Background: Patients with unprovoked venous thromboembolism (VTE) can be stratified according to their recurrence risk based on their sex, the VTE location, and D-dimer measured 3 weeks after anticoagulation by the Vienna Prediction Model. We aimed to expand the model to also assess the recurrence risk from later points on.
Methods and results: Five hundred and fifty-three patients with a first VTE were followed for a median of 68 months. We excluded patients with VTE provoked by a transient risk factor or female hormone intake, with a natural inhibitor deficiency, the lupus anticoagulant, or cancer. The study end point was recurrent VTE, which occurred in 150 patients. D-dimer levels did not substantially increase over time. Subdistribution hazard ratios (95% confidence intervals) dynamically changed from 2.43 (1.57 to 3.77) at 3 weeks to 2.27 (1.48 to 3.48), 1.98 (1.30 to 3.02) , and 1.73 (1.11 to 2.69) at 3, 9, and 15 months in men versus women, from 1.84 (1.00 to 3.43) to 1.68 (0.91 to 3.10), 1.49 (0.79 to 2.81) , and 1.44 (0.76 to 2.72) in patients with proximal deep vein thrombosis or pulmonary embolism compared with calf vein thrombosis, and from 1.30 (1.07 to 1.58) to 1.27 (1.06 to 1.51), 1.20 (1.02 to 1.41), and 1.13 (0.95 to 1.36) per doubling D-dimer. Using a dynamic landmark competing risks regression approach, we generated nomograms and a web-based calculator to calculate risk scores and recurrence rates from multiple times after anticoagulation.
Conclusions: Risk of recurrent VTE after discontinuation of anticoagulation can be predicted from multiple random time points by integrating the patient's sex, location of first VTE, and serial D-dimer measurements.
Keywords: Vienna Prediction Model; d‐dimer; recurrence; venous thrombosis.