Anticoagulation: bleeding occurred more often in patients with comorbid disease and commonly from the GI tract.

Clinical bottom line (level 2a)

  1. The most common sites of bleeding for patients on heparin or warfarin were:
    • gastrointestinal tract
    • urinary tract
    • soft tissues
    • nasopharynx
  2. The duration of anticoagulant therapy was a critical determinant of the overall risk for bleeding- the highest risk occurs during the first month of therapy with warfarin.
  3. The type and severity of comorbid illness were the most important risk factors for bleeding on heparin or warfarin. Cerebrovascular disease, serious heart disease, liver dysfunction and renal insufficiency were associated with an increased risk for anticoagulant-related bleeding. Age had not been clearly established as a risk factor.
  4. Concurrent intake of non-steroidal anti-inflammatory agents, especially aspirin, increased the risk of bleeding on heparin or warfarin.
  5. Diagnostic evaluation of gastrointestinal bleeding and gross haematuria may lead to the diagnosis of previously unknown lesions in one third of patients.
  6. Anticoagulant-related bleeding might be decreased by the use of less intense warfarin therapy (International Normalized Ratio [INR], 2.0 to 3.0)- major bleed (NNT = 76 at unknown) . An INR <3.5 made a major or minor bleed less likely (LR-0.13) .
Landefeld and Beyth: American Journal of Medicine 1993; 95: 315-328
Expires June 2003

The study

Systematic review of prospective studies of
  • Patients: patients on warfarin or heparin
  • Outcome: fatal, major minor bleeding


    • Articles found in all using MEDLINE, 1986-1991 (search terms: heparin, warfarin and adverse effects ) and bibliographies of review articles and original articles were also checked

    Selection criteria: as above
    Appraisal criteria: studies were scored according to set criteria. Unclear if independent multiple reviewers.
    Articles excluded if: not stated

    8 studies on heparin, 25 on warfarin and 25 on bleeding risk were found.
    • Unclear how studies were combined- no meta-analysis.
    • Patients were on warfarin for cerebrovascular disease, atrial fibrillation, coronary artery disease and hip replacement.
    There was no indication of a test for heterogeneity.

    The evidence

    Outcome Time to outcome CEREERRRR
    (95% CI)    		ARR
    (95% CI)    		NNT
    (95% CI)
    warfarin: fatal bleed 1.8 years 7
    (0.20%)    		 37
    (0.94%)    		 -382%
    (-979% to -115%)    		 -0.75%
    (-1.08% to -0.41%)    		 130
    (NNT = 93 to infinity;
    NNH = 240 to infinity)
    warfarin: major bleed 1.8 years 18
    (0.50%)    		 130
    (3.31%)    		 -558%
    (-975% to -303%)    		 -2.80%
    (-3.41% to -2.20%)    		 36
    (29 to 45)
    warfarin: major or minor bleed 1.8 years 130
    (3.63%)    		 569
    (14.5%)    		 -299%
    (-380% to -232%)    		 -10.9%
    (-12.1% to -9.59%)    		 9
    (8 to 10)
    low vs high intensity of anticoagulation: fatal bleed unknown 0
    (0.00%)    		 7
    (0.66%)    		 -100%
    (% to %)    		 -0.66%
    (-1.14% to -0.17%)    		 -152
    (-580 to -88)
    low vs high intensity: major bleed unknown 7
    (0.66%)    		 21
    (1.97%)    		 -198%
    (-597% to -27%)    		 -1.31%
    (-2.28% to -0.34%)    		 -76
    (-290 to -44)
    low vs high intensity: major or minor bleed unknown 44
    (4.16%)    		 130
    (12.2%)    		 -193%
    (-308% to -111%)    		 -8.04%
    (-10.4% to -5.74%)    		 -12
    (-17 to -10)
    with cerebrovascular disease (control vs warfarin): fatal bleed 12 months 2
    (0.45%)    		 21
    (4.68%)    		 -950%
    (-4352% to -148%)    		 -4.23%
    (-6.28% to -2.18%)    		 -24
    (-46 to -16)
    with cerebrovascular disease (control vs warfarin): major bleed 12 months 6
    (1.34%)    		 34
    (7.57%)    		 -467%
    (-1236% to -140%)    		 -6.24%
    (-8.90% to -3.57%)    		 -16
    (-28 to -11)


    diagnostic test major or minor bleed no bleed LR+
    (95% CI)    		 LR-
    (95% CI)
    INR >3.5 2.4
    ( to )     		0.13
    ( to )
    total

    outcome time to outcome number of patients/total number %
    (95% CI)
    fatal bleed on heparin ? 4/937 0.43%
    (0.009% to 0.84%)
    major bleed on heparin ? 56/937 5.98%
    (4.46% to 7.49%)
    minor bleed on heparin ? 150/937 16%
    (13.7% to 18.4%)
    heparin: site of bleeding soft tissue/wound ? 148/474 31%
    (27% to 35%)
    heparin: GI tract ? 130/474 27%
    (23% to 31%)
    heparin: urinary tract ? 90/474 19%
    (16% to 23%)
    heparin: nasopharynx ? 30/474 6.3%
    (4.1% to 8.5%)
    heparin: intracranial ? 9/474 1.9%
    (0.7% to 3.1%)
    heparin: retroperitoneal ? 12/474 2.5%
    (1.1% to 3.9%)
    warfarin: soft tissue/wound ? 39/187 21%
    (15% to 27%)
    warfarin: GI tract ? 28/187 15%
    (9.9% to 20%)
    warfarin: urinary tract ? 28/187 15%
    (9.9% to 20%)
    warfarin: nasopharynx ? 60/187 32%
    (25% to 39%)
    warfarin: intracranial ? 7/187 3.7%
    (1.0% to 6.5%)
    warfarin: retroperitoneal ? 2/187 1.1%
    (0.0% to 2.5%)

    • There were no good RCTs on heparin- all information was from inception cohort studies. ?50% of bleeds due to heparin.
    • Event on warfarin (25 trials):
      • fatal bleed: 26/4328: yearly risk 0.60% (95% CI: 0.37% to 0.83%)
      • major bleed: 130/4318: yearly risk 3.0% (95% CI: 2.5% to 3.5%)
      • major or minor bleed: 415/4318: yearly risk 9.6% (95% CI: 8.7% to 10%)
    • Diagnostic evaluation of GI bleeding and gross haematuria may lead to the diagnosis of previously unknown lesions in 30% of patients (15% of cases are malignant).
    • 1960's: 685 patients (7 RCTs)- INR range 1.5-7.5:
      • fatal bleed- 2%
      • major bleed- 9%
      • major or minor bleed- 24%
      1970's: 122 (3 RCTs)- INR range 1.8-4.4:
      • fatal bleed- 0.8%
      • major bleed- 7%
      • major or minor bleed- 19%
      1980-91: 2226 (11 RCTs)- INR range 1.5-9.0:
      • fatal bleed- 0.4% (p<0.01)
      • major bleed- 2% (p<0.01)
      • major or minor bleed- 13% (p<0.01)
    • 10-64% of intracranial bleeds are fatal.
    • Risk factors for bleeding: therapy:
      • length of anticoagulation- some studies suggest that major bleeding risk decreases with time:
      • -3% in first month
      • -then 0.8%/month for first year
      • -then 0.3% month after that
      • heparin: ?2-5 times greater than warfarin (?artefact of patients in studies)
      • in-patients
    • Risk factors for bleeding: patient:
      • past history of GI bleed
      • cerebrovascular disease (see above)
      • serious heart disease
      • renal insufficiency
      • liver disease in in-patients
      • cancer
      • Hct <0.30
      • on NSAIDs

    Comments

    1. Some of the bleeding rates mentioned seem too high to be credible.
    2. Are various trials really combinable?. Anticoagulation used on a wide range of patients for a wide range of indications. General statements are probably valid especially since seen frequently in such a broad range.
    3. Different rates of bleeding seen in different trials and over time may be related to varying definitions of major or minor bleeds. However, its difficult to explain the decrease in fatal bleed rate over time by this argument.

    Citation

    1. Landefeld CS, and Beyth RJ: Anticoagulant-related bleeding: clinical epidemiology, prediction and prevention. American Journal of Medicine 1993; 95: 315-328
    Search Terms: anticoag* in Cochrane
    Contributor: Chris Ball and Clare Wotton, June 2000
    Reviewer:

    Clinical Question.
    Patient patients
    Intervention or Exposure warfarin or heparin
    Outcome bleeding