Hemorrhagic Complications of Anticoagulant Treatment Mark N. Levine, M.D., M.Sc., Chainnan jack Hirsh, M.D., F.C.C.P. Seth Landefeld, M.D. Gary Raskob, M.Sc.
Thebleeding. major complication of anticoagulant therapy is In this review, the incidence of hemorrhage in patients receiving oral anticoagulants or heparin and the clinical and laboratory risk factors that predispose to bleeding are discussed. The published reports are classified into one of five categories based on the strength of the study design, and whenever possible, conclusions are based on results of randomized control trials. 1•2 Hemorrhage was classified as major if it was intracranial or retroperitoneal, if it led directly to death, or if it resulted in hospitalization or transfusion. 1•2 All other bleeding episodes were classified as minor. The details of bleeding events and definition of bleeding varied considerably between studies. Details of the methodology used to select relevant articles can be found in our reviews in the two previous symposia of the American College of Chest Physicians. 1•2 Since these previous reviews, important new information has emerged from randomized trials on the relationship between intensity of oral anticoagulant therapy and hemorrhage and on the risk of hemorrhage in patients receiving oral anticoagulants for atrial fibrillation. In this review, we include a section on heparin-associated bleeding that was not included in the previous reviews. 1·2 Although bleeding is considered the major side effect of anticoagulant therapy, it should not be considered in isolation of potential benefit, ie, reduction in thrombosis. ORAL ANTICOAGULANTS
The major determinants of oral anticoagulant-induced bleeding are the intensity of the anticoagulant effect, patient characteristics, the use of drugs that interfere with hemostasis, and the length of therapy.
Relationship Between Intensity of Anticoagulant Therapy and Hemorrhage The intensity of the anticoagulant response to warfarin is influenced by the dose of warfarin, the patient's vitamin K intake, interindividual variability in anticoagulant response to a given dosage, and other factors that alter the pharmacokinetics and pharmacodynamics of warfarin, of which the most important are concomitant medication and liver function. 352$
There is a strong relationship between the intensity of anticoagulant therapy and the risk of bleeding that has been reported in patients with deep vein thrombosis, tissue heart valves, and mechanical heart valves. In patients with venous thrombosis, Hull and colleagues3 reported a strong relationship between the intensity of warfarin treatment and the risk of bleeding (level I). The patients randomized to receive less intense warfarin (targeted international normalized ratio [INR] = 2.0) had significantly less bleeding (4 percent vs 22 percent) than those receiving more intense therapy (targeted INR = 2.5 to 4.5). In the 13 patients who bled, the INR was greater than 3.0. In a trial reported by Turpie et al, 4 patients with tissue heart valves were randomized to either standard intensity warfarin (INR 2.5 to 4.0) or less intense warfarin (INR 2.0 to 2.5) (level I). The rate of clinically important bleeding over three months was reduced from 14 percent in the standard intensity group to 6 percent in the low intensity group without loss of antithrombotic efficacy. In a reported trial by Saour et al, 5 patients with prosthetic heart valves were randomized to either moderate intensity warfarin (targeted INR 2.65) or high-intensity warfarin (targeted INR 9.0) (level I). Bleeding was more common in the higher-intensity group. Of the 122 patients randomized to the moderate intensity, 4 experienced major bleeding and 22 experienced minor bleeding. In the 125 patients who received the higher-intensity warfarin, there were 9 major bleeds (of which 2 were fatal intracranial) and 44 minor bleeds. The difference in total and minor bleeding between groups was statistically significant. It is important to note, however, that the targeted intensity of the higher-intensity arm in this trial is much higher than the currently recommended intensity. In a trial recently reported by Altman et al, 6 patients with mechanical valves received either acenocoumarol (INR 3.0 to 4.5) or acenocourmarol (INR 2.0 to 2.99) (level I). All patients received concurrent aspirin and dipyridamole. There was a statistically significant reduction in bleeding in patients who received the less intense therapy: 21 percent vs 4 percent. Hemorrhagic Complications (Lavine et al)
Finally, results from the recent Boston randomized trial indicate that lower-dose oral anticoagulants (INR 1.5 to 2. 7) can be administered safely in patients with atrial fibrillation. 7
Relationship Between Risk of Bleeding and lhtient Characteristics Major bleeding is especially likely in patients with ischemic cerebral vascular disease and venous thromboembolism, possibly owing to the higher prevalence of associated risk factors in these conditions. 1·2 In venous thromboembolism, these comorbid conditions (cancer, recent surgery, and paraplegia) predispose the patient to thrombosis. In cerebrovascular disease, major bleeding is usually intracerebral, possibly because of associated hypertension or the vascular disease per se. The risk of bleeding during warfarin therapy is probably also related to patient characteristics other than the indication for therapy. Results of observational studies (level IV) suggest that the risk of bleeding is related to age, history of past bleeding, and specific comorbid conditions. However, the presence of such factors should not be considered as contraindications to anticoagulant therapy, but clinical judgment may need to be used in certain situations. Several studies have found that the frequency of bleeding during warfarin therapy is higher in older patients,s- 15 although some other studies disagree. 16- 19 The only two studies of large inception cohorts9·10 found an increased bleeding risk in older patients, with a relative risk for major bleeding of 3.2 for patients aged 65 years or older in one study. 9 Past gastrointestinal bleeding is also an indicator of increased risk of bleeding during warfarin therapy. 9 In contrast, peptic ulcer disease alone, without past bleeding, has not been associated with an increased risk of bleeding. 9·10•20 Several comorbid diseases have been associated
with bleeding during warfarin therapy; these include treated hypertension, 9 •10•12·14.2 1 cerebrovascular disease, 9 serious heart disease, 9·11 and renal insufficiency. 9 Although many other patient characteristics have been associated with bleeding during warfarin therapy, the data supporting these findings are not compelling. For example, a few studies have noticed an increased frequency of bleeding among women treated with warfarin, 10•11 •13 although most have not. 9·12·15·17 ·18 Although most experienced clinicians believe that either alcoholism or liver disease increases the risk of bleeding during long-term warfarin therapy, two recent studies did not find such an association. 10•11 However, these negative findings may reflect the infrequency with which such patients are treated with warfarin and, perhaps, inaccurate documentation of alcoholism in medical records. Two recent observational studies10·21 reported higher frequencies of warfarin-related bleeding in patients with atrial fibrillation than would be expected on the basis of results from randomized trials; this contrast may reflect differences in the intensity of anticoagulant effect or in the frequency of other risk factors for bleeding. Independent risk factors for bleeding during longterm warfarin therapy were recently identified:9 •22 age 65 years or older, history of past gastrointestinal bleeding, stroke (present or past), atrial fibrillatidn, and one or more of three specific co morbid conditions (recent myocardial infarction, renal insufficiency, •or severe anemia). Hypertension was not associated with increased risk for bleeding after controlling for these risk factors. In addition, there was a suggestion that patients who bled with the prothrombin time (IT) in the therapeutic range were subsequently found to have an underlying pathologic lesion as the cause of bleeding. These studies suggest that the yield of finding an underlying occult bleeding source is higher if the bleed occurs when the INR is within the therapeutic range. However, it would seem prudent
Table l - lschemic Cerebrovascular Disease Bleeding Study Baker"' Baker"', Fisher'" Enger and Boyesen"' Hill et al 31 McDowell et al'" Olsson et al 21
Treatment Warfarin/dicurnarol No Rx Dicurnarol Placebo Phenylindanedione Placebo Phenindione (high) Phenindione (low) Warfarin No Rx Warfarin ASA + dipyridarnole
Patient No. 78 77
224 219 52 48 71 71
95 105 68 67
Major(%)
Fatal(%)
Targeted INR
(39.7)* (6.5) (39.3)* (4.6) (19.2)*
10 (12.8)
4 (5.1)
5.6
3
(3.9)
l (1.3)
(8.0) (0.9) (7.7)
(21.1)* (1.4) (17.9)* (1.0) (11.8) (4.5)
18 2 4 0 5 1 2 1 2 0
(7.0) (1.4) (2.1) (1.0) (2.9)
Total(%) 31 5 88 10 10 0 15 1 17 1 8 3
12 0 3 0 5 0 2 1 2 0
(5.4)
4.5-7.2
(5.8)
1.5-2.l
(7.0)
2.0-2.5
(2.1) (1.0) (2.9)
4.4-7.5 1.9-3.7
*p
Thebleeding. major complication of anticoagulant therapy is In this review, the incidence of hemorrhage in patients receiving oral anticoagulants or heparin and the clinical and laboratory risk factors that predispose to bleeding are discussed. The published reports are classified into one of five categories based on the strength of the study design, and whenever possible, conclusions are based on results of randomized control trials. 1•2 Hemorrhage was classified as major if it was intracranial or retroperitoneal, if it led directly to death, or if it resulted in hospitalization or transfusion. 1•2 All other bleeding episodes were classified as minor. The details of bleeding events and definition of bleeding varied considerably between studies. Details of the methodology used to select relevant articles can be found in our reviews in the two previous symposia of the American College of Chest Physicians. 1•2 Since these previous reviews, important new information has emerged from randomized trials on the relationship between intensity of oral anticoagulant therapy and hemorrhage and on the risk of hemorrhage in patients receiving oral anticoagulants for atrial fibrillation. In this review, we include a section on heparin-associated bleeding that was not included in the previous reviews. 1·2 Although bleeding is considered the major side effect of anticoagulant therapy, it should not be considered in isolation of potential benefit, ie, reduction in thrombosis. ORAL ANTICOAGULANTS
The major determinants of oral anticoagulant-induced bleeding are the intensity of the anticoagulant effect, patient characteristics, the use of drugs that interfere with hemostasis, and the length of therapy.
Relationship Between Intensity of Anticoagulant Therapy and Hemorrhage The intensity of the anticoagulant response to warfarin is influenced by the dose of warfarin, the patient's vitamin K intake, interindividual variability in anticoagulant response to a given dosage, and other factors that alter the pharmacokinetics and pharmacodynamics of warfarin, of which the most important are concomitant medication and liver function. 352$
There is a strong relationship between the intensity of anticoagulant therapy and the risk of bleeding that has been reported in patients with deep vein thrombosis, tissue heart valves, and mechanical heart valves. In patients with venous thrombosis, Hull and colleagues3 reported a strong relationship between the intensity of warfarin treatment and the risk of bleeding (level I). The patients randomized to receive less intense warfarin (targeted international normalized ratio [INR] = 2.0) had significantly less bleeding (4 percent vs 22 percent) than those receiving more intense therapy (targeted INR = 2.5 to 4.5). In the 13 patients who bled, the INR was greater than 3.0. In a trial reported by Turpie et al, 4 patients with tissue heart valves were randomized to either standard intensity warfarin (INR 2.5 to 4.0) or less intense warfarin (INR 2.0 to 2.5) (level I). The rate of clinically important bleeding over three months was reduced from 14 percent in the standard intensity group to 6 percent in the low intensity group without loss of antithrombotic efficacy. In a reported trial by Saour et al, 5 patients with prosthetic heart valves were randomized to either moderate intensity warfarin (targeted INR 2.65) or high-intensity warfarin (targeted INR 9.0) (level I). Bleeding was more common in the higher-intensity group. Of the 122 patients randomized to the moderate intensity, 4 experienced major bleeding and 22 experienced minor bleeding. In the 125 patients who received the higher-intensity warfarin, there were 9 major bleeds (of which 2 were fatal intracranial) and 44 minor bleeds. The difference in total and minor bleeding between groups was statistically significant. It is important to note, however, that the targeted intensity of the higher-intensity arm in this trial is much higher than the currently recommended intensity. In a trial recently reported by Altman et al, 6 patients with mechanical valves received either acenocoumarol (INR 3.0 to 4.5) or acenocourmarol (INR 2.0 to 2.99) (level I). All patients received concurrent aspirin and dipyridamole. There was a statistically significant reduction in bleeding in patients who received the less intense therapy: 21 percent vs 4 percent. Hemorrhagic Complications (Lavine et al)
Finally, results from the recent Boston randomized trial indicate that lower-dose oral anticoagulants (INR 1.5 to 2. 7) can be administered safely in patients with atrial fibrillation. 7
Relationship Between Risk of Bleeding and lhtient Characteristics Major bleeding is especially likely in patients with ischemic cerebral vascular disease and venous thromboembolism, possibly owing to the higher prevalence of associated risk factors in these conditions. 1·2 In venous thromboembolism, these comorbid conditions (cancer, recent surgery, and paraplegia) predispose the patient to thrombosis. In cerebrovascular disease, major bleeding is usually intracerebral, possibly because of associated hypertension or the vascular disease per se. The risk of bleeding during warfarin therapy is probably also related to patient characteristics other than the indication for therapy. Results of observational studies (level IV) suggest that the risk of bleeding is related to age, history of past bleeding, and specific comorbid conditions. However, the presence of such factors should not be considered as contraindications to anticoagulant therapy, but clinical judgment may need to be used in certain situations. Several studies have found that the frequency of bleeding during warfarin therapy is higher in older patients,s- 15 although some other studies disagree. 16- 19 The only two studies of large inception cohorts9·10 found an increased bleeding risk in older patients, with a relative risk for major bleeding of 3.2 for patients aged 65 years or older in one study. 9 Past gastrointestinal bleeding is also an indicator of increased risk of bleeding during warfarin therapy. 9 In contrast, peptic ulcer disease alone, without past bleeding, has not been associated with an increased risk of bleeding. 9·10•20 Several comorbid diseases have been associated
with bleeding during warfarin therapy; these include treated hypertension, 9 •10•12·14.2 1 cerebrovascular disease, 9 serious heart disease, 9·11 and renal insufficiency. 9 Although many other patient characteristics have been associated with bleeding during warfarin therapy, the data supporting these findings are not compelling. For example, a few studies have noticed an increased frequency of bleeding among women treated with warfarin, 10•11 •13 although most have not. 9·12·15·17 ·18 Although most experienced clinicians believe that either alcoholism or liver disease increases the risk of bleeding during long-term warfarin therapy, two recent studies did not find such an association. 10•11 However, these negative findings may reflect the infrequency with which such patients are treated with warfarin and, perhaps, inaccurate documentation of alcoholism in medical records. Two recent observational studies10·21 reported higher frequencies of warfarin-related bleeding in patients with atrial fibrillation than would be expected on the basis of results from randomized trials; this contrast may reflect differences in the intensity of anticoagulant effect or in the frequency of other risk factors for bleeding. Independent risk factors for bleeding during longterm warfarin therapy were recently identified:9 •22 age 65 years or older, history of past gastrointestinal bleeding, stroke (present or past), atrial fibrillatidn, and one or more of three specific co morbid conditions (recent myocardial infarction, renal insufficiency, •or severe anemia). Hypertension was not associated with increased risk for bleeding after controlling for these risk factors. In addition, there was a suggestion that patients who bled with the prothrombin time (IT) in the therapeutic range were subsequently found to have an underlying pathologic lesion as the cause of bleeding. These studies suggest that the yield of finding an underlying occult bleeding source is higher if the bleed occurs when the INR is within the therapeutic range. However, it would seem prudent
Table l - lschemic Cerebrovascular Disease Bleeding Study Baker"' Baker"', Fisher'" Enger and Boyesen"' Hill et al 31 McDowell et al'" Olsson et al 21
Treatment Warfarin/dicurnarol No Rx Dicurnarol Placebo Phenylindanedione Placebo Phenindione (high) Phenindione (low) Warfarin No Rx Warfarin ASA + dipyridarnole
Patient No. 78 77
224 219 52 48 71 71
95 105 68 67
Major(%)
Fatal(%)
Targeted INR
(39.7)* (6.5) (39.3)* (4.6) (19.2)*
10 (12.8)
4 (5.1)
5.6
3
(3.9)
l (1.3)
(8.0) (0.9) (7.7)
(21.1)* (1.4) (17.9)* (1.0) (11.8) (4.5)
18 2 4 0 5 1 2 1 2 0
(7.0) (1.4) (2.1) (1.0) (2.9)
Total(%) 31 5 88 10 10 0 15 1 17 1 8 3
12 0 3 0 5 0 2 1 2 0
(5.4)
4.5-7.2
(5.8)
1.5-2.l
(7.0)
2.0-2.5
(2.1) (1.0) (2.9)
4.4-7.5 1.9-3.7
*p
