Thromb Haemost 1994; 71(03): 286-291
DOI: 10.1055/s-0038-1642432
Original Article
Schattauer GmbH Stuttgart

Duration of Oral Anticoagulant Therapy after Proximal Deep Vein Thrombosis: a Decision Analysis

François P Sarasin
The Clinique Médicale 1 and Division of Angiology and Haemostasis, Hôpital Cantonal, University of Geneva Medical School, Geneva, Switzerland
,
Henri Bounameaux
The Clinique Médicale 1 and Division of Angiology and Haemostasis, Hôpital Cantonal, University of Geneva Medical School, Geneva, Switzerland
› Author Affiliations
Further Information

Publication History

Received: 17 March 1993

Accepted after revision 18 December 1993

Publication Date:
06 July 2018 (online)

Summary

The optimal duration of oral anticoagulant therapy following proximal deep vein thrombosis (DVT) in the lower limbs remains controversial. To compare the risk benefit tradeoffs for different treatment durations (6 to 24 weeks) we constructed a Markov-based decision analysis model which explicitly balances the time-dependent declining risk of recurrent thrombosis and pulmonary embolism against the risk of major hemorrhagic complications. Specifically, we determined the threshold below which the risk of recurrent DVT exceeds the risk of major hemorrhage if anticoagulant therapy is discontinued, and above which the benefits provided by oral anticoagulants are outweighed by their risk.

Our model shows that for patients with a low hemorrhagic risk (0.5%/month), the benefit yielded by oral anticoagulants breaks off beyond the 4th month of therapy, while patients with moderate (1%/month) to high (2%/month) bleeding risk will no longer benefit from the therapy after 3 or 2.5 months, respectively.

In conclusion, our model supports the validity of the usually recommended duration of 3 months of oral anticoagulation after proximal vein thrombosis in the lower limbs, but suggests that this duration should be modulated between 2.5 and 4 months depending upon individual bleeding risk. Since clinical trials can hardly handle the complexity of the addressed issue, such a model may prove very helpful in daily clinical practice.