Once and only once.

V enous thrombosis is a fairly common disorder that manifests mainly as deep vein thrombosis of the leg or pulmonary embolism. It affects 1 in 1000 individuals annually and has a substantial mortality rate. 1 Its causes are both genetic (prothrombotic abnormalities sometimes referred to as a group as thrombophilia) and acquired. 2 Among the latter are cancer, female hormone use, and all forms of prolonged immobility (surgery, trauma, plaster casts, medical diseases). Venous thrombosis has a high recurrence rate of several percent per year; hence, efforts to reduce this risk are in order. 3–5 A targeted approach is to use antithrombotic pro-phylaxis liberally during transient circumstances of increased risk. However, because a substantial number of recurrent events occur without a clear provoking factor, this approach will prevent only a fraction of all recurrent events. There is little doubt that indefinite treatment with anticoagulants will reduce the risk of recurrence, but it will be at the price of an increased number of major, debilitating, and in some cases fatal hemorrhages. Given that current treatment with antico-agulants still carries an annual risk of major hemorrhage of 1% to 2%, indiscriminate indefinite treatment of every patient with a first thrombosis is clearly not in order. The challenge therefore is to identify those patients who are at an elevated risk of recurrence and investigate the potential benefit of prolonged or indeterminate anticoagulant prophylaxis in these patients. One group that is a candidate for such an approach is made up of patients with nontransient risk factors for thrombosis (ie, thrombophilia). One would think that in this era of evidence-based medicine such a risky strategy would be recommended only in light of clear evidence pointing to the benefits outweighing the evidence. Quite surprisingly, indeterminately prolonged anticoagulant treatment is often prescribed not only in the absence of any evidence but in the presence of evidence pointing in the opposite direction. What would studies look like that would convince us that in a particular circumstance anticoagulation would be beneficial? They would ideally be randomized trials of patients with a certain condition, contrasting those with and without anticoagulation. Outcomes would be thrombosis and hemorrhage, and the prevention of thrombosis would need to outweigh the excess of hemorrhage with treatment; hence, these studies would need to be large to be powerful enough to allow meaningful comparisons. This is the typical method by which the benefit of anticoagulation has been …