Anticoagulation: bleeding is more likely with more co-morbid conditions and supratherapeutic levels.

Clinical bottom line (level 1a)

  1. Risk factors for bleeding on anticoagulation therapy include:
    • number of co-morbid conditions
    • using iv heparin in elderly patients
    • raised PT or aPTT levels
    • worsening liver function during therapy
  2. A fifth of in-patients ranked at high risk have a major bleed.
  3. No patients ranked at low-risk have a major bleed.
Landefeld et al: American Journal of Medicine 1990; 89: 569-577
Expires June 2003

The study

Inception cohort study with objective outcomes, adjusted for confounding factors, validated in an independent set of patients.
Setting: two tertiary medical centres, USA

1011 patients (aged ?, ?% male) in derivation and validation sets
Note:
  • Clinical prediction rule in an independent set of patients.
  • Proportional hazards multivariate analysis of risk factors. Following were found to be significant:
    • 1. Comorbid illness
    • serious cardiac illness (acute MI, hypotension- systolic <90 mmHg on admission or subsequent hypotension requiring aortic balloon pump)
    • liver dysfunction (hyperbiliaemia or macrocytosis)
    • renal insufficiency (Cr>1.5 mg/dl)
    • poor general condition (cancer or severe anaemia- Hct<30%)
    • 2. Worsening liver dysfunction if hyperbiliaemia at start of therapy and rose to = mg/dl without evidence of non-hepatic cause
  • Risk factors and point weight:
    • 1 specific co-morbid condition at start- 1 point
    • 2 specific co-morbid conditions- 2 points
    • 3+ specific co-morbid conditions- 3 points
    • 60-79 years old- 2 points
    • 80 years old or more- 4 points
    • maximum PT or aPTT ratio 2.0-2.9- 1 point
    • maximum ratio = 3.0- 2 points
    • worsening liver dysfunction during therapy- 2 points
    • total points possible- 11 points


    • Control Group: (n = 617, 617 analysed): Deviation set: retrospective study of patients in Boston started on anticoagulation therapy 1977-1983.
    Experimental Group: (n = 394, 394 analysed): Validation set: prospective study of patients in Cleveland started on anticoagulation therapy 1986-1989. Patients scored according to bleeding risk scale predicted from Boston study, and physicians made independent prediction of 'risk of major bleed during anticoagulation in hospital' in 316 patients. Gold standard- major bleed.
    Patients were anticoagulated with warfarin, iv heparin, subcutaneous heparin >10000 units/day, streptokinase or urokinase for >10 days.
    100% followed for 10 days

    The evidence

    prediction rule bleeding risk
    Outcome Time to outcome CER RR
    (95% CI)    		 NNF
    (95% CI)
    high (5 to 6) unknown 12/63
    (19%)    		 2.6
    (1.5 to 4.3)     		3
    (2 to 11)
    moderate (2 to 4) unknown 15/96
    (16%)    		 2.0
    (1.3 to 3.1)     		6
    (3 to 21)
    low (0 to 1) unknown 6/235
    (2.6%)    		 0.29
    (0.14 to 0.59)     		-54
    (-94 to -45)

    prediction rule bleeding risk with clinical prediction
    Outcome Time to outcome CER RR
    (95% CI)    		 NNF
    (95% CI)
    high (5 to 6) and clinical prediction moderate, high or very high unknown 11/48
    (23%)    		 3.1
    (2.0 to 4.9)     		2
    (1 to 4)
    moderate (2 to 4) and clinical prediction low unknown 6/61
    (9.8%)    		 1.1
    (0.58 to 2.3)     		102
    (-24 to 8)
    low (0 to 1) and clinical prediction low unknown 0/87
    (0.0%)    		 0.0
    (0.0 to 0.33)     		inf
    (30 to inf)

  • Boston study:
    • 28/617 patients had a major bleed, ie. 4.5 % (95% CI: 2.8% to 6.2%)
    • 1/617 died, ie. 0.16% (95% CI: 0.0% to 0.48%)
  • Cleveland study:
    • 33/394 patients had a major bleed, ie. 8.4% (95% CI: 5.6% to 11%)
    • 0/394 died
  • There was a significant difference in bleeding rate between hospitals (p<0.05).

    • Comments

    1. A number of patients were given thrombolytics- not clearly separated from other patients. Makes risk of bleeding much higher, and consequently the study is less useful.
    2. Fewer Cleveland patients were anticoagulated for cardiac surgery than Boston patients and more for venous thromboembolism. Fewer had liver dysfunction. More were on heparin and fewer on warfarin- ?effect on scoring.

    Citation

    1. Landefeld CS, McGuire E, Rosenblatt MW, et al: A bleeding risk index for estimating the probability of major bleeding in hospitalized patients starting anticoagulant therapy. American Journal of Medicine 1990; 89: 569-577
    Search Terms: heparin, warfarin and adverse effects
    Contributor: Chris Ball and Clare Wotton, June 2000
    Reviewer:

    Clinical Question.
    Patient on anticoagulation therapy
    Intervention or Exposure clinical prediction rule
    Outcome predicting major bleeds