EM guidemap - Upper GI bleed

Click on any of the headings or subheadings to rapidly navigate to the relevant section of the guidemap

Introduction

Definitions and general principles

History of the present illness

Examination

Diagnostic testing

ED management Appendix
Introduction

- this guidemap is focused on the ED management of an upper GI bleed - note that there is a seperate guidemap that deals with the clinical problem of a lower GI bleed in the ED setting

- this guidemap is mainly focused on the ED management phase, and it does not discuss any disease-entities or supply detailed information about definitive in-hospital management (including the further investigation of endoscopy-negative upper GI bleeds)

- this guidemap merely offers a conceptual approach to the immediate ED management of an upper GI bleed, and it also offers basic guidance on the need for surgical consultation and the need for ICU admission

- this guidemap is mainly focused on the management of large upper GI bleeds that present to the ED with hematemesis and/or melena +/- hemodynamic instability

Definitions and general principles

- an upper GI bleed is defined as a bleed that originates from a gastro-intestinal source that is proximal to the ligament of Treitz (distal duodenum)

- the clinical presentation of an upper GI bleed mainly depends on the rate of bleeding, and whether the patient actively vomits blood or whether his bleed manifests solely as melena or hematochezia

- a slow upper GI bleed may merely present with a progressive anemia + heme-occult positive stools

- melena (black, tarry foul-smelling stool) usually signifies the presence of >50ml of blood and it usually implies a GI source proximal to the cecum (the black color is due to the effect of bacteria on blood as it passes through the colon - it takes > 12 hours of transit time for melena to develop)

(* melena can rarely develop as a result of bleeding from right colonic sources => always exclude an upper GI bleed before pursuing proximal colonic pathology)

- hematochezia (maroon bloody stools) can come from a proximal GI source if intestinal transit times are fast, but is usually signifies a lower GI source, and it usually implies a volume of blood of >200ml

- hematemesis is bloody vomitus and it may be fresh bright-red blood or older coffee-ground material, and it signifies an upper GIT bleed

- hematemesis accompanies melena in ~ 50% of upper GIT bleeds, and it usually signifies a larger, brisker bleed

- a patient may occasionally present without any objective signs of GI blood loss, and the patient may merely present with symptoms of hypovolemia (weakness, lightheadedness or syncope, dyspnea or angina if the patient has underlying CAD)

- common causes of upper GI bleeding in adults include PUD (50% of adult cases), esophageal varices (10-15% of adult cases), gastritis (10% of adult cases) and Mallory-Weiss tears (5-10% of adult cases)

- other common causes in an adult (> 1% incidence) include:- esophagitis, neoplasm, marginal ulcer, or vascular ectasia (upper gastrointestinal angiodysplasia or angiomas are mostly idiopathic, but may also be found as part of the Osler-Weber-Rendu syndrome and in patients with chronic renal failure, aortic stenosis, CREST and von Willebrands disease)

- rare causes in an adult (< 1% incidence) include:- Dieulafoy's erosion (a large submucosal artery that protrudes through the gastric mucosa and that is unassociated with a gastric ulcer), hemobilia, connective tissue disorders, aorto-enteric fistula, mesenteric ischemia, radiation gastritis, gastric volvulus and blood dyscrasias

(* see the appendix for common causes of upper GI bleeding in pediatric patients based on age)

- the mortality of acute upper GI bleeding has remained unchanged for the past 50 years (~ 7 - 10% mortality) and there is a much higher mortality in patients, who have bright red blood in the gastric aspirate

History of the present illness

- if the patient has a history of an acute hematemesis, obtain information on the amount of blood and the duration of bleeding

- inquire about any history of melena and symptoms of hypovolemia (weakness, lightheadedness or syncope, dyspnea or angina if the patient has underlying CAD)

- any associated upper abdominal pain suggests PUD, or rarely gastritis, or uncommon pathology (pancreatic or biliary pathology eroding into the upper GIT, aortic-enteric fistula, gastric malignancy)

- recent dyspepsia or heartburn suggests esophagitis or gastritis or PUD

- recent violent retching or vigorous coughing suggests a possible Mallory-Weiss tear, especially if the patient has liberally consumed alcohol

- a previous history of upper GI bleeds is found in patients with esophageal varices and PUD (60% of rebleeding is from the same site)

- inquire about any underlying PUD, liver disease and esophageal varices, gastric malignancy or any bleeding diathesis (hematological disorder, anticoagulants, end stage liver disease)

- inquire about previous aortic graft surgery (recent fever and abdominal tenderness suggests an aorto-enteric fistula)

- a history of pancreatitis suggests the possibility of rupture of a peri-pancreatic pseudoaneurysm or vein into a pseudopancreatic cyst and secondary erosion into the GIT (or simple leakage via the pancreatic duct into the duodenum)

- inquire about recent nosebleeds, which can also subsequently present with coffee-ground vomitus or hematemesis

- a personal or family history of recurrent nosebleeds suggests Osler-Weber-Rendu syndrome (hereditary hemorrhagic telangiectasia)

- inquire about heavy alcohol consumption (causes alcoholic gastropathy, gastritis, PUD and cirrhosis with secondary esophageal varices) or recent aspirin/NSIAD use (causes gastritis and PUD)

Examination

- first determine the patient's vital signs

- check the nose and throat for evidence of a recent upper airway bleed

- check the skin for cutaneous manifestation of cancer (acanthosis nigricans), bleeding diathesis (bruising or petechia), connective tissue disease (Ehlers-Danlos syndrome) or genetic diseases (Peutz-Jeghers disease, Osler-Rendu-Weber syndrome) or end-stage liver disease (spider nevi, palmar erythema, caput medusae)

- check for jaundice (cirrhosis, or hemobilia = combination of jaundice + abdominal pain + blood-stained bile)

- check the abdomen for tenderness (PUD perforation, pancreatitis, hemobilia), or masses (enlarged liver, gastric or pancreatic masses), or pulsations (AAA), or bruits (AAA, aorto-enteric fistulas), or hyperactive bowel sounds (large upper GI bleed) or ascites (end-stage liver disease) or testicular atrophy  (end-stage liver disease) or hepatosplenomegaly (liver disease and portal hypertension)

- perform a rectal exam looking for evidence of melena, hematochezia or heme-occult positive stools

Diagnostic testing

Nasogastric aspiration

- a NGT should be inserted in all patients who present with hematemesis or melena or hematochezia

- the NGT should be inserted into the stomach and manually aspirated

(* continuous suction is contra-indicated because it may cause gastric mucosal injury if the stomach wall is constantly sucked against the catheter tip; those same mucosal injuries may also confound the endoscopic determination of the cause of the GI bleeding)

- the presence of grossly visible blood or coffee-grounds in the gastric aspirate confirms an upper GI bleed

(* there is no reason to perform a heme occult test on marginally dark gastric fluid to see if it is heme positive, because the test is inaccurate and the result does not affect clinical management => see the outpatient management of an upper GIT bleed)

- a negative gastric aspirate does not exclude an upper GI bleed - the gastric aspirate could be negative because the bleeding is intermittent; also, >10% of bleeding duodenal ulcers have a negative gastric aspirate because the blood does not reflux back into the stomach

(* aspiration of non-bloody bile virtually excludes a duodenal bleed, although clinicians have difficulty deciding whether aspirated gastric fluid definitely contains bile based on a visual inspection of the aspirated fluid)

- the NGT should be promptly removed if there is no evidence of active bleeding - leaving the tube in situ for a prolonged period can cause mucosal lesions and is uncomfortable for the patient

(* insertion of a NGT is not contra-indicated in the presence of known esophageal varices or a suspected Mallory-Weiss tear)

- if the patient is actively bleeding => a larger tube (34F) should be used for tap water irrigation in order to remove large clots (the patient should be placed in the left lateral decubitus and Trendelenburg position)

(* tap water irrigation should not be performed without first confirming correct NGT placement on an upright chest X-ray; remember that the absence of free air on an upright chest x-ray may also help exclude a gastric perforation, which is a contra-indication to gastric lavage; tap water should be used in 100 - 200cc aliquots until the aspirate clears => continued bleeding suggests a massive bleed requiring emergency gastro-enterology consultation + emergency endoscopy +/- emergency surgery; ice water lavage does not help stop the bleeding and it lowers the core temperature and it is therefore not indicated; there is no advantage to using sterile water or normal saline solution when performing gastric lavage)

Chest X-ray

- has no diagnostic utility in the absence of any clinical suspicion of pulmonary aspiration, or a gastric perforation, or a large full-thickness Mallory-Weiss tear causing secondary pleural space soiling

Abdominal X-ray

- has no clinical utility in the diagnostic evaluation of an upper GI bleed due to PUD, gastritis, esophageal varices or esophagitis

BUN

-  a BUN level > 40 in the presence of a normal serum creatinine is suggestive of an upper GI bleed

Hematocrit

- the absolute level does not correlate with the presence, or degree, of an upper GI bleed

- the hematocrit can be normal despite overt evidence of hypovolemic shock, because it takes many hours for the hematocrit to stabilize following an acute bleed

- a low initial hemoglobin could be due to occult GI bleeding preceding the ED visit, or it could be due to an anemia that is unrelated to the GI bleed eg. chronic anemia secondary to liver disease or renal failure

- the final hemoglobin level depends on the amount of continued bleeding, but it is also affected by the volume of crystalloid fluid +/- packed cells infused during that same time period

- serial hemoglobin levels are used to roughly quantify the degree of active bleeding and the adequacy of red blood cell replacement

Coagulation studies

- a PT and PTT and platelet count are routinely indicated, especially if the patient has a history of liver disease, a known coagulopathy or anticoagulant use

Blood typing

- a type and screen should routinely be performed

- a type-and-crossmatch should be performed if there is any evidence of significant bleeding or hemodynamic instability => the number of ordered units depends on the degree of bleeding and the hemodynamic status of the patient (usually 2 - 4 units in an adult)

Liver function tests

- indicative of baseline liver function in patients with liver disease and should routinely be ordered

ECG

- should be performed in all elderly patients with a significant GIT bleed because there is a high incidence of silent AMIs secondary to hypovolemic shock

- any ischemic ECG changes warrants patient admission to an ICU to exclude secondary cardiac ischemia

ED management 

Fluid recuscitation

- first ensure hemodynamic stability by normal saline fluid administration

- fluid administration is tailored to the patient's needs; caution is advisable in elderly patients at risk of CHF, or in renal failure patients and in patients with cirrhosis; vasopressors are contra-indicated in hypovolemic patients

- packed cells should be administered if > 30cc/kg of normal saline is required to acutely recuscitate a hypovolemic patient; the threshold should be lower if the patient is obviously anemic

- the requirement for packed cells depends on the initial hemoglobin level + rate of bleeding + presence of underlying ischemic heart disease => the "target" hemoglobin level should be flexible and ~ 10 g/dl in elderly patients with ischemic heart disease and poor cardiovascular reserve, and 6 - 8 g/dl in young, previously healthy patients

(* rigid guidelines for the administration of blood products should be avoided => use your clinical judgement => the transfusion requirment should be based on the:- i) patient's age, ii) presence of co-morbidities, iii) patient's cardiovascular reserve, iv) baseline hematocrit, v) rate of bleeding and vi) clinical efficacy of therapy

- serial vital signs + serial hemoglobin levels are the best indicators of the balance between ongoing bleeding and ongoing fluid recuscitation

Correct any coagulopathy

- correct any coagulopathy prn with 10 - 15 ml/kg of FFP (if PT > 1.5) and/or platelet transfusions (if platelet count < 50,000/cu.mm)

(* avoid giving FFP and platelets based on an empirical formula relating to the number of units of transfused packed cells; transfusion decisions should also be affected by the presence of other coagulopathies eg. DIC, or the presence of qualitative platelet defects induced by renal failure or aspirin/NSIADs)

- do not empirically give large doses of vitamin K (10mg) to patients taking warfarin for critically important reasons (eg. mechanical heart valve) because the patient will become resistant to coumadin for an extended period of time => use FFP to temporarily correct the anticoagulant-induced coagulopathy during the acute bleeding period => prothromin complexes are only used if FFP is not effective

(* small doses of sc vitamin K - 1mg - may be acceptable)

Emergency endoscopy

- patients with i) active upper GI bleeding manifested by hematemesis + hemodynamic instability, ii) massive upper GI bleeding, or iii) suspected esophageal variceal bleeding should undergo emergency endoscopy after initiating fluid and medical  recuscitation

- gastric lavage with room temperature tap water should be used to clear the stomach prn prior to endoscopy

- endoscopy may need to be performed in the operating room if there is massive bleeding and life-threatening shock => emergency surgery may be necessary

- endotracheal intubation may be required to protect the airway from aspiration if the bleeding is massive or if the patient has significantly altered mental status => temporarily delay upper endoscopy until airway management and hemodynamic stability is secured

- endoscopy should also be performed promptly if the patient has a history of a previous abdominal aortic aneurysm repair - in order to exclude the possibility of an aorto-enteric fistula (70% are found in the duodenum and 90% present with a herald bleed prior to a massive bleed)

(*failure to recognize a "herald" bleed may result in death from a subsequent sudden exanguinating bleed)

- endoscopy can be temporarily delayed if the non-variceal bleeding patient (who is also not at risk of bleeding from an aorto-enteric fistula) has self-limited bleeding => diagnostic accuracy is not altered if endoscopy is performed within 24 hours; middle-of-the-night diagnostic endoscopy should be avoided in stable patients with self-limited bleeding if well-trained personnel are not readily available

- endoscopy is successful in making the diagnosis in 80 - 95% of patients with an upper GI bleed if it is performed within 24 hours of a recent bleed => 10 % of patients have a negative upper endoscopy study because i) the bleeding may have ceased and the lesion may have healed rapidly eg. Mallory-Weiss tear, ii) the true source of the bleeding was from an epistaxis or hemoptysis, iii) the bleeding lesion is technically difficult to see eg. Dieulafoy's lesion, iv) the bleeding is distal to the proximal duodenum  and inaccesible to upper endoscopy, or v) the story/evidence of upper GI bleeding was fabricated eg. Munchausens syndrome with blood from another source

(* there is no role for diagnostic upper GI radiography in patients with upper GI bleeds)

Surgical consultation

- indications for surgical consultation include:-

Empiric acid suppression therapy

- there is some evidence that empiric administration of H2 antagonists (eg. ranatidine) may decrease the rate of re-bleeding in non-variceal bleeding patients with ulcers

- continuous IV infusion may be better than bolus administration

- acid suppression using omeprazole is an alternative treatment that is gaining favor

- there is no need to administer these agents in the ED => defer the decision to the admitting physician

Vasopressin and octreotide acetate

- vasopressin has only been shown to be beneficial for esophageal variceal bleeding, and it should not be used empirically for upper GI bleeding of undetermined cause

- vasopressin is less effective than endoscopic therapy, but it can be initiated as a stop-gap measure until endoscopy can be performed (or if endoscopy is unsuccessful)

- vasopressin can be given via selective arterial infusion (eg. superior mesenteric artery) or via a central venous line => the simpler IV route is preferred because there is no significant difference in effect between intra-arterial and intravenous infusions

- a standard mixture of vasopressin is 100 units in 250 mg 5DW (0.4 units/ml) => starting dose of 0.3 units/min for 30 minutes followed by increments of 0.3 unit/min until hemostasis is achieved, or side-effects develop, or a maximum dose of 0.9 units/min is reached

- vasopressin has significant side-effects (myocardial ischemia and infarction, mesenteric ischemia or infarction, cerebral ischemia, cutaneous ischemic necrosis and acrocyanosis) => vasopressin should be used with extreme caution in patients with underlying CAD or vascular disease => a concomitant NTG infusion is often administered if vasopressin must be utilized in CAD patients (presuming that the systolic blood pressure is > 100mmHg)

- octreotide acetate, a long-acting synthetic analog of somastain, reduces portal hypertension in patients with an acute variceal bleed

- octreotide is as effective as vasopressin, but it has fewer side-effects (given as a 50 - 100 mcg IV bolus and then infused at 25 - 50 mcg/hour) => it is now the preferred agent to induce splanchnic vasocontriction and reduce portal venous pressure in patients with a variceal bleed

- octreotide can be used in addition to endoscopic therapy of esophageal varices, or alone if endoscopy is unavailable or unsuccessful

Balloon tamponade

- effective in temporarily stopping bleeding from esophageal varices; can be used prior to endoscopy as a stop-gap measure or after failed therapeutic endoscopy

- it is associated with a high rate of complications

- accurate balloon placement requires special expertise => it has therefore fallen into disfavor and is less frequently utilized

- should never be inserted blindly by inexperienced personnel

Shunt surgery for variceal bleeding

- may be required if endoscopic and medical therapy fails to control an acute esophago-gastric variceal bleed

- transjugular intrahepatic portosystemic shunt (TIPS) is a radiologic alternative to a surgical shunt and it is now used more frequently - especially in patients who are poor surgical candidates or who are awaiting liver transplantation

Arteriography

- arteriographic embolisation therapy is used selectively - if bleeding is so massive that endoscopic evaluation and therapy is not feasible, or if vascular anomalies or hemobilia are the source of the bleeding

- transcatheter angiographic embolization of the bleeding artery responsible for an ulcer hemorrhage may be indicated if endoscopic therapy fails and the patient is a poor operative candidate => embolization is successful in ~ 50% of cases

(* there is a only small risk of iatrogenic GI infarction following embolization therapy of proximal bleeds because of the good blood supply)

Emergency surgery

- may be required if ulcer bleeding is unresponsive to endoscopic therapy and/or arteriographic embolization therapy

- also required if endoscopy reveals a non-ulcer bleeding site in the duodenum in a patient with a previous history of aortic aneurysm surgery

- rarely required for ongoing bleeding from rare entities such as angiodysplasia or Dieulafoy's lesion

Outpatient management

- if a patient presents to the ED with a history of hematemesis, but has no clinical evidence of active bleeding + the nasogastric aspirate is negative (or equivocally positive for coffee-ground material) => observe the patient for 6 hours => it may be acceptable to discharge the patient if there is no evidence of GI bleeding in the ED and the following criteriae are fulfilled:-

Appendix

Clinical risk factors suggesting a high probability of a poor outcome in patients with non-variceal bleeds => consider ICU admission

Some definite ICU admission criteriae
Risk of re-bleeding, need for surgery and mortality rate based on the endoscopic appearance of an ulcer
Ulcer appearance
 Rebleeding risk Need for surgery Mortality rate
Clean base 5% 0.5% 2%
Flat spot 10% 6% 3%
Adherent clot 22% 10% 7%
Visible vessel 43% 34% 11%
Active bleeding 55% 35% 11%

Hemobilia

- presence of blood in the bile and is clinically associated with the triad of jaundice and biliary colic and an upper GIT bleed in < 50% of cases

- due to diseases that cause a communication between the biliary tree and the vascular system

- the diagnosis is made endoscopically when blood is seen coming from the ampulla of Vater and/or by angiography

- arterial embolization is the treatment of choice => hepatic artery ligation if embolization therapy fails

Common causes of an upper GI bleed in pediatric patients - based on age

Infant (< 3 months)

Toddler (< 2 years) Pre-schooler (< 5 years) School-age child (> 5 years) Rare causes of upper GI bleeding in pediatric patients Disclaimer: My EM guidemaps reflect my personal approach to problem-solving/managing clinical cases in an ED setting and they should not be regarded as the standard of care. They merely represent the personal opinions of the author and they should only be used in clinical practice if the reader-user has substantial reason to believe that the clinical advice contained in the guidemaps is valid and accurate. The guidemaps are not meant to be "authoritative" and the reader-user should consult standard medical textbooks and expert opinion articles/guidelines for more authoritative advice. The reader-user should particularly confirm all drug doses, their indications and contra-indications, prior to their use.