While hematemesis is clearly a symptom of upper GI bleeding, black tarry melenic stool predicts an upper GI bleeding source, and brighter colors of red in the stool are more often associated with a distal colonic bleeding source. However, color of bloody stool may not always be helpful in predicting the level of GI bleeding and is subject to interpretation variability of both patients and physicians. A pocket-sized color card is helpful in confirming the stool color, as described by the patient, and suggesting the level of bleeding in the GI tract [10]. When bright blood in the stool (implying a lower GI bleed) is associated with hemodynamic compromise, as many as 11% of patients may have an upper GI bleeding source, even if the NG aspirate is negative [3,4]. In this setting, an upper endoscopy may be the first endoscopic evaluation even though the presenting symptom is hematochezia.

When bleeding is suspected to originate proximal to the jejunum, esophagogastroduodenoscopy (upper endoscopy) is the diagnostic procedure of choice. This identifies the bleeding source in 80% to 90% of cases with a high degree of accuracy, provides therapeutic options, and carries low morbidity [3,4]. Endoscopy has the added advantage of detecting prognostic signs (Table 91.2) and classifies bleeding stigmata as high or low risk for rebleeding based on the Forrest grade (Table 91.3) [11].

Even when an exact diagnosis cannot be made, localizing the bleeding to a specific region within the upper GI tract can be helpful to the surgeon (if resection is indicated) or interventional radiologist (if embolization of the bleeding vessel is recommended). Erythromycin or metoclopramide can be administered intravenously to induce gastric emptying and clear the stomach of blood and clots prior to endoscopy [12,13]; repeated lavage with saline through a wide-bore orogastric tube also may be used for this purpose. However, routine gastric lavage may not be necessary and is not endorsed by the authors. Complications related to endoscopy are higher when the procedure is performed on an emergency basis.

The timing of endoscopy in upper GI bleeding continues to be evaluated. Endoscopy within 12 hours of presentation increased the use of endoscopic therapy but did not reduce rebleeding rates or improve survival rates [14]. However, endoscopy within 24 hours did demonstrate a reduction in the length of hospital stay and need for surgical intervention [15,16]. Patients with bloody NG aspirate did benefit from
endoscopy within 12 hours to reduce the blood transfusion requirements and length of hospital stay [14]. Therefore, endoscopy offered within 24 hours of upper GI bleeding presentation appears appropriate, with consideration for early endoscopy within 12 hours for patients with bloody NG aspirate or clinical suspicion for high-risk lesions [17]. The benefit of a repeat “second-look” endoscopy is an area of investigation, especially in the presence of factors associated with an increased risk of rebleeding (history of peptic ulcer disease, previous ulcer bleeding, presence of shock at presentation, ulcers > 2 cm, large underlying bleeding vessel ≥ 2 mm diameter, and ulcers located in lesser curve of stomach or posterior/superior duodenal bulb) [18,19]. A meta-analysis in 2003 concluded that second-look endoscopy was associated with a decreased risk of recurrent bleeding but did not alter subsequent surgery rates or mortality [20]. Scheduled repeat endoscopy therefore is not routinely recommended, but can be considered on an individual case basis if clinical signs of recurrent bleeding are present or if there are questions about adequate hemostasis [21,22].

If a small bowel lesion is suspected after a negative upper endoscopy, a longer endoscope can be used to evaluate the proximal small bowel (push enteroscopy), which allows visual inspection and endoscopic hemostasis of bleeding lesions as far distal as the proximal jejunum [23]. Further evaluation of small bowel bleeding lesions can be provided by capsule endoscopy. Disadvantages of capsule endoscopy in acute bleeding include the lack of accurate localization of visualized lesions, and the fact that the test is not performed in real time [23]. Bleeding lesions beyond the reach of a push enteroscope can potentially be approached using single- and double-balloon enteroscopy, techniques that allow for visualization of most of the small bowel. Balloons at the endoscope tip and an overtube can be consecutively inflated and deflated while inserting and pulling out the endoscope to allow bowel to pleat over the overtube, thus allowing deep endoscope insertion into the small bowel, either through the mouth or the anus [23].

When a distal lower GI bleeding source is suspected, early sigmoidoscopy may be helpful if the bleeding is not of a magnitude that would prevent adequate visualization. For most situations, however, colonoscopy replaces sigmoidoscopy in the diagnostic approach. Early colonoscopy provides a higher yield of the bleeding source compared to radiologic studies, especially when performed within 24 hours of presentation [24,25]. In patients with severe hematochezia and diverticulosis, urgent colonoscopy (within 6 to 12 hours of hospitalization or diagnosis of hematochezia) after rapid bowel purge can provide endoscopic treatment of diverticular hemorrhage and may prevent recurrent bleeding and decrease the need for surgery [26]. Only 20% of patients with lower GI bleeding, however, have a lesion amenable to endoscopic intervention [27]. Even when the exact cause of bleeding cannot be determined, colonoscopy may localize fresh blood to a segment of colon and direct further therapies such as angiotherapy or surgery. Patients with subacute bleeding or hemorrhage that has ceased can undergo adequate bowel preparation followed by semiurgent colonoscopy [28].

The technetium-99m–labeled red blood cell scan performed at the bedside offers a noninvasive diagnostic approach to patients suspected of having GI bleeding originating beyond the reach of an endoscope, especially in unstable patients where bowel preparation or endoscopy cannot be safely performed. Although bleeding rates as low as 0.1 mL per minute can be detected by this method, the patient should have evidence of ongoing bleeding during the study [7]. If the test localizes bleeding, angiography or endoscopy (push enteroscopy, colonoscopy, double-balloon enteroscopy, capsule endoscopy) is needed to confirm the site, to further define the cause, and to offer therapy for ongoing bleeding [7]. If the test is negative, colonoscopy followed by capsule endoscopy is usually performed to evaluate potential colonic and small bowel bleeding sources [23].

Because a more rapid bleeding rate is necessary for a positive arteriogram (0.5 mL per minute), this procedure typically is performed after active bleeding is documented on a radionuclide bleeding scan [7]. However, because of the intermittent nature of bleeding and the variable timing of mesenteric arteriography, a positive red blood cell scan does not always result in a diagnostic arteriogram [7]. Arteriography is also useful for upper GI bleeding sources not visualized on upper endoscopy because of rapid bleeding or a blood-filled stomach.

Endotherapy offers a convenient and expedient method for treatment of GI bleeding. Although endotherapy was primarily used for the treatment of upper GI and peptic ulcer bleeding, these modalities can also be applied to patients with lower GI bleeding [4,29,30,31]. Endotherapy is indicated for all patients with high-risk lesions because of the significant risk of persistent or recurrent bleeding (22% to 55%) and even death
if left untreated [3,4,32]. Randomized trials demonstrate that endotherapy for upper GI bleeding decreases further bleeding, shortens hospital stay, decreases transfusions, decreases emergency surgery, decreases mortality, and lowers costs [17]. Optimal therapy for adherent clots remains controversial; a recent meta-analysis demonstrated reduced rebleeding rates (RR 0.35, 95% CI 0.14–0.82) with endoscopic removal of clot and treating of the uncovered lesion, but no change in length of hospitalization, need for surgery, transfusion requirements, and mortality compared to only medical therapy [33,34,35].