Focused History

In addition to the presenting symptoms, one should solicit any prior history of GI bleeding, peptic ulcer disease, bleeding diathesis or chronic anticoagulation, renal or liver disease, alcohol abuse, or nonsteroidal anti-inflammatory drug (NSAID) use. It is important to assess the possibility of cirrhosis or other causes of portal hypertension, because management of variceal bleeding remains distinct from that resulting from other UGIB causes. The history may suggest bleeding etiologies. Retching or vomiting episodes immediately preceding the onset of UGIB suggest a Mallory-Weiss tear. A prior abdominal aortic aneurysm repair should prompt consideration of an aortoenteric fistula. Recent instrumentation of the pancreas, liver, or biliary tract should raise the suspicion of hemobilia or hemosuccus pancreaticus. Chronic epistaxis and skin telangiectasias indicate potential hereditary hemorrhagic telangiectasia (HHT, also known as Osler-Weber-Rendu syndrome).

With a clear history of vomiting bright red blood or “coffee grounds” (blood present in the stomach long enough to be acidified by gastric acid turns brown), the localization of UGIB is straightforward. However, occasional bleeding from the posterior pharynx or the lung may be confused with UGIB (see also Chapter 79). Unfortunately, a history of melena, resulting from bacterial degradation of hemoglobin, remains nonspecific. Melena commonly arises from brisk UGIB, as well as from a small bowel source (distal of the ligament of Treitz) or a slow bleeding from the right colon.

Focused Physical Examination and Laboratory Evaluation

The physical examination begins with evaluating the patient’s hemodynamic status. In addition to directing immediate resuscitation, initial vital signs have prognostic importance: 50% of patients presenting with shock have rebleeding episodes. An orthostatic pulse rise of more than 20 beats per minute implies an acute blood loss of at least 500 mL, whereas an accompanying fall in diastolic pressure of 10 mm Hg or more implies a loss of at least 1000 mL.

The initial examination should survey for stigmata of chronic liver disease (such as spider angiomas, gynecomastia, palmar erythema, ascites, and splenomegaly) or findings suggestive of an underlying malignancy. Cutaneous manifestations of diseases associated with GI bleeding, such as perioral petechiae in HHT, may also be detected.

Initial laboratory evaluation should include a complete blood count (CBC) with a platelet count, coagulation studies (prothrombin time [PT] and partial thromboplastin time [PTT]), and the determination of serum electrolytes, creatinine, blood urea nitrogen (BUN), bilirubin, and liver-associated enzyme levels. The initial blood draw should include a sample to be sent to the blood bank for immediate typing and cross matching.

Intestinal metabolism of blood raises serum BUN so that a BUN:creatinine ratio > 20 (with both BUN and creatinine expressed in mg/dL) supports the diagnosis of UGIB. However, this nonspecific finding can be seen in hypovolemia alone.

Gastric Lavage

Even with an obvious history of UGIB, gastric lavage is indicated to clear the stomach in anticipation of endoscopy. Gastric lavage decreases the risk of aspiration for the patient and improves endoscopic visualization. Sometimes, a large-bore tube (e.g., an Ewald tube) unlikely to be occluded by blood clots may be needed. No therapeutic advantage derives from the use of iced saline (versus room temperature tap water or saline) for the lavage fluid. Without evidence of recent bleeding (red blood or coffee grounds) in the initial gastric aspirate, the nasogastric tube may be removed.

A 30% mortality rate has been reported when both the gastric aspirate and stool contain red blood. However, up to 16% of patients with active UGIB may have clear gastric fluid on lavage because of intermittent bleeding or postpyloric bleeding without blood reflux into the stomach. Identification of bile in the gastric aspirate is notoriously inaccurate and is not evidence for the absence of UGIB.

General Care

Initial management is individualized based on the patient’s hemodynamics, bleeding rate, and comorbidities. General recommendations for the ICU patient start with assuring ample intravenous access (Box 61.2). For hypotensive patients experiencing exsanguination, use short large-bore (7-8 Fr) catheters (often called rapid infusion catheters or trauma lines) in peripheral veins (or longer large-bore catheters in the internal jugular or femoral vein) with corresponding wide-bore tubing and special three-way stopcocks plus a blood warmer. Alternatively, 8 Fr catheter introducer sheaths, routinely used for insertion of pulmonary artery catheters, can also be used (but again without small-bore stopcocks). Initially, give normal saline or Ringer’s lactate solution, titrated to keep the heart rate at less than 100 beats per minute and the systolic blood pressure (BP) higher than 100 mm Hg or the mean BP > 60 to 65 mm Hg, if possible. Once available, preferentially replace lost blood volume by transfusing pRBCs. Transfusions remain critical for cirrhotics, who tend to redistribute crystalloids to the extravascular space and acquire massive total body fluid overload. Transfusion timing and thresholds depend on the patient’s hemodynamic stability, underlying conditions, comorbidities, and risk of further bleeding. As a general rule, early consultation with a gastroenterologist and, as appropriate, with an interventional radiologist and general surgeon is recommended for ICU patients with UGIB.

During the first few hours after a bleeding episode, plasma volume and red blood cell mass decrease proportionately, so that the Hgb often remains normal despite significant bleeding. Later, after the plasma volume has expanded from crystalloid therapy, the Hgb may underestimate the quantity of red blood cells present. Transfusion goals can be summarized as (1) pRBCs given to improve oxygen delivery and provide a buffer in case further bleeding occurs, (2) fresh frozen plasma to correct coagulation defects, and (3) platelets to treat thrombocytopenia or platelet dysfunction. Maintaining the Hgb at ~10 g/dL was the traditional target for most patients with UGIBs, but a recent large controlled clinical trial published in 2013 by Villanueva et al indicated that using a threshold of 7 gm/dL was superior to a threshold of 9 gm/dL. In addition, in cases with known portal hypertension and bleeding from gastric or esophageal varices, an Hgb goal of 7 to 8 g/dL adequately resuscitates blood volume without increasing portal pressure and the risk of rebleeding.

Standard ICU monitoring should include continuous electrocardiographic and blood pressure monitoring, the latter either by an arterial catheter or a frequently cycled automated cuff. Patients with a history of congestive heart failure or other significant heart disease should be candidates for central venous or even pulmonary arterial pressure monitoring. Patients with respiratory insufficiency, or altered mental status at increased risk for aspiration, should undergo endotracheal intubation. Similarly, patients with active hematemesis should maintain the left lateral decubitus position to decrease aspiration risk, and tracheal intubation should be considered for airway protection. The importance of frequent repeated clinical assessments of the patient’s condition by ICU staff cannot be overemphasized.

Endoscopic and Angiographic Interventions

Once the patient is stabilized (and rarely in the unstable patient experiencing exsanguination), urgent endoscopy should be considered. Endoscopy is indicated in resuscitated patients with active hemorrhage, blood product transfusion requirements, persistent hypovolemia, known or suspected portal hypertension, or suspected aortoenteric fistula. Patients who rebleed after initial stabilization should also undergo urgent endoscopy. Endoscopy can be deferred for up to 24 hours in patients with self-limited bleeding and no hemodynamic instability. Esophagogastroduodenoscopy (EGD) correctly identifies the source of bleeding in most cases and also provides valuable prognostic information while allowing the initiation of proper therapy. The accuracy in identifying the bleeding source is highest within the first 12 to 18 hours of hospital admission (approximately 90%) and falls by 30% or more after 24 hours. Accurate identification of endoscopic features can predict the risk of rebleeding (Table 61.2). Because oral radiologic contrast studies offer no therapeutic benefit, and contrast agents may interfere with subsequent endoscopy, they have no role in the initial evaluation of UGIB.