Upper GI bleeding: a clinical prediction rule can help identify patients at risk of rebleeding or dying.

Clinical bottom line (level 1a)

  1. Patients with an upper GI bleed can be ranked using a clinical prediction rule for their risk of dying or rebleeding.
  2. The clinical prediction rule uses age, degree of shock on admission, comorbidity, diagnosis and major stigmata of haemorrhage on endoscopy.
  3. Less than 1% of patients with a score of 0 die, and around 5% have a rebleed.
  4. A third of patients with a score of 5 or more die, and a third rebleed.
Rockall et al: Gut 1996; 38: 316-321
Rockall et al: British Medical Journal 1995; 311: 222-226
Expires October 2002

The study

Inception cohort study with objective outcomes, adjusted for confounding factors, validated in an independent set of patients.

Setting: 74 acute hospitals, UK

4185 patients (aged 54 to 80; mean 66, 57% male) with clinical evidence of acute upper GI bleeding (haematemesis or melaena) or a history of bleeding within the last 10 days (84% emergency admissions)

Excluded if
  • aged < 16
  • GI bleed in hospital
  • underwent endoscopy for another reason



    • Factors studied:
  • age, degree of shock on presentation, comorbidity, diagnosis, major stigmata of recent haemorrhage




    • Logistic regression analysis was used to identify the risk factors that predicted mortality and rebleeding.

    ?100% followed for hospital stay
    Outcomes studied:
  • death
  • death due to further haemorrhage
  • death if admitted as an emergency
  • death if bleed as an inpatient
  • death if pre-endoscopy score: 0
  • death if pre-endoscopy score: 1
  • death if pre-endoscopy score: 2
  • death if pre-endoscopy score: 3
  • death if pre-endoscopy score: 4
  • death if pre-endoscopy score: 5
  • death if pre-endoscopy score: 6
  • death if pre-endoscopy score: 7
  • death if post-endoscopy score 0 to 2
  • death if post-endoscopy score: 3
  • death if post-endoscopy score: 4
  • death if post-endoscopy score: 5
  • death if post-endoscopy score: 6
  • death if post-endoscopy score: 7
  • death if post-endoscopy score: 8+
  • further bleeding continued bleeding requiring surgery or rebleeding
  • rebleeding if post-endoscopy score 0 to 2
  • rebleeding if post-endoscopy score: 3
  • rebleeding if post-endoscopy score: 4
  • rebleeding if post-endoscopy score: 5
  • rebleeding if post-endoscopy score: 6
  • rebleeding if post-endoscopy score: 7
  • rebleeding if post-endoscopy score: 8+

    • Clinical prediction rule: sum the score for each factor. 2 scores can be derived: pre and post-endoscopy.
    • age
      • < 60: (0)
      • 60 to 79: (1)
      • >79: (2)
    • shock
      • 'no shock': systolic bp = 100, pulse <100: (0)
      • 'tachycardia': systolic bp = 100, pulse more than 100 beats/min: (1)
      • 'hypotension': systolic bp < 100: (2)
    • comorbidity
      • no major co-morbidity: (0)
      • cardiac failure, sachemic heart disease, any major co-morbidity: (2)
      • renal failure, liver failure, disseminated malignancy: (3)
    • diagnosis
      • Mallory-Weiss tear, no lesion identified and no stigmata of recent haemorrhage: (0)
      • all other diagnoses: (1)
      • malignancy of upper GI tract: (2)
    • major stigmata of recent haemorrhage
      • none or dark spot seen: (0)
      • blood in upper GI tract, adherent clot, visible or spurting vessel: (2)
    • The clinical prediction rule was validated in an independent set of 1625 patients.

    The evidence

    outcome time to outcome number of patients/total number %
    (95% CI)    		 NNF
    (95% CI)
    death hospital stay 585/4142 14.12%
    (13.06% to 15.18%)    		 7
    (7 to 8)
    death due to further haemorrhage hospital stay 272/4142 6.57%
    (5.81% to 7.32%)    		 15
    (14 to 17)
    death if admitted as an emergency hospital stay 365/3472 10.51%
    (9.49% to 11.53%)    		 10
    (9 to 11)
    death if bleed as an inpatient hospital stay 191/578 33.04%
    (29.21% to 36.88%)    		 3
    (3 to 3)
    death if pre-endoscopy score: 0 hospital stay 0/246 0.0%
    (0.0% to 0%)
    death if pre-endoscopy score: 1 hospital stay 6/201 2.99%
    (0.63% to 5.34%)    		 34
    (19 to 158)
    death if pre-endoscopy score: 2 hospital stay 14/249 5.62%
    (2.76% to 8.48%)    		 18
    (12 to 36)
    death if pre-endoscopy score: 3 hospital stay 38/311 12.22%
    (8.57% to 15.86%)    		 8
    (6 to 12)
    death if pre-endoscopy score: 4 hospital stay 77/364 21.15%
    (16.95% to 25.35%)    		 5
    (4 to 6)
    death if pre-endoscopy score: 5 hospital stay 47/134 35.07%
    (26.99% to 43.15%)    		 3
    (2 to 4)
    death if pre-endoscopy score: 6 hospital stay 42/68 61.76%
    (50.21% to 73.32%)    		 2
    (1 to 2)
    death if pre-endoscopy score: 7 hospital stay 6/8 75%
    (44.99% to 105.01%)    		 1
    (1 to 2)
    death if post-endoscopy score 0 to 2 hospital stay 0/321 0.0%
    (0.0% to 0%)
    death if post-endoscopy score: 3 hospital stay 3/162 1.85%
    (-0.224% to 3.93%)    		 54
    (-446 to 25)
    death if post-endoscopy score: 4 hospital stay 14/176 7.95%
    (3.95% to 11.95%)    		 13
    (8 to 25)
    death if post-endoscopy score: 5 hospital stay 21/199 10.55%
    (6.28% to 14.82%)    		 9
    (7 to 16)
    death if post-endoscopy score: 6 hospital stay 16/137 11.68%
    (6.3% to 17.06%)    		 9
    (6 to 16)
    death if post-endoscopy score: 7 hospital stay 24/106 22.64%
    (14.67% to 30.61%)    		 4
    (3 to 7)
    death if post-endoscopy score: 8+ hospital stay 36/89 40.45%
    (30.25% to 50.65%)    		 2
    (2 to 3)
    further bleeding hospital stay 736/4142 17.77%
    (16.6% to 18.93%)    		 6
    (5 to 6)
    rebleeding if post-endoscopy score 0 to 2 hospital stay 19/321 5.92%
    (3.33% to 8.5%)    		 17
    (12 to 30)
    rebleeding if post-endoscopy score: 3 hospital stay 19/162 11.73%
    (6.77% to 16.68%)    		 9
    (6 to 15)
    rebleeding if post-endoscopy score: 4 hospital stay 27/176 15.34%
    (10.01% to 20.67%)    		 7
    (5 to 10)
    rebleeding if post-endoscopy score: 5 hospital stay 49/199 24.62%
    (18.63% to 30.61%)    		 4
    (3 to 5)
    rebleeding if post-endoscopy score: 6 hospital stay 37/137 27.01%
    (19.57% to 34.44%)    		 4
    (3 to 5)
    rebleeding if post-endoscopy score: 7 hospital stay 39/106 36.79%
    (27.61% to 45.97%)    		 3
    (2 to 4)
    rebleeding if post-endoscopy score: 8+ hospital stay 33/89 37.08%
    (27.04% to 47.11%)    		 3
    (2 to 4)

    Comments

    1. The data reported comes from the validation set.

    Citation

    1. Rockall TA, Logan RF, Devlin HB, et al: Risk assessment after acute upper gastrointestinal haemorrhage. Gut 1996; 38: 316-321
    2. Rockall TA, Logan RF, Devlin HB: incidence of and mortality from acute upper gastrointestinal haemorrhage in the United Kingdom. British Medical Journal 1995; 311: 222-226
    Contributor: Chris Ball and Musab Hayatli, October 1999
    Reviewer: Horand Meier

    Clinical Question.
    Patient upper GI bleed
    Intervention or Exposure clinical prediction rule
    Outcome rebleeding, death