sequential approach consist of:

  • Pharmacological treatment

  • Endoscopy

  • Angiography and/or additional diagnostic imaging

  • Surgery

Pharmacological treatment can be initiated during the resuscitation or shortly thereafter. Unless contraindicated, proton pump inhibitors (PPI) are used routinely in almost all cases of upper GI bleeding, while in cases of suspected variceal hemorrhage specific protocols call for the simultaneous administration of vasopressin or somatostatin or their analogs. For patients with lower GI bleeding, a quick preparation of the colon, ideally with polyethylene glycol solution (NuLYTELY or GoLYTELY), is usually necessary to allow adequate visualization.

Whatever the type of GI bleeding, prompt endoscopy is essential as it not only can identify the source of bleeding but can also offer therapeutic options. If timely endoscopy is not available or cannot be done, angiography is an alternative to identify and potentially treat the source of bleeding. Angiography can detect a bleeding vessel if the rate of hemorrhage is greater than or equal to 0.5 mL/min. Once an arterial source is identified, treatment with embolization is performed with a success rate greater than 90%. Ischemic complications are rare in the upper GI tract due to the abundance of collaterals but more common in the lower GI tract.

The role of computed tomography (CT) angiography is less clear. Although CT angiography appears to be as accurate as conventional angiography in identifying the source of massive GI bleeding, it is not therapeutic. If a bleeding site is identified using CT angiography, an additional procedure (i.e. surgery or conventional angiography) will be required for treatment. Therefore, CT angiography should be employed only when conventional angiography is unavailable.

Autologous tagged red blood cell scan with technetium (99mTc) is another option to localize bleeding. It is more sensitive than angiography as it can detect bleeding rates as low as 0.1 mL/min. Once the source of bleeding is identified, treatment is determined by the patient’s hemodynamic stability. In the hemodynamically unstable patient, emergency surgery is performed, while another attempt at angiographic embolization can be made in the stable patient.

Upper GI hemorrhage

Patients with upper GI bleeding tend to present with hematemesis. Upon evaluation, the nasogastric tube aspirate is often bloody. These patients also commonly have melena. If the bleeding is significant, it can be accompanied by hematochezia. There is a multitude of potential etiologies for upper GI bleeding; the two most common are peptic ulcer disease and variceal hemorrhage. The initial approaches to these differ and they will be described later in detail. Less common causes include Dieulafoy’s lesion, stress ulceration, Mallory-Weiss tear, esophagitis and aortoenteric fistulas. However, the initial approach for these is very similar to the one for for peptic ulcer disease.

Peptic ulcer disease

Peptic ulcer disease (PUD) is the most frequent cause of upper GI bleeding. Common etiologies include the use of nonsteroidal anti-inflammatory drugs (NSAIDS) and Helicobacter pylori infection. If present, eradication of H. pylori is unnecessary in the acute setting but is important for long-term prevention of recurrence.

Significant PUD bleeding is frequently due to a posterior penetrating ulcer in the first portion of the duodenum that erodes into the gastroduodenal artery. The bleeding usually stops spontaneously in approximately 80% of cases. However, the remaining 20% of patients will have recurrent bleeding or it will not stop at all. Historically, surgery was indicated to control the bleeding in a significant number of patients, but it is rarely required these days as other less invasive therapies are available.

Unless there is a contraindication, a proton pump inhibitor (PPI) should be used. PPIs are effective in reducing the intragastric pH and the need for endoscopic treatment, and appear to be more effective than histamine receptor 2 (H2) blockers or somatostatin. Administration of high-dose omeprazole after endoscopy (80 mg bolus injection followed by 8 mg/hr continuous infusion for 72 hours, and then 40 mg/day orally for 1 week) compared to the standard dose has also been found to decrease the incidence of recurrent bleeding, reduce the need for surgery and prevent shock that leads to death.

Prompt endoscopy is essential as it allows identification and treatment of the bleeding vessel. The administration of the prokinetic agent erythromycin (3 mg/kg IV) before endoscopy can improve visibility during the procedure. Endoscopy within 24 hours has been strongly recommended, with some groups performing the endoscopy within 3 hours of admission. This approach has been associated with a significant decrease in recurrent bleeding, the need for surgery and length of stay.

Local injection of epinephrine, thermocoagulation and endoscopic clips are all used to stop bleeding. When used alone or in combination, these methods are successful in over 90% of the cases. A detailed description of these endoscopic procedures is beyond the scope of this chapter and will therefore be discussed only briefly to inform the intensivist about possible complications. Injection of diluted epinephrine (1:10,000) is effective in stopping bleeding, especially when used in higher volumes.

Several proposed mechanisms include direct vasoconstriction, a tamponade effect related to the volume of injection and induction of platelet aggregation. Although less frequent with high-volume injections, rebleeding can occur in up to 20% of the patients treated with epinephrine alone. Thermal coagulation is more effective than epinephrine in preventing rebleeding and the need for surgery; however, it carries an approximately 1% risk of perforation. Hemoclips can also be applied endoscopically to control a bleeding vessel. Although very safe, they are difficult to deploy and require a certain degree of expertise that is simply not available in many centers.

Even after initial endoscopic control, rebleeding occurs in up to 20% of patients, resulting in a significant increase in morbidity and mortality. Identification and treatment of those ulcers with potential for rebleeding is necessary to avoid complications. The Forrest classification is very useful at predicting recurrent bleeding. There are three categories based on endoscopic findings. Forrest I lesions are those that are actively bleeding at the time of the endoscopy. Forrest II lesions have stigmata of recent bleeding. Forrest II lesions are further classified into Forrest IIa lesions, where a non-bleeding vessel is visible, and Forrest IIb lesions, defined by the presence of an adherent clot.

Forest III lesions are those that do not have signs of recent bleeding. Forrest I and II lesions have a high incidence of rebleeding. Nearly all Forrest I lesions rebleed. Forrest IIa lesions have rebleeding rates of approximately 40% and a mortality rate of 11%. Endoscopic treatment is recommended for Forrest I and IIa lesions.The role of endoscopic treatment for Forrest IIb lesions is less clear, as consensus is lacking on what constitutes an adherent clot. Many clinicians are reluctant to mobilize a clot because of the potential to restart bleeding. Nevertheless, administration of PPIs at high doses as described above can help stabilize the clot and decrease the rebleeding rate.

Angiography is indicated for patients who continue to bleed despite endoscopic treatment or in whom endoscopy cannot be performed. Embolization of an actively bleeding vessel has a high success rate. If a bleeding site cannot be identified, blind embolization is not recommended. However, if the source of bleeding was previously identified endoscopically, blind embolization of the suspected vessels can be as effective as targeted embolization.

A metallic clip placed at the time of endoscopy may be helpful in identifying the area where bleeding occurred. If rebleeding takes place, a new embolization can be reattempted for a combined success rate of 95%. A dreaded complication of embolization is contrast-induced acute renal failure, particularly in combination with volume depletion caused by hemorrhage. Duodenal ischemia is rare and usually can be treated conservatively with PPIs. Mortality is in the range of 10-45%, likely related to the associated patient comorbidities.

A minority of patients require surgery to control the bleeding. Because surgery is usually performed after failure of endoscopic and/or angiographic embolization, the patients are usually sicker and have a higher mortality rate. Although the reported mortality with surgery is similar to angiographic embolization, the mortality rate after failed embolization has been reported to be as high as 83%.

The procedure of choice is oversewing of the bleeding vessels with or without an acid-reducing procedure such as a vagotomy and pyloroplasty, a vagotomy and antrectomy or a highly selective vagotomy. With the current availability of highly effective acid-suppressive medications and H. pylori treatment, extensive surgery is usually not indicated in these very sick patients.

Variceal hemorrhage

The second most common cause of upper GI bleeding is variceal hemorrhage. Varices are present in approximately 50% of patients with cirrhosis and become more prominent with advanced liver disease. They develop commonly in the lower esophagus and stomach secondary to portal hypertension. The hepatic vein pressure gradient is the main determinant of the propensity for variceal bleeding. The thin-walled varicose veins, located in the weak lamina propria of the lower esophagus, are especially predisposed to rupture and bleeding.

Despite recent improvements in the medical management of these patients, mortality remains high at 20% in the first 6 weeks. The rebleeding rate ranges between 30% and 40% in the first 6 weeks and is associated with a mortality of 30%. Because of the high mortality rate, prompt management is crucial. When the diagnosis of variceal hemorrhage is suspected, if available, the vasopressin analog terlipressin should be administered promptly as this is the only agent proven to reduce mortality. It is administered at a dose of 2 mg every 4-6 hours for the first 48 hours, followed by half this dose for up to 5 days.

In the United States, where terlipressin is unavailable, somatostatin and the somatostatin analog octreotide are valid therapeutic options. These medications are effective at controlling the bleeding in up to 80% of cases. They can also facilitate endoscopic visualization by reducing the rate of bleeding.

Somatostatin causes splanchnic vasoconstriction and decreases portal pressure. It is administered as an initial bolus of 250 mcg, which can be repeated up to three times, followed by an infusion of 250 mcg/h for up to 5 days to prevent rebleeding. The role of octreotide is less clear. Although it has a longer half-life compared to somatostatin, its hemodynamic effects are not as pronounced. It is used as a continuous infusion at 25 or 50 mcg/hour preceded by a 50- or 100-mcg bolus.

Although somatostatin and octreotide have similar success rates of controlling variceal hemorrhage, use of these agents alone has not been shown to decrease mortality. Therefore, their use is recommended in combination with endoscopic therapy.

Historically vasopressin has been used successfully to stop acute variceal hemorrhaging. However, its administration is associated with significant side effects (myocardial infarct, mesenteric and limb ischemia, cerebrovascular accidents and hyponatremia) in a significant number of patients and it requires the concurrent administration of intravenous nitroglycerin (10-50 mcg/minute). Vasopressin is administered as an infusion starting at 0.4 U/minute and can be incrementally increased to 1 U/minute.

Although it controls bleeding in up to 80% of cases, vasopressin does not decrease mortality. This is most likely due to the severe ischemic events it causes in other vital organs. Therefore, vasopressin should be reserved for those rare circumstances when more effective drugs are not available.

Antibiotic prophylaxis should be started early since up to 20% of patients with cirrhosis and gastrointestinal bleeding develop infections. Antibiotics have been shown to significantly reduce the number of infections, prevent early rebleeding and improve survival in these patients. Ceftriaxone 1 gm IV daily is the recommended antibiotic. Studies show it to be superior to fluoroquinolones. However, for those patients with a penicillin allergy, fluoroquinolones can be used as an alternative.

Patients should be admitted to an ICU for close monitoring and management. Endotracheal intubation is frequently recommended as these patients are at risk of aspiration or may be confused due to hepatic encephalopathy. Resuscitation with crystalloids and packed red blood cells needs to be initiated promptly to avoid hypovolemia and subsequent complications like renal failure. However, hypervolemia should be avoided as this could worsen variceal bleeding.

Upper endoscopy should be performed next to determine the exact source of bleeding. Despite esophageal and gastric varices being a common source of bleeding in cirrhotic patients, up to 25% of cases will have another bleeding source.

Esophageal varices can be treated with sclerotherapy or band ligation. The type of endoscopic treatment depends on the experience and expertise of the endoscopist as well as the magnitude of bleeding. Band ligation appears to be superior in reducing the risk of long-term bleeding and has been associated with fewer complications. However, it can be difficult to perform in massive bleeds, and in these cases, sclerotherapy is preferred.

The success of endoscopic treatment is similar to vasoactive drugs, with a rate of 80-85%. However, despite having similar success rates to vasoactive drugs for controlling bleeding, a recent Cochrane meta-analysis found that sclerotherapy could not be recommended as a first-line treatment given its higher complication rates. Its use should be reserved for pharmacological failures. Conversely, the combination of pharmacological and endoscopic treatment in selected patients appears to be superior to vasoactive drugs alone.

Gastric varices bleed less frequently but more intensely than esophageal varices, leading to higher transfusion requirements and mortality. The recommended treatment is sclerotherapy with histoacryl glue (N-butyl-2-cyanoacrylate). Complications include ulceration of the mucosa and embolic events from the glue.

With massive bleeding, a Sengstaken-Blakemore (S-B) tube, or one of its variations, can be inserted as a temporizing measure. The S-B tube has fallen out of favor as more effective therapies have become available. The risks of its use include aspiration, esophageal rupture and an inability to control the bleeding. However, for selected patients, its placement can be life-saving. The S-B tube has a gastric and an esophageal balloon. The placement should be preceded by endotracheal intubation.

The tube is introduced through the nose or, more frequently, through the mouth into the stomach. The position of the distal tube in the stomach is confirmed by insufflation of air combined with auscultation. Radiographic confirmation has been advocated to avoid insufflation of the gastric balloon in the esophagus resulting in esophageal rupture. After confirming the position of the S-B tube, the gastric balloon is inflated with 500 mL of saline and gentle traction of approximately 1 kg (1 liter IV bag) is applied.

This maneuver stops bleeding from varices high in the fundus and blocks collaterals from the stomach to the submucosa of the esophagus. If the bleeding stops, the balloon is secured in place using a football helmet or a pulley system connected to an IV pole. If esophageal bleeding persists, the esophageal balloon is inflated next. The inflation port is connected to a manometer and pressure is gradually increased until the bleeding stops or it reaches 45 mmHg.

Once the bleeding is controlled, the S-B tube is kept in place, usually for 24 hours, before deflating the balloons. This allows preparation for other therapies such as endoscopy or a transjugular intrahepatic portosystemic shunt (TIPS). If bleeding recurs after deflation, the balloons can be reinflated.

If the previously described treatments are not successful, TIPS can be attempted. TIPS placement reduces the hepatic vein pressure gradient and effectively controls bleeding in 90-100% of cases. However, emergency TIPS placement is associated with a high incidence of hepatic encephalopathy and a mortality rate as high as 50%. In the event that TIPS placement is not available, a surgical portosystemic shunt should be considered.

One surgical approach is the interposition of an 8-mm ringed polytetrafluoroethylene shunt between the portal vein and the inferior vena cava. Because of the relatively small diameter of the graft, this creates only a “partial” portocaval shunt with preservation of hepatopedal flow, leading to a lower incidence of encephalopathy. In experienced hands, the graft has a 95% patency rate at 7 years. However, even its advocates advise against it if the patient is a candidate for liver transplantation.

A splenorenal shunt (with or without splenectomy) is another surgical option. The operative mortality for this procedure is reported to be less than the portocaval shunt; however, this has not been validated in the literature. Although some studies report similar complication and mortality rates compared to TIPS, only a few specialized centers have surgeons experienced enough in performing these procedures on very sick patients.

Lower GI hemorrhage

Patients bleeding from the colon or rectum present with hematochezia or blood per rectum. Occasionally, bleeding from the right colon may present as melena due to longer transit time, allowing the oxidation of iron in the hemoglobin.

Common causes of lower GI hemorrhage include colonic diverticula, angiodysplasia, inflammatory bowel diseases and neoplasms. As in bleedings originating in other areas of the GI tract, a sequential approach is recommended to establish the diagnosis and to treat the patient. Despite the disparity in etiologies, the algorithm to diagnose and treat colonic bleed is remarkably similar. Again, as a general rule, this sequential approach consist of:

  • Colonoscopy

  • Angiography and/or other diagnostic imaging

  • Surgery

While the initial evaluation and resuscitation is taking place, an attempt should be made to prepare the colon prior to the colonoscopy. Unless contraindicated, this can be achieved by administering 240 mL of a polyethylene glycol (PEG) electrolyte solution (GoLYTELY or NuLYTELY) every 10 minutes until the stool is free of solid matter. Although in patients who cannot drink it can be administered through a nasogastric tube, special care is needed to prevent and avoid aspiration. The use of laxatives to prepare the colon is contraindicated in patients with any type of gastrointestinal obstruction, perforation or toxic megacolon.

After quick cleansing of the bowel, endoscopy can be diagnostic and therapeutic. If the colon cannot be prepared and/or endoscopy is unsuccessful, angiography is the next step.

Because rectal bleeding can be diagnosed without a full colonoscopy and approached locally, it will be discussed separately.

Colonic diverticula

Diverticulosis of the colon is an acquired condition that results in herniation of the mucosa through the muscular layer of the colon. Colonic diverticula are prevalent in the Western hemisphere but rare in Africa and Asia. Diverticulosis develops with age, being rare before age 40, but with a prevalence greater than 65% by age 80. The cause of bleeding is erosion of a vessel (vasa recta) into the diverticulum.

Although diverticula are more common in the left colon, a significant number of bleeds originate from the right side. Bleeding stops spontaneously in 80% of cases. Identification of the source is usually made with endoscopy, which can also be therapeutic. Treatment includes thermocoagulation or epinephrine injections. The use of hemoclips has also been described. If endoscopy cannot be performed, angiography can be diagnostic and therapeutic. The reported success rate with super-selective embolization exceeds 85% with a very low incidence of clinically significant ischemia (<5%).

Because bleeding tends to be intermittent and an estimated bleeding rate of more than 0.5 mL/min is necessary for angiographic visualization, the angiography will sometimes be negative. In such cases, provocative angiography, as described previously, can be helpful, leading to identification and successful embolization of the offending vessel in one third of patients.

If angiography is not possible or available, CT scan localization may be an option. A limited surgical resection can be undertaken if the source of bleeding is identified. Surgery is indicated when bleeding does not stop. Mortality is high in these patients, approaching 20% even in tertiary centers. It is critical to identify the site of bleeding before surgery. In retrospective reviews, blind segmental resection is prone to failure and has been associated with a high mortality rate. A subtotal colectomy rather than segmental resection is therefore recommended in those cases where the source cannot be identified preoperatively.


Angiodysplasias are small, ectatic blood vessels found in the mucosa and submucosa of the GI tract. They are also known as arterial venous malformations (AVMs) or vascular ectasias. Their prevalence is around 1% of the population. Angiodysplasias are usually asymptomatic and occur more often in older individuals. Most lesions are located in the right colon, although they can be found in other areas of the GI tract. Even though angiodysplasias are arguably the second most common cause of lower GI bleeding, the reported frequency as a cause for lower GI bleeding varies widely from 3-37%.

Identification by colonoscopy is usually difficult, particularly after administration of opioids or cold water irrigation, as they decrease mucosal blood flow. The avoidance of narcotics during colonoscopy or the use of a narcotic antagonist has been proposed as a means to facilitate endoscopic visualization of angiodysplasia in small studies. However, this has not been validated in large studies.

Even when angiodysplasias are identified by colonoscopy, they should not be assumed to be the cause of lower GI bleeding. Endoscopic treatment entails thermal probe coagulation and injection of epinephrine. It is recommended that in actively bleeding angiodysplasias, treatment should start in the periphery of the lesions to address the feeding vessels before treating the source of hemorrhage.

Angiography is reserved for patients who cannot undergo colonoscopy because of massive bleeding or in those with nondiagnostic colonoscopies. The diagnosis of angiodysplasia is made by identification of early filling of ectatic veins. Transcatheter embolization can be therapeutic but carries the risk of inducing significant ischemia.

In selected cases, CT angiography can be as accurate as angiography in establishing the diagnosis and localizing the source of bleeding. If the bleeding cannot be controlled by endoscopic means or angiographic embolization, surgery is indicated.

Inflammatory bowel disease

Severe inflammation of the colon accounts for 1-5% of lower GI hemorrhaging. Severe life-threatening bleeding is the indication for up to 10% of emergency colectomies performed in ulcerative colitis patients. About 1% of patients with Crohn’s disease also will need emergency surgery. Medical treatment should be initiated first. Endoscopy can be diagnostic and occasionally therapeutic. Angiography may also have a role in the diagnosis and treatment of these patients.

Surgery should be reserved for patients who continue to bleed or experience recurrent bleeding. Colonic bleeding due to ulcerative colitis is usually associated with severe pan-colitis and is treated with a subtotal colectomy and ileostomy. Identification of the source of hemorrhage is very important in Crohn’s disease as the bleeding originates in the small bowel in approximately two thirds of patients. Limited resections are recommended for these patients.


Although chronic bleeding from colonic adenocarcinoma or polyps is common, massive hemorrhage requiring ICU admission is rare. Colonic polyps can bleed spontaneously or, more commonly, after colonoscopic resection. Nevertheless, in recent series, the incidence of significant post-polypectomy bleeding has been very low (less than 1%). Major bleeding after polyp resection is usually arterial. It can be treated endoscopically with a combination of thermocoagulation and epinephrine injections.

Rectal bleeding

Massive rectal bleeding is uncommon, accounting for around only 2% of lower GI hemorrhage. Hemorrhoids are the most frequent cause; however, substantial bleeding is rare except in patients with portal hypertension. If the bleeding does not stop spontaneously, band ligation is a safe option.

In the elderly population, solitary rectal ulcer must be considered as a possible etiology of lower GI bleeding. Patients with these ulcers are usually bedridden and debilitated with multiple medical comorbidities. Bleeding can be profuse and life-threatening. Multiple therapeutic approaches have been described to control the bleeding, including endoscopic thermocoagulation, vascular clips and transanal sutures.

Another cause of rectal bleeding is from radiation changes of the rectum after treatment for prostate cancer. The mechanism is radiation-induced endarteritis obliterans, which results in neovascularization with telangiectasias that are prone to bleeding. The presentation of bleeding usually occurs within months of radiation, but can be delayed for several years.

Ischemic proctitis, Dieulafoy’s lesions not associated with solitary rectal ulcers and stercoral ulcers are less common causes of rectal bleeding and usually treated with some variation of the methods described above.

Small bowel bleeding

A normal upper endoscopy and colonoscopy should alert the clinician to a possible small bowel origin of bleeding. Traditionally, bleeding originating in the small bowel has been classified as lower GI bleeding. However, new diagnostic and therapeutic implications merit its description as a distinct entity. The reported incidence is between 1% and 7% of patients who present with blood per rectum.

Prakash and Zuckerman described a cohort of patients whose source of bleeding was in the small bowel. In their series, patients bleeding from the small bowel had significantly worse outcomes compared to patients with colonic bleeding. Also, the patients with small bowel hemorrhage underwent a higher number of studies before the diagnosis of the source was made.

The most common cause of small bowel bleeding is angiodysplasia, followed by small bowel tumors. Angiodysplasia is associated with advanced age and chronic renal failure. Less frequently, the bleeding is caused by inflammation in Crohn’s disease, Meckel’s diverticulum, NSAID or aspirin use, Dieulafoy’s lesions and varices in portal hypertension.

The diagnosis is difficult and usually reached through a combination of studies. Because of its wide availability, 99mTc-tagged RBC scan is the most commonly used test. However, because of the high rate of false-positive and false-negative results, its use cannot be recommended. The intermittent nature of bleeding also results in angiography being positive in only about 50% of the cases. Radiographic contrast studies of the small bowel are not advocated. Small bowel follow-through has a very poor sensitivity in identifying the source of bleeding.

Enteroclysis, a procedure where the contrast is instilled directly into the duodenum together with methylcellulose and air to improve visualization, is only marginally better.

Some experts advocate push enteroscopy, whereas others are proponents of capsule endoscopy. Push enteroscopy with a dedicated video enteroscope allows for visualization of the proximal 50-100 cm of the small bowel. The advantage of push enteroscopy is that it can be both diagnostic and therapeutic. Although capsule endoscopy appears to be more sensitive, some of the patients may need additional procedures to control the bleeding afterwards. In patients who are actively bleeding, capsule endoscopy has a sensitivity of up to 92%. However, the sensitivity decreases to less than 50% in cases with intermittent bleeding.

Double balloon enteroscopy is a relatively new technique that allows for complete visualization of the small bowel. It uses a dedicated 200-cm enteroscope with 2 balloons. One of them is attached to the tip of an endoscope and the other to the tip of a flexible overtube. The balloons are sequentially inflated, allowing them to grip the mucosa of the small bowel and advance the endoscope without looping. The enteroscope can be inserted orally, transanally or both to examine the entire small bowel. Like push endoscopy, it can take biopsies and potentially treat the bleeding source. Unfortunately, double balloon enteroscopy is not yet widely available.

Meckel’s diverticulum

Meckel’s diverticulum is the most common congenital anomaly of the gut, with a prevalence of around 1% of the population. Complications are rare; however, heterotopic gastric mucosa in the diverticulum can occasionally cause bleeding. This occurs most frequently in children but should be considered in any patient younger than 40 who has a small bowel bleed. If the diagnosis is not established by the above techniques, Technetium-99m pertechnetate scintigraphy (Meckel’s scan) can be used. Surgical resection is the treatment of choice.

Finally, for patients who continue to bleed, exploratory laparotomy with or without intraoperative enteroscopy is an option to identify and treat the source of bleeding.


Upper GI bleeding is more common than lower GI bleeding, with an annual incidence of hospitalization in the U.S. of approximately 100 cases per 100,000 adults, in contrast to about 20 per 100,000 patients with lower GI bleeding. When compared to patients with upper GI hemorrhage, patients with lower GI hemorrhage are less likely to experience shock and usually require fewer blood transfusions. GI hemorrhage is most common in the elderly, with a reported incidence as high as 500 cases per 100,000 people per year. With the increasing elderly population, this number is expected to steadily increase in the coming years.

The mortality reported in the majority of studies has remained basically unchanged over the past 60 years despite major improvements in the management of GI hemorrhage. This persistently high mortality has been attributed to an aging population with many associated comorbidities, leading to a high death rate independent of the GI hemorrhage, with 80% of the mortality attributable to other causes.