Upper GI Bleeding

Introduction

• Any bleeding above ligament of Treitz
• Increasing age, concurrent illnesses and recurrent haemorrhage are associated with increased morbidity and mortality
• Melena is due to upper GI bleeds 70% of the time and lower GI in 20-30%
  • Need 150mL of blood in gut for at least 8 hours
  • Does not in itself carry poor prognosis

• Haematochezia in upper GI bleed carries 3x risk of death

Aetiology

• Peptic ulcer disease (35-50%)
• Erosive gastritis/oesophagitis/duodenitis – 15%
• Varices – 10%
• Mallory-Weiss tears 10-15%

Pathophysiology

• Peptic ulcer disease
  • Most common cause of upper GI bleeding
  • 20% of cases have gastric or duodenal ulcers on endoscopy
  • Decreasing incidence as smoking rates and NSAID use decline and H. pylori eradication is promoted
• Erosive gastritis and oesophagitis
  • Alcohol and NSAID’s, infection, toxic ingestion, radiation and stress ulcers
  • Candida, HSV, CMV and HIV are potential sources of oesophageal bleeding from infection
• Oesophageal and gastric varices
  • Cause of UGI bleeding in cirrhotic patients 59% of the time (peptic ulcer in 16%)
  • In-hospital mortality for any GI bleed in cirrhotics is 2x that of normal population
  • Child-Pugh classification is most important prognostic factor
• Mallory-weiss tear
  • Bleeding secondary to longitudinal tear at GOJ
  • Associated with binge drinking, DKA or chemotherapy
  • Only 1/3 of these patients have history of recurrent vomiting/retching
• Dieulafoy lesions
  • Arteries of GI tract that protrude through submucosa
  • Mostly lesser curvature of stomach and 95% within 6cm of GOJ
  • Intermittent massive GI bleeding and difficult to diagnose endoscopically
• Other
  • AV malformations, malignancy, masquerades (ENT bleeding), aortoenteric fistula from graft

Diagnosis

• Hx of aortic graft = Aortoenteric fistula until proven otherwise
  • Often have herald bleed then exsanguination
  • Often diagnosed on CT after herald bleed

• Bright red or maroon stools is upper GI 14% of the time
• Occult presentations with syncope, angina, hypotension, tachycardia, weakness, confusion or cardiac arrest

Examination

• Visually inspect vomitus
• Subtle reduced pulse pressure
• Paradoxical bradycardia may be seen with profound hypovolaemia
• Baseline hypertension and beta-blocker/CCB may masks signs
• Look for signs of liver disease
• Facial lesions, cutaneous macules or telangiectasias may suggest Peutz-Jeghers, Rendu-Osler-Weber or Gardner’s syndromes
• ENT for occult bleeding source
• Abdo
• PR

Labs

• FBC, Chem20, Coags, G&H, stool guaiac
• Urea:creatinine >30 suggests UGI bleed
• ECG
• Lactate
  • Single elevated lactate is a sentinal for severe illness
  • Rising levels is a clear predictor of in-hospital mortality

NG tube

• Can confirm diagnosis and stratify risk

• Aspirate may show blood
• High-risk lesions more likely in bloody aspirates
• Early NG is associated with decreased time to endoscopy
• Negative NG aspirate does not rule out UGI bleed
• Intermittent bleeding, pyloric spasm or oedema preventing reflux of duodenal blood

• NG aspirates positive in 23% of patients without haematemesis who have occult UGI bleeding
• Visual inspection is the most reliable test (vs. stool guaiac card use with UGI aspirates)
• No evidence that NG passage increases risk of bleeding in known varices

CXR

• Not necessary
• Indications
  • Suspected aspiration
  • Co-existent cardiorespiratory disease
  • Suspected perforation (only 70-80% sensitive)
  • Elderly
  • Very low risk
  • <60yo
  • No major comorbidities (esp. liver disease/varices)
  • No history of red haematemesis
  • No frequent melena
  • No haematochezia
  • Negative NG aspirate
  • Haemodynamically stable in ED
  • Normal lab studies (normal Hb)

Clinical scores

Rockall score

• Pre-endoscopy

• Age <60, 60-79, >80
• Degree of shock – None, HR >100, SBP <100
• Comorbidities – None, any, renal/liver/disseminated malignancy

• In some studies, score of 0 is very low risk of mortality but in others was not consistent
• If not low risk, consider urgent endoscopy and ICU
• Blatchford score of 0 probably a better indicator of low risk patients suitable for outpatient care (<1% chance of requiring intervention)

Glasgow-Blatchford Score

• Hb
• BUN
• SBP
• Sex
• HR >100
• Melena
• Recent syncope
• Hepatic disease
• Cardiac failure
• Low risk = score 0
• Any score >0 is high-risk for intervention, transfusion, endoscopy or surgery
• Stanley et al. confirmed score of 1 also low-risk

Modified Blatchford

• Only pulse, BP, BUN and Hb
• Performs as well as original

AIMS65

• Albumin < 30
• INR > 1.5
• Altered mental status
• SBP <90
• Age >65
• Superior to Blatchford in predicting mortality, but Blatchford superior in predicting transfusion
• Superior in predicting in-hospital mortality, ICU admission and LOS
• Blatchford better for predicting low risk and suitable outpatient care

Treatment

Medical treatment

• Transfuse
  • Liberal transfusion to >90 can cause harm
  • Restrictive targeting >70 with target of >90 for older patients with comorbidites who are not tolerating acute anaemia is recommended
• Reverse anticoagulants if able
  • INR >1.5 is significant predictor of mortality in UGI bleed on anticoagulants
• Target Plt >50

• Tranexamic acid has shown no benefit in UGI bleed
  • Cochrane review (2014) showed may reduce mortality, may reduce re-bleeding and advised only to use in sickest patients
  • HALT-IT trial showed no mortality benefit at 5 days
• PPI 80mg bolus then 8mg/hr
  • Allows platelet aggregation (pH >6.0), reduces need for surgery, length of stay and signs of bleeding
  • Reduces likelihood of bleeding or need for intervention during endoscopy
  • Reduces re-bleeding risk and blood transfusion
  • Does not improve mortality
  • Up to 81% of patients with known varices have an alternative source of bleeding at endoscopy so PPI is crucial
• Octreotide
  • Long-acting somatostatin analogue inhibit gastric acid secretion, reduces blood flow to gastroduodenal mucosa and causes splanchnic vasoconstriction
  • 50mcg bolus then 25-50mcg/hr
  • No clear benefit on mortality
  • Decreases risk of persistent bleeding and rebleeding in peptic ulcer disease and can be considered while awaiting endoscopy
  • Terlipressin has been shown to reduce mortality in one study
  • Head-to-head vs. octreotride shown to be non-inferior with no reduction in in-hospital mortality (NEJM)
• Antibiotics
  • Cirrhotic patients have increased risk of gut bacterial translocation during bleeding episodes
  • Most common infections are UTI, SBP, bacteraemia and pneumonia
  • 20% of cirrhotic upper GI bleeders have infection at presentation and another 50% develop infection while hospitalised
  • Cipro 400mg IV or Ceftriaxone 1g IV reduce infectious complications, reduce mortality and should be started as soon as possible
  • Antibiotics
    • Reduces (Cochrane)
      • Infectious complications (RR 0.40)
      • All-cause mortality (RR 0.75)
      • Bacterial infection mortality (RR 0.43
      • Rebleeding (RR 0.53)
      • Length of stay (-2 days)
• Erythromycin and metoclopramide are promotility agents used prior to endoscopy for improved visualisation

Endoscopy

• Early endoscopy (<24 hours) is recommended for most patients as is associated with a significant cost reduction and decreased LOS
• Even earlier endoscopy (<6 hours) has shown reduced hospital LOS but increased mortality for inadequately resuscitated patients
• Lau et al. performed a randomised trial in a single-centre in Hong Kong to evaluate whether endoscopy within 6 hours vs. 6-24 hours from initial Gastroenterologist Consultation improved outcomes in patients at high-risk of further bleeding or death
  • Patients were over 18 and had Glasgow-Blatchford scores of 12 or higher
  • Patients who remained hypotensive despite initial resuscitation attempts were excluded
  • 516 patients were enrolled with only 10% of patients in the urgent (<6hrs) group and 7% in the early group having variceal bleeding
  • Only 28% of patients in the urgent group actually had endoscopy within 6 hours (mean 9.9 hrs) vs. 24.7 hrs in the early group
  • All-cause mortality at 30 days, further bleeding episodes, hospital LOS, transfusion and surgery/embolisation did not differ between groups
  • 8% of patients in the early group underwent emergent endoscopy for recurrent bleeding
• Endoscopy findings
  • Oozing ulcer site = 85-90% bleeding recurrence
  • Protuberant vessel = 35-55%
  • Adherent clot = 30-40%
  • Flat spot = 5-10%
  • None = 5%
• Proximal posteroinferior wall of duodenal bulb and high lesser curvature of stomach are sites of particularly high risk of heavy bleeding due to underlying arteries

• Completed Rockall score
  • Can be used after completed endoscopy to predict mortality
  • Helpful to identify patients that warrant intensive care and are at risk of death
  • Less helpful in predicting low-risk outpatient candidates
  • Criteria
    • Age <60, 60-79 or >80
    • Degree of shock
    • Comorbidities
    • Diagnosis on endoscopy (Mallory-Weiss and no lesion/stigmata lowest risk)
    • Stigmata of haemorrhage on endoscopy
• Varices
  • Endoscopic sclerotherapy effective for 80-90% of acute variceal bleeding
  • Gastric varices are more difficult to sclerose
  • Variceal ligation (banding) is an alternative with less complications

Balloon tamponade

• Effective short-term solution for life-threatening bleeding
• Temporary stabilisation
• Sengstaken-Blakemore tube and Minnesota tube (added oesophageal suction port above oesophageal balloon)
• Risk of mucosal ulceration, oesophageal/gastric rupture, asphyxiation from tracheal compression and aspiration
• Should be intubated prior
• Gastric balloon inflated first with 250-350mL of water but must be certain in the stomach
• If bleeding does not stop, inflate oesophageal balloon with manometer reading <50mmHg
• Confirm with X-ray
• Apply 1kg traction by attaching distal end of tubing to 1L bag of saline hung off Iv pole
• Duration of use should be <24 hours

Transjugular intrahepatic portosystemic shunt (TIPS)

• 90% success rate
• Reduces portal pressure and variceal bleeding
• Intra-abdominal haemorrhage and stent occlusion are important complications
• Hepatic encephalopathy occurs in 25-60% of patients

Surgery

• For refractory bleeding to medical/endoscopic therapy or suspicion of perforation
• If variceal:
  • TIPS and non-shunt oesophageal transection/GOJ devascularisation
• In non-variceal bleeding:
  • Percutaneous embolisation or subtotal/total gastrectomy 

Surgery for variceal bleeding

• Overall mortality 15-20%
• Reserved for failed endoscopic +- TIPS
• Staple transection of oesophagus and portocaval shunt formation are options
• Hepatic encephalopathy is a major complication of shunt surgery and future liver transplant is no longer possible after this

Acute stress ulcers

• Mostly gastric fundus likely due to mucosal hypoxia and hypoperfusion
• 8-45% of ICU patients
• Acid suppression therapy reduces risk of this and GI haemorrhage
• Risk of gastric bacterial overgrowth and risk of nosocomial pneumonia

Last Updated on December 29, 2021 by Andrew Crofton