Decision Making in Emergency Critical Care
SECTION 7 - Gastrointestinal and Hematological Critical Care
Parvathi A. Myer and Shai Friedland
LOWER GASTROINTESTINAL BLEEDING
Lower gastrointestinal bleeding (LGIB), defined as bleeding below the ligament of Treitz, accounts for 20% of all acute GI bleeding.1 In the United States, the incidence of LGIB ranges from 20.5 to 27 cases per 100,000 adults.2 Compared to acute upper GI bleeding, patients with acute LGIB experience less shock, require fewer blood transfusions, and have a significantly higher hemoglobin level.3 The mortality rate of acute LGIB is 2% to 4%, and bleeding stops spontaneously in 80% to 85% of patients.3
The causes of hemodynamically significant hematochezia include diverticular bleeding (17% to 40%), angiodysplasia (9% to 21%), colitis (including ischemic, infectious, chronic inflammatory bowel disease [IBD], or radiation injury) (2% to 30%), neoplasia and postpolypectomy bleeding (11% to 14%), anorectal disease (4% to 10%), upper GI bleeding (1% to 11%), and small bowel bleeding (2% to 9%).4
Diverticular bleeding is typically due to arterial bleeding, presents as painless hematochezia, and increases in prevalence with age. Most diverticula are in the left colon, but the majority of bleeding diverticula are in the right colon.5 Although diverticular bleeding stops spontaneously in 80% of cases, the cumulative rebleeding rate is approximately 25% after 4 years.6
Angiodysplasias (or vascular ectasias/angioectasias) appear as red, mucosal lesions on colonoscopy and account for up to 30% of LGIB. They commonly occur in the right colon and increase in frequency with age. Most angiodysplasias do not bleed; bleeding is precipitated by platelet dysfunction and coagulopathy and is more frequent in patients with renal failure.5 Radiation proctopathy can result in telangiectasias, which can lead to rectal bleeding that is typically low volume and chronic. In a minority of patients, telangiectasias are due to hereditary hemorrhagic telangiectasia.5
Ischemic colitis classically presents with mild abdominal pain associated with self-limited hematochezia. The etiology of ischemic colitis is frequently a decrease in mesenteric blood flow in a watershed distribution due to low blood pressure or vasospasm.5 Ischemic colitis risk factors include underlying cardiovascular disease, atherosclerosis, and advanced age.5 Endoscopically, the colonic mucosa appears edematous and can show areas of submucosal hemorrhage or necrosis.
Mucosal inflammation can be caused by IBD or an infectious colitis. In 50% of patients with inflammation-associated bleeding, bleeding will stop spontaneously; however, 35% will rebleed.7 Nonsteroidal anti-inflammatory drug (NSAID) use has been noted to greatly increase the risk of bleeding in these conditions.5
Neoplasia and Postpolypectomy Bleeding
Two to nine percent of hematochezia is due to colon cancer and presents most commonly as occult or low-volume bleeding. Bleeding may occur up to 14 days after colonoscopic polypectomy, usually because of arterial bleeding from the stalk of the polyp.5 Postpolypectomy bleeding accounts for 2% to 8% of acute LGIB.3
Hemorrhoids account for 2% to 9% of hemodynamically significant LGIB4. Solitary rectal ulcers resulting from internal rectal prolapse can also produce rectal bleeding. Rectal varices (enlarged blood vessels) can also cause hematochezia and significant GI bleeding.
A Dieulafoy lesion is an artery that is exposed through a colonic mucosal defect and can bleed profusely.5 They are classically observed in the proximal stomach but can occur anywhere in the GI tract including the colon and rectum. These lesions can be difficult to locate endoscopically.
Initial Evaluation and Risk Stratification
A complete patient history and physical exam are essential during the initial emergency department (ED) evaluation. The reported color of the stool—maroon, bright red, or black and tarry (melena)—and confirmatory rectal exam are key to determining if the bleeding stems from an upper or lower gastrointestinal (UGI or LGI) source. The rectal exam also helps to assess rate of bleeding and the presence of anorectal pathology. Patient history should note duration and frequency of bleeding, associated symptoms (light-headedness, dizziness, palpitations, syncope, abdominal pain, and fevers), sick contacts, travel history, and prior episodes of GI bleeding.8 The use of NSAIDs, family or personal history of colon cancer, prior radiation exposure, IBD history, liver disease, coagulopathy, and weight loss are important details to obtain.5 In the setting of brisk hematochezia, clinicians must always maintain a high index of suspicion for the presence of a brisk upper GI bleed (UGIB), as 11% to 15% of patients with hematochezia are bleeding due to massive UGIB.9
Patients with brisk bleeding or showing signs of hemodynamic instability including orthostatic hypotension, chest pain, dyspnea, and tachypnea should be monitored in the intensive care unit (ICU). ICU placement should also be considered for LGIB patients with significant comorbidities or those requiring two or more units of packed red blood cells (PRBCs). Initial laboratory studies should include a complete blood count, serum chemistry including BUN and creatinine, and an international normalized ratio (INR).5 In addition to volume resuscitation, patients with active bleeding and an INR > 1.5 or a platelet count <50,000/μL will require transfusion of fresh frozen plasma or platelets, respectively.5 A recent randomized clinical trial (RCT) in patients with acute UGIB suggests that a restrictive strategy of transfusion for hemoglobin <7 g/dL may be optimal.10 This study provides new guidelines for blood transfusion in the setting of GI bleeding and represents a departure from previous recommendations for transfusion in critically ill patients established by the Transfusion Requirements in Critical Care (TRICC) Investigators study.11 Patients with underlying cardiac disease should still be transfused when the hemoglobin drops below 9 g/dL.2
To help identify patients with severe LGIB, the following validated risk factors should be assessed: heart rate > 100, systolic blood pressure < 115 mm Hg, syncope, nontender abdominal exam, bleeding within the first 4 hours of evaluation, aspirin use, and greater than two comorbid conditions (Charlson Comorbidity Index).12,13 In a patient with a LGIB and hemodynamic instability, placement of a nasogastric tube (NGT) and evaluation of the aspirate for bile or blood help identify whether brisk UGIB is the cause of the patient's rectal bleeding. If NGT aspirate is clear, and an esophagogastroduodenoscopy (EGD) is not deemed necessary, colonoscopy should be planned within 12 to 48 hours of arrival to the ED. All patients should have a clear liquid diet until 4 hours before the procedure. Four liters of GoLytely or an equivalent polyethylene glycol solution should be administered at a rate of a liter per hour until 4 hours before the procedure.
Colonoscopy is the preferred modality for identifying the source of an LGIB and achieving hemostasis with the fewest complications.5 The risk of serious complications from a colonoscopy is approximately 1 in 1,000 procedures.5 In patients with severe LGIB due to diverticular bleeding, colonoscopic treatment may prevent recurrent bleeding and decrease the need for surgery.9
Timing of Colonoscopy
The timing of colonoscopy has been the subject of several RCTs.8,14 The most recent considered 85 patients with hematochezia with the following attributes: heart rate > 100 bpm, systolic blood pressure < 100 mm Hg, orthostatic change in heart rate or blood pressure > 20 mm Hg, hemoglobin drop > 1.5 g/dL, or requiring blood transfusion. These patients all underwent EGD within 6 hours to exclude brisk UGI bleeding; 13 (15%) were found to have an UGI source. Of the remainder, 36 patients were randomized to an urgent colonoscopy group (<12 hours) and 36 were randomized to an elective colonoscopy group (36 to 60 hours). There were no differences in clinical outcomes, including recurrent bleeding, units of blood transfused, hospital days, need for subsequent interventions, treatment of bleeding, hospital charges, or length of stay.8 The study concluded that the timing of colonoscopy between 12 and 60 hours from initial ED evaluation made no difference in clinical or economic outcomes.8
In patients with brisk LGIBs unable to undergo colonoscopy due to hemodynamic instability or with bleeding that is too brisk to permit successful colon preparation, immediate angiography, radionucleotide scintigraphy (tagged red blood cell [RBC] scan), computed tomography (CT), or surgery may be necessary.12,15
A minimum colonic bleeding rate of 0.5 to 1 mL/min is required to be detected by angiography; the higher the bleeding rate, the more likely the study will be able to localize its source.12 A spontaneous bacterial peritonitis (SBP) < 90 and the need for more than 5 units of PRBCs in 24 hours are also predictors that angiography will locate the source of bleeding.16 Therapeutically, angiography works by embolization of the arterial branches that feed the bleeding site. Embolization has been demonstrated to be more effective than vasoconstrictor infusion and carries a lower risk of bowel infarction.12,15Complications of angiography include contrast allergies, nephrotoxicity, hematomas, thrombosis, and vascular dissections.15,12
Radionucleotide scintigraphy, or tagged RBC scanning, detects LGIB at rates as low as 0.05 to 0.1 mL/min12; source detection is improved when the scans are positive within 2 hours.12 No randomized trials have compared tagged RBC scanning to angiography, and studies regarding the diagnostic advantage of performing tagged RBC scans prior to angiography are equivocal.12,15 One RCT compared colonoscopy with tagged RBC scan following angiography and demonstrated colonoscopy to be the superior diagnostic test.14 A tagged RBC scan has the advantage of being noninvasive and not requiring special patient preparation; its disadvantage, however, is that it provides no therapeutic option for controlling a bleeding site once identified. In the setting of a brisk LGI bleed, tagged RBC scans remain valuable when time-consuming colonoscopy preparation prevents urgent localization of the bleeding source. RBC scans are also useful at the time of colonoscopy if the bleeding site cannot be identified or hemostasis cannot be achieved endoscopically. In this setting, the tagged RBC scan would be used to localize the site of bleeding, followed by angiography for hemostasis.
Multidetector CTs have improved imaging time and the ability to detect arterial bleeds; in animal models, they have detected bleeding rates as low as 0.3 to 0.5 mL/min.12 The main disadvantage of CT, as with the tagged RBC scan, is the inability to provide therapeutic options. Other disadvantages include radiation exposure, false-positive rates, contrast allergies, and potential contrast-induced nephrotoxicity.12 CT appears to be highly effective in detecting vascular ectasias.17
Surgery is performed when LGIB is recurrent and other measures, such as colonoscopy or angiography, have proven unsuccessful. Preoperative angiography can localize the bleeding source and appears to be associated with decreased rebleeding rates.15
NONVARICEAL UPPER GASTROINTESTINAL BLEEDING
Nonvariceal upper gastrointestinal bleeding (NVUGIB) has a mortality rate of 10% to 14%18 and imposes a significant clinical and economical burden on the U.S. health care system. Cases range from 48 to 160 per 100,000 adults per year,19 with an associated mean length of hospital stay of 2.7 to 4.4 days.20
Initial Evaluation and Risk Stratification
Initial evaluation—as with any critically ill patient—begins with an assessment of airway, breathing, and circulation (ABC). Once the ABCs are attended to, patients may be stratified as either high- or low-risk for re-bleeding and mortality using clinical assessment, laboratory data, and endoscopic criteria for risk of rebleeding. This type of risk stratification can help the emergency physician and gastroenterologist determine the appropriate timing of endoscopy. Clinical predictors of increased risk include age over 65, multiple comorbidities, hemodynamic instability, melena, poor overall functional status, hematochezia, hematemesis, and bloody nasogastric aspirate. Concerning laboratory data include a low initial hemoglobin, and/or elevated BUN, creatinine, or serum aminotransferase.19 Endoscopic predictors of increased risk for rebleeding include arterial bleeding, nonbleeding visible vessel or adherent clot, ulcer size greater than 2 cm, ulcer location in the posterior lesser gastric curvature or posterior duodenal wall, and varices or cancer.19
Initiation of intravenous proton pump inhibitor (PPI) therapy with an 80-mg bolus followed by an 8 mg/hour infusion rate is recommended for all NVUGIB. High-dose PPI therapy decreases the proportion of patients that will present with endoscopic findings that place them at risk of significant hemorrhage and/or in need of therapeutic intervention (e.g., active bleeding, nonbleeding visible vessel, or adherent clot). High-dose PPI therapy in NVUGIB does not reduce mortality, rebleeding rate, the need for blood transfusions, or the need for surgery.21,22
The TRICC trial is a landmark study that has historically guided ICU blood transfusion strategies. This study stratified ICU-admitted patients who had been given blood transfusions into a restrictive (transfused when hemoglobin dropped below 7 g/dL) versus liberal (when hemoglobin dropped below 10 g/dL) transfusion strategy. The trial found that less critically ill patients (Acute Physiology and Chronic Health Evaluation II score < 20) and patients under the age of 55 had significantly decreased mortality rates when transfused with the restrictive transfusion strategy. Patients with significant underlying cardiac disease had improved mortality rates with the liberal transfusion strategy.11
A 2013 landmark RCT has provided strong new evidence, and established a new standard, for blood transfusion in patients with active GI bleeding.10 Unlike the TRICC trial, which surveyed ICU-admitted patients requiring blood transfusions for all causes, this trial was performed exclusively in patients with acute UGI bleeding. Furthermore, the 2013 trial studied a lower hemoglobin cutoff for the liberal transfusion strategy. Of the 921 patients enrolled in the trial, 421 patients were randomized to a restrictive transfusion strategy when the hemoglobin dropped below 7 g/dL; and 460 patients were randomized to a liberal transfusion strategy when the hemoglobin dropped below 9 g/dL. Patients in the restrictive group had higher probability of survival at 6 weeks, lower recurrence of further bleeding, and lower risks of adverse events.10 Patients with cardiovascular disease, hypotension (systolic blood pressure < 90 mm Hg), or thought to be hemoconcentrated due to low systemic volume should still be considered for more liberal transfusion strategies.2
Infusion of erythromycin 250 mg 30 minutes before endoscopy stimulates gastric emptying and has been known to increase endoscopic visualization, increase diagnostic yield, and decrease the need for repeat endoscopy in randomized trials.23 Only two small studies have evaluated the benefit of metoclopramide, with no significant benefits noted.23
Correction of Coagulopathy
Correction of coagulopathy is recommended but should not delay early endoscopy.19 Data on the correction of coagulopathy for patients with NVUGIB are sparse and often contradictory. For patients on anticoagulation therapy, the threshold for correcting the INR varies widely in different studies.19 The International Consensus on Nonvariceal Upper Gastrointestinal Bleeding, which stresses the importance of early endoscopic intervention in NVUGIB, provides a general recommendation to correct supratherapeutic INRs prior to endoscopy.19
Endoscopic findings are useful in predicting an individual patients' risk of rebleeding. Initial endoscopic findings that place a patient at highest risk for rebleed and warrant intervention include ulcers with active bleeding, nonbleeding visible vessels, and adherent clot with an underlying vessel visualized after clot removal. Current guidelines recommend the use of clips or thermal coagulation alone and in combination with epinephrine; monotherapy with epinephrine alone is no longer recommended.25 Patients with flat pigmented spots and clean-based ulcers are at low risk and generally do not benefit from endoscopic treatment.
Timing of Endoscopy
In patients who are hemodynamically stable with no significant comorbidities, endoscopy should be performed within 24 hours, following which patients can often be discharged home if demonstrated to have low-risk endoscopic findings (clean-based ulcers or ulcers with flat pigmented spots).2 In patients with a more concerning clinical profile (tachycardia, hypotension, bloody emesis on NG lavage), endoscopy within 12 hours is recommended, as this may improve clinical outcomes.2 In low-risk hemodynamically stable patients, expedited endoscopy resulted in earlier hospital discharge and lowered costs. No clinical outcome data exist, however, to support emergent endoscopy in the low-risk group.
Discharge of Low-Risk Patients from the ED
Multiple assessment scoring systems exist to risk-stratify patients and to predict mortality and the need for clinical intervention such as blood transfusions, endoscopic treatment, or surgery. Some of these systems, such as the Rockall system, require endoscopic criteria to risk-stratify patients, making them less helpful in the ED; others, such as Glasgow-Blatchford Bleeding Score (GBS), require only clinical and laboratory data.24 A prospective study performed in the United Kingdom compared the GBS to the Rockall system for predicting mortality and need for clinical intervention (blood transfusions, endoscopic treatment, or surgery). Of the 676 patients presenting with acute GI bleeding, 105 received a GBS score of 0, indicating a low risk of need for intervention and an ability to be safely discharged from the ED without endoscopy if they meet the following additional criteria: urea nitrogen < 18.2 mg/dL, hemoglobin > 13.0 g/dL for men and 12.0 g/dL for women, systolic blood pressure > 100 mm Hg, pulse < 100 bpm, and absence of melena, syncope, cardiac failure, and liver disease. The study used receiver operator characteristic (ROC) scores to compare each score's ability to predict mortality and the need for clinical intervention. The GBS outperformed the Rockall system in predicting both measures.24 A follow-up study tested the GBS in clinical practice; of 123 patients with UGI bleeding, 84 (68%) were characterized as low risk (GBS score of 0) and were successfully managed in the outpatient setting. No clinical interventions were required and no deaths occurred.24
VARICEAL UPPER GASTROINTESTINAL BLEEDING
Patients with suspected acute gastroesophageal variceal bleeding should be admitted to an ICU for management and resuscitation.26 Gastric and esophageal varices are formed as an end result of cirrhosis. Cirrhosis results from advanced liver disease and is characterized by hepatic tissue fibrosis, which leads to a structural resistance to hepatic blood flow and intrahepatic vasoconstriction due to an associated decrease in nitric oxide production. These changes result in portal venous system hypertension and the formation of a collateral (gastric and esophageal) circulation. Elevated portal pressures persist, however, because of resistance to portal flow within the collateral circulation and increased portal venous blood flow from concurrent splanchnic vasodilation.26 Fifty percent of patients with cirrhosis will have gastroesophageal varices, and variceal wall tension is the primary determinant of variceal rupture. Variceal hemorrhage typically occurs when the hepatovenous portal gradient is over 12 mm Hg.26
Based on the findings of the 2013 trial discussed above, blood transfusions should now target a hemoglobin of 7 to 8 g/dL10,26; excessive transfusion and vigorous saline infusion should be avoided because of resulting increases in portal pressures and increased risk of variceal rebleed.10,26 Data from the 2013 study recommended a similar restrictive transfusion strategy for cirrhotics with variceal bleeding as for patients with NVUGIB. Survival improved in all patients assigned to the lower transfusion threshold (<7 g/dL compared to 9 g/dL); this benefit was magnified in the subgroup of patients with cirrhosis and a Child-Pugh class A or B disease.10,27 Compared to the restrictive strategy group, cirrhotics in the liberal transfusion strategy had significantly higher portal pressure gradients.10
Octreotide causes splanchnic vasoconstriction and is thought to decrease vasodilatory peptides such as glucagon, thereby helping to counteract the increased portal venous blood flow from splanchnic vasodilation seen in cirrhotics.26 Current guidelines recommend octreotide be given as an initial 50-μg bolus followed by a 50 μg/h infusion for 3 to 5 days following initial presentation of variceal bleeding.26
Patients with cirrhosis with gastroesophageal variceal bleeding are at high risk of bacterial infections including SBP and bacterial peritonitis, which cause increased risk of variceal rebleed and increased overall mortality.26 Current guidelines recommend that patients receive antibiotics pre-endoscopy to cover gram-negative bacteria and for a total of 7 days after initial GI bleed. Recommended antimicrobials include norfloxacin 400 mg PO BID or ciprofloxacin 500 mg IV BID for patients unable to tolerate oral intake. In areas of high fluoroquinolone resistance, ceftriaxone 1 g/day is preferred.26
There are two endoscopic methods for treating esophageal varices. The first, endoscopic variceal band ligation (EVL), deploys bands across varices with stigmata of recent hemorrhage with subsequent necrosis and sloughing of the varix. The second, sclerotherapy, injects a sclerosing agent such as cyanoacrylate into a bleeding varix to obtain hemostasis. A meta-analysis of 10 RCTs demonstrated EVL to be superior in overall outcomes when compared to sclerotherapy (pooled relative risk of 0.53).28 When EVL is not available or technically infeasible, sclerotherapy should be used.26
Data on endoscopic management of bleeding gastric varices are minimal. In contrast to esophageal varices, sclerotherapy is recommended over EVL. TIPS should be considered when bleeding continues despite endoscopic attempts for hemostasis.
Transjugular Intrahepatic Portosystemic Shunt
Transjugular intrahepatic portosystemic shunt (TIPS) is a procedure performed by interventional radiology. It utilizes an expandable metal stent that creates a connection between the hepatic vein and the intrahepatic portal vein to help decrease portal pressure in the setting of acute variceal GI bleeding. TIPS may be considered in patients who are Child-Pugh class A or B with variceal bleeding and have failed endoscopic and medical therapy.26,27
The management of gastroesophageal bleeding requires a focused patient history and physical exam, close hemodynamic monitoring, rapid resuscitation using a restrictive transfusion strategy, and prompt endoscopic evaluation. In variceal bleeding, an octreotide infusion to promote splanchnic vasoconstriction and prophylactic antibiotics to prevent bacterial translocation are also recommended. EVL, sclerotherapy, and TIPS are three validated management options for refractory variceal bleeding.
CI, confidence interval; NNT, number needed to treat; OR, odds ratio.
1.Zuccaro G, Jr. Management of the adult patient with acute lower gastrointestinal bleeding. American College of Gastroenterology. Practice Parameters Committee. Am J Gastroenterol. 1998;93:1202–1208.
2.Laine L, Jensen DM. Management of patients with ulcer bleeding. Am J Gastroenterol. 2012;107:345–360; quiz 361.
3.Farrell JJ, Friedman LS. Review article: the management of lower gastrointestinal bleeding. Aliment Pharmacol Ther. 2005;21:1281–1298.
4.Zuckerman GR, Prakash C. Acute lower intestinal bleeding. Part II: etiology, therapy, and outcomes. Gastrointest Endosc. 1999;49:228–238.
5.Barnert J, Messmann H. Diagnosis and management of lower gastrointestinal bleeding. Nat Rev Gastroenterol Hepatol. 2009;6:637–646.
6.Longstreth GF. Epidemiology and outcome of patients hospitalized with acute lower gastrointestinal hemorrhage: a population-based study. Am J Gastroenterol. 1997;92:419–424.
7.Robert JR, Sachar DB, Greenstein AJ. Severe gastrointestinal hemorrhage in Crohn's disease. Ann Surg. 1991;213:207–211.
8.Laine L, Shah A. Randomized trial of urgent vs. elective colonoscopy in patients hospitalized with lower GI bleeding. Am J Gastroenterol. 2010;105:2636–2641; quiz 2642.
9.Jensen DM, Machicado GA. Diagnosis and treatment of severe hematochezia. The role of urgent colonoscopy after purge. Gastroenterology. 1988;95:1569–1574.
10.Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med. 2013;368:11–21.
11.Hebert PC, Wells G, Tweeddale M, et al. Does transfusion practice affect mortality in critically ill patients? Transfusion Requirements in Critical Care (TRICC) Investigators and the Canadian Critical Care Trials Group. Am J Respir Crit Care Med. 1997;155:1618–1623.
12.Strate LL, Naumann CR. The role of colonoscopy and radiological procedures in the management of acute lower intestinal bleeding. Clin Gastroenterol Hepatol. 2010;8:333–343; quiz e44.
13.Strate LL, Orav EJ, Syngal S. Early predictors of severity in acute lower intestinal tract bleeding. Arch Intern Med. 2003;163:838–843.
14.Green BT, Rockey DC, Portwood G, et al. Urgent colonoscopy for evaluation and management of acute lower gastrointestinal hemorrhage: a randomized controlled trial. Am J Gastroenterol. 2005;100:2395–2402.
15.Rockey DC. Lower gastrointestinal bleeding. Gastroenterology. 2006;130:165–171.
16.Abbas SM, Bissett IP, Holden A, et al. Clinical variables associated with positive angiographic localization of lower gastrointestinal bleeding. ANZ J Surg. 2005;75:953–957.
17.Junquera F, Quiroga S, Saperas E, et al. Accuracy of helical computed tomographic angiography for the diagnosis of colonic angiodysplasia. Gastroenterology. 2000;119:293–299.
18.van Leerdam ME, Vreeburg EM, Rauws EA, et al. Acute upper GI bleeding: did anything change? Time trend analysis of incidence and outcome of acute upper GI bleeding between 1993/1994 and 2000. Am J Gastroenterol. 2003;98:1494–1499.
19.Barkun A, Bardou M, Marshall JK. Consensus recommendations for managing patients with nonvariceal upper gastrointestinal bleeding. Ann Intern Med. 2003;139:843–857.
20.Viviane A, Alan BN. Estimates of costs of hospital stay for variceal and nonvariceal upper gastrointestinal bleeding in the United States. Value Health. 2008;11:1–3.
21.Leontiadis GI, Sharma VK, Howden CW. Proton pump inhibitor therapy for peptic ulcer bleeding: cochrane collaboration meta-analysis of randomized controlled trials. Mayo Clin Proc. 2007;82:286–296.
22.Lau JY, Leung WK, Wu JC, et al. Omeprazole before endoscopy in patients with gastrointestinal bleeding. N Engl J Med. 2007;356:1631–1640.
23.Barkun AN, Bardou M, Martel M, et al. Prokinetics in acute upper GI bleeding: a meta-analysis. Gastrointest Endosc. 2010;72:1138–1145.
24.Stanley AJ, Ashley D, Dalton HR, et al. Outpatient management of patients with low-risk upper-gastrointestinal haemorrhage: multicentre validation and prospective evaluation. Lancet. 2009;373:42–47.
25.Greenspoon J, Barkun A, Bardou M, et al. Management of patients with nonvariceal upper gastrointestinal bleeding. Clin Gastroenterol Hepatol. 2012;10:234–239.
26.Garcia-Tsao G, Sanyal AJ, Grace ND, et al. Prevention and management of gastroesophageal varices and variceal hemorrhage in cirrhosis. Am J Gastroenterol. 2007;102:2086–2102.
27.Child CG, Turcotte JG. Surgery and portal hypertension. Major Probl Clin Surg. 1964;1:1–85.
28.Garcia-Pagan JC, Bosch J. Endoscopic band ligation in the treatment of portal hypertension. Nat Clin Pract Gastroenterol Hepatol. 2005;2:526–535.