A comparison of three months of anticoagulation with extended anticoagulation for a first episode of idiopathic venous thromboembolism 

Authors Kearon C, Gent M, Hirsh J, Weitz J, Kovacs M, Anderson D, Turpie A et al. 
Source New England Journal of Medicine. 340:901-7. March 25, 1999. 
Institutions Multi-institutional in Canada and Northwestern University Medical School (USA).
Support Dupont Pharma; Medical Research Council and the Heart and Stroke Foundation of Canada;  Ministry of Health of Ontario.


Recent studies suggest that patients who have a first episode of venous thromboembolism are at greater risk for recurrence after anticoagulation is discontinued if they did not have a transient risk factor at the time of the initial event.  These patients either have persistent risk factors or had no known risk factors at the time of their initial thromboembolism.

Based on this, it was speculated that patients who present with idiopathic venous thromboembolism should benefit from oral anticoagulation for longer than the conventional three months.  This study was designed to test that hypothesis. After three months of warfarin, patients with idiopathic venous thromboembolism (DVT or pulmonary embolism) were randomized to an additional 24 months of anticoagulation or placebo.



Eligible:  patients with a first episode of idiopathic venous thromboembolism, treated initially with heparin and then with three months of warfarin.  Venous thromboembolism was defined as symptomatic deep vein thrombosis of the leg or pulmonary embolism.

Exclusion criteria included

  • Predisposing factors for thromboembolism (which would make it non-idiopathic): fracture, hospitalization for at least three days or general anesthesia (all within the previous three months); cancer within the previous 5 years; known hypercoagulable state.

  • Factors that would interfere with anticoagulant therapy or monitoring, principally: contraindication to or requirement for anticoagulant therapy; treatment with NSAIDS or anti-platelet medications (aspirin up to 160 mg daily was allowed); potential pregnancy; bleeding disorder; inability to comply with the regimen.


After all patients had received 3 months of warfarin, patients were randomized to receive either 24  months of anticoagulation or placebo.  Patients randomized to placebo had blood tests just like those randomized to anticoagulation, and their placebo dose was adjusted according to a sham INR.  Those on warfarin had their dose adjusted to maintain an INR of 2.0 to 3.0. 


At baseline, patients underwent lung-scanning, compression ultrasound of the legs and impedence plethysmography (where possible).  They were evaluated every three months for any signs or symptoms of recurrent thromboembolism and were also instructed to return for immediate evaluation if any symptoms suggesting DVT or PE developed.  Evaluation in case of suspected DVT consisted of ultrasound and venography if ultrasound was negative; evaluation for suspected PE included lung scanning and, if negative, ultrasound and/or venography and/or pulmonary angiography.

At baseline, blood was obtained for factor V Leiden, the G20210A prothrombin gene mutation, anticardiolipin antibodies and the lupus anticoagulant, but the results of those assays were not made available to the clinicians.


The primary endpoint was a comparison in the rates of recurrent venous thromboembolism in the two groups.  Based on previous studies and a power analysis, it was planned that 190 patients would be enrolled, with follow-up for one year after the last patient was enrolled. One interim analysis was planned after 150 patients had been enrolled, with a decision to terminate the study early if there was unequivocal evidence for efficacy in the treatment group (p<0.001).



241 patients met the inclusion criteria at the time of their initial event.  Of these, 37 were excluded after 3 months of anticoagulation therapy (mainly for subsequent diagnosis of cancer, geographic inaccessibility or indication/contra-indication for anticoagulant therapy).  Of the remaining 204 patients, 162 gave informed consent and were randomized to either warfarin (79 patients) or placebo (82).

Baseline characteristics included:

  • Age 59 years; male sex 60%
  • DVT only at presentation: 73% in placebo group, 76% in warfarin group
  • Pulmonary embolism at presentation (with or without DVT): 27% vs. 24%

Length of treatment and compliance

Mean duration of follow-up was 9 months in the placebo group and 12 months in the warfarin group (the shorter duration of follow-up in the placebo group was largely due to recurrences of thromboembolism in this group).

The study drug was discontinued prematurely in 14 patients assigned to warfarin (8 at patient's request, 3 for major bleeding) and in 13 patients assigned to placebo (7 at patient's request).

The mean INR was 2.5 in the warfarin group. It was estimated to be under 2.0 for 22% of the time, and greater than 3.0 for 14% of the time.

Recurrent venous thromboembolism

The trial was halted early, and recruitment stopped, after the interim analysis.

Among the 79 patients assigned to warfarin, there was only episode of recurrent venous thromboembolism. This was a pulmonary embolus in a patient who had discontinued warfaring 14 months earlier because of upper GI bleeding.

Among the 83 patients assigned to placebo, there were 17 episodes of recurrent thromboembolism: 11 DVT's and 6 pulmonary emboli (one fatal).

The rate of recurrent thromboembolism was 1.3 percent per patient-year in the warfarin group, vs. 27.4% per patient-year in the placebo group (P<0.001).

All episodes of recurrent venous thromboembolism were idiopathic (without obvious precipitating factors).

Bleeding complications and death

There were three major bleeding episodes, none fatal, among patients assigned to warfarin: two episodes of GI bleeding (INRs were 5.4 and 2.9) and one episode of genitourinary bleeding (INR greater than 10).  There was no major bleeding in the placebo group.

There was one death in the warfarin group (pneumonia) and three in the placebo group (pulmonary embolism, coronary disease and leukemia).

Risk factors for thromboembolism

The prevalence of factor V Leiden was 26%, of G20210A was 5% and of antiphospholipid antibody was 5%.

The only risk factor that was found to be significantly associated with recurrent thromboembolism was the lupus anticoagulant, which was present in 2 out of 16 patients assigned to placebo who had a recurrence and in 1 out of 61 assigned to placebo who did not have a recurrence.

Author's discussion

The authors note that the rate of recurrent thromboembolism in the placebo group was higher in this study than in other recent studies. They speculate that, in those other studies, some patients may have been misclassified as having had idiopathic thromboembolism at the time of diagnosis when, in fact, they had thromboembolism due to a transient risk factor.

They note that extending warfarin therapy for longer than 3 months markedly decreased the incidence of recurrent thromboembolism, but at the price of a higher incidence of major bleeding (non fatal in this study).  They speculate that continuing anticoagulation at a lower target INR (under 2.0) might be effective in reducing recurrent events, at a lower risk of bleeding.

Finally, they note that the optimal duration of continued anticoagulation in these patients remains unknown.



This well-designed study of anticoagulation in patients with idiopathic thromboembolism demonstrated a very marked reduction in events among patients whose anticoagulation was extended beyond three months.

As the authors note, their study does not provide evidence for the optimal duration of treatment of these patients.  At 12 months follow-up, there were only 61 patients left in the study, from the original 162.  Looking at the cumulative probability curves for recurrence (shown in the article), it would appear that the benefit extended to at least 6 months.

These results cannot be extrapolated directly to patients who have a thromboembolic event secondary to a defined, transient or modifiable risk factor (such as surgery, trauma, bedrest), in whom the removal of the risk factor would be expected to lower the risk of recurrence in the absence of anticoagulation.

Whether or not anticoagulation at a lower INR would also be effective is an important question that needs to be addressed in further studies.  Lower levels of anticoagulation have been disappointing in the treatment of atrial fibrillation, but venous thromboembolism is a different disorder from atrial thromboembolism.

June 9, 1999


References related to this article from the NLM's PubMed database. 

Reader Comments


August 12, 1999

Letters to the editor about this article, in the August 12 NEJM.  Does anticoagulation reduce or merely delay recurrent thrombosis?  Optimum duration of treatment?


Submit a comment

Journal Club homepage

Site Meter