Albert Einstein Medical Center, Philadelphia, PA, USA
Definition of disease
Bleeding that originates anywhere in the GI tract and may manifest as hematemesis, hematochezia, melena, or maroon‐colored stools.
Upper GI bleeding originates proximal to the ligament of Treitz, whereas lower GI bleeding originates distal to the ligament of Treitz.
The incidence of acute upper GI bleeding is 50–100 per 100 000 persons per year in the USA.
The incidence of lower GI bleeding is 20 cases per 100 000 individuals per year.
The most common cause of acute upper GI hemorrhage is peptic ulcer disease. Additional causes include but are not limited to non‐steroidal anti‐inflammatory (NSAID) use, stress gastritis, gastroesophageal varices, and Mallory–Weiss tears.
The most common causes of lower GI bleed include diverticulosis (24–47%), colitis (6–36%), neoplasms (9–17%), and angiodysplasia (2–12%).
Upper GI bleeding:
Duodenal ulcers occur more frequently than gastric ulcers and bleeding may occur with ulcerative erosion through the posterior wall into the gastroduodenal artery, resulting in hemorrhage or hematemesis.
NSAID use is also associated with mucosal injury that may result in ulcerative disease causing acute upper GI hemorrhage.
Stress gastritis as a cause of upper GI bleeds results from superficial gastric ulcers due to altered gastric mucosal blood flow and impaired clearance of hydrogen ions from the mucosa.
Gastroesophageal varices may cause upper GI bleeding in cirrhotic patients.
Dieulafoy vascular malformations are generally due to unusual large arteries running through the gastric submucosa which may result in bleeding due to gastric erosion.
Lower GI bleeding:
The most common cause of lower GI bleeding is diverticulosis. Bleeding from diverticulosis results from a perforated vasa recta at the neck or apex of a diverticulum.
Colonic angiodysplasia arises from age‐related degeneration of previously normal intestinal submucosal veins and overlying mucosal capillaries.
Ischemic colitis is due to impaired local microvascular perfusion of the colonic wall.
Aspirin 81 mg
Aspirin plus NSAIDs
Aspirin plus antiplatelet
Respiratory failure/mechanical ventilation
Patients with gastric and duodenal ulcers should have screening biopsies obtained for H. pylori. Patients who test positive should be treated.
With suspected upper GI bleeds, PPIs should be started on presentation until the cause of bleeding is confirmed. PPIs should be used as stress ulcer prophylaxis in critically ill patients to prevent GI bleeding.
The risk of ulcer formation in patients taking NSAIDs is significantly reduced with prophylactic use of PPI or histamine‐2 (H2) receptor antagonists. PPIs are more effective in protecting patients on NSAIDs than H2 receptor blockers.
Beta‐blockers can be used for primary prophylaxis of bleeding from varices. Beta‐blockers have a 9% absolute risk reduction of GI bleeding episodes.
Endoscopic surveillance of upper GI ulcers and detection and treatment of H. pylori leads to a reduction in GI bleeds.
Decreasing the use of antiplatelet and anticoagulant agents as appropriate reduces GI bleeding episodes.
PPIs should be used over H2 blockers to reduce rebleeding episodes after successful endoscopic therapy. PPIs may promote hemostasis via neutralization of gastric acid leading to stabilization of blood clots. PPI therapy decreases rebleeding rate, in addition to decreasing the length of hospital stay and blood transfusion needed.
Beta‐blockers have a 21% absolute risk reduction of recurrent GI bleeds.
Endoscopic treatments can prevent rebleeding. Specifically, endoscopy plus prophylactic treatment of varices decrease bleeding risk. Banding ligation is superior to injection sclerotherapy treatment and beta‐blockers alone. Endoscopy plus PPI treatment has a lower risk of rebleeding than PPI treatment alone (11.6% risk of bleed versus 1.1%).
Treatment for H. pylori if indicated helps reduce the recurrence of GI bleeding compared with no treatment or chronic antisecretory treatment alone.
Somatostatin analogs like octreotide and prophylactic antibiotics in cirrhotic patients can reduce rebleeding from variceal causes.
Upper GI bleeding may present as hematemesis or melena (dark or tarry black stool); lower GI bleeding may present as hematochezia (maroon stool, bright red blood per rectum).
Hematemesis and melena are generally indicative of a bleeding source proximal to the ligament of Treitz. Frank hematemesis is of greater clinical concern, whereas coffee ground emesis may be indicative of a more limited upper GI bleed.
Melena occurs due to a source proximal to the ligament of Treitz (90%), oropharynx, small bowel, or colon. May see bloody aspirate on nasogastric lavage or identify source via upper endoscopy.
Hematochezia may present secondary to a massive upper GI bleed which may result in orthostatic hypotension or due to lower GI bleed. A lower source may be supported by a negative nasogastric aspirate and a bleeding scan or colonoscopy.
When taking a history, it is important to elicit the patient’s general medical history to understand their comorbidities as well as an accurate medication history. Specifically, a history of epigastric pain with NSAID use may lead one to think of peptic ulcerative disease as a source of GI bleed.
A recent history of vomiting and retching with associated hematemesis may be secondary to Mallory–Weiss.
A history of alcohol use may suggest gastroesophageal variceal bleed or portal hypertension.
In patients who have had an aortic graft repair an aorto‐enteric fistula should be ruled out.
Weight loss in a smoker with vague abdominal pain and dysphagia may point to a malignancy.
A history of medical comorbidities including renal or hepatic dysfunction are important in identifying coagulopathies. A medication history including use of NSAIDs or antiplatelet or anticoagulant agents is important in planning patient resuscitation and treatment.
The physical exam should begin with an examination of the patient’s general status. A patient who is obtunded, tachypneic, tachycardic, and diaphoretic is of great concern and resuscitation should be initiated immediately.
Orthostatic hypotension is indicative of a 15% blood volume loss and supine hypotension may indicate a 40% blood volume loss. Additionally, one should inspect the patient for signs of anemia including cool, clammy, mottled, or pale skin or mucous membranes.
An examination of the abdomen may demonstrate epigastric abdominal pain consistent with peptic ulcer disease versus peritoneal signs consistent with a possible perforated viscus.
A rectal exam should be performed to identify any obvious source of distal rectal bleed including hemorrhoids or fissures.
Useful clinical decision rules and calculators
The International Consensus Upper Gastrointestinal Bleeding Conference Group recommends using mortality risk stratification tools. Two scoring systems used for GI bleeding include the Rockall score and the Blatchford score.
The Rockall score (www.mdcalc.com/rockall‐score‐upper‐gi‐bleeding‐complete) incorporates age, shock, comorbidity, diagnosis, and endoscopic stigmata of recent hemorrhage (range 0–11). Although validated for risk stratification, the Rockall score also predicts recurrent bleeding. Rebleeding occurred in <5% of patients, with mortality 0–0.2% among patients with scores 0–2. For scores ≥5 one‐fourth to one‐half of patients experienced a rebleed.
The Glasgow Blatchford score (www.mdcalc.com/glasgow‐blatchford‐bleeding‐score‐gbs) can be calculated on patient presentation since no endoscopic information is required. The Blatchford score incorporates the blood urea nitrogen, hemoglobin, systolic blood pressure, pulse, and presence of melena, syncope, hepatic disease, and/or cardiac failure. The score ranges from 0 to 23 with an increasing score having a greater association with need for urgent endoscopic intervention. The modified Blatchford was found to outperform the Rockall score and full Blatchford score in regard to predicting the need for clinical intervention, rebleeding, and mortality.
The AIMS65 score (www.mdcalc.com/aims65‐score‐upper‐gi‐bleeding‐mortality) also uses data available prior to endoscopy and has a high accuracy for predicting inpatient mortality among upper GI bleed patients.
List of diagnostic tests
Patients presenting with GI bleeds should be evaluated with routine labs including CBC and chemistries to evaluate for anemia and resuscitation status. Additionally, liver function tests, a coagulation panel, and type and cross should be obtained to assess the patient’s coagulopathy and to prepare for blood transfusion if necessary.
An ECG and cardiac enzymes should be obtained in patients who have sustained a sizeable blood loss or have resultant hypotension particularly in the face of a cardiac history in order to evaluate for ischemic cardiomyopathy.
A nasogastric (NG) tube can provide a simple diagnostic test to differentiate upper and lower GI bleeding since it is inexpensive and safe with a sensitivity of 42% and specificity of 91%. The NG tube should be placed and lavaged. Bloody output identifies a likely source of upper GI bleed and has been shown to result in a shorter time to endoscopy. Clear output is equivocal and requires further investigation, whereas non‐bloody bilious output suggests no upper GI bleed.
List of imaging techniques
Upper endoscopy is the diagnostic modality of choice for acute upper GI bleed with a high sensitivity and specificity for localizing bleeding, and can be utilized for treatment as well. Colonoscopy should be completed if hematochezia or melena are found in a patient with a negative upper endoscopy.
Technetium‐99m sulfur colloid allows for detection of GI bleeding at rates from 0.1 to 0.5 mL/min. For sulfur colloid scans, the patient must be actively bleeding during the limited time the label is present. With technetium labeling of RBCs, greater sensitivity for detection of GI bleeding was noted. With this mode, patients with intermittent bleeding can be scanned several times over a 24 hour period. The anatomic accuracy for tagged RBC scans is 70–85%, and it may serve as a screening for angiography.
CT angiography serves as a fast, widely available, and minimally invasive diagnostic tool that detects bleeding at rates of 0.3–0.5 mL/min. A sensitivity of 85% and specificity of 92% were reported in a meta‐analysis of 22 studies. The study found greater precision when it came to localizing the site of bleeding. CT angiography, however, lacks therapeutic capability, requires radiation exposure, and utilizes intravenous contrast, which can cause allergic reactions or nephropathy.
Angiography requires a rate of active blood loss of 0.5–1.0 mL/min. Negative arteriograms can be reduced with screening radionuclide imaging. If no prior localization exists, the superior mesenteric artery is examined first, followed by the inferior mesenteric or celiac vessels with success rates of 25–70%. Angiography does not require bowel preparation and provides accurate anatomic location in addition to potential therapeutic intervention with transcatheter embolization techniques.
Potential pitfalls/common errors made regarding diagnosis of disease
Radionuclide scanning with sulfur colloid or tagged RBCs and CT angiographic diagnostic imaging techniques require active bleeding at the time of diagnosis. Most GI bleeding stops spontaneously and thus if the bleeding has ceased, localization is difficult to achieve. Additionally, to localize with the sulfur colloid label the patient must be actively bleeding in the few minutes the label is present, while for a tagged RBC study the patient may be intermittently scanned for 24 hours in order to localize the lesion.
The accuracy rates for radionuclide studies are incredibly variable with accuracy ranging from 24% to 91%. The difficulty localizing the bleeding lesion is based on the need for active bleeding as well as complicating factors such as a redundant bleeding left colon appearing as a right‐sided GI bleed.