Risk of recurrence after a first unprovoked venous thromboembolism: external validation of the Vienna Prediction Model with pooled individual patient data

Abstract

In order to stratify patients with a first unprovoked venous thromboembolism (VTE) according to their recurrence risk and to identify those who would actually benefit from indefinite anticoagulation, three prediction models have been developed so far; none of them has been yet externally validated. To externally validate the Vienna Prediction Model (VPM), a prediction guide for estimating the recurrence risk after a first unprovoked VTE developed through Cox modeling and including sex, D-dimer and index VTE site as predictors. Nine hundred and four patients pooled from five prospective studies evaluating the prognostic value of D-dimer for VTE recurrence served as the validation cohort. The validity of the VPM in stratifying patients according to their relative recurrence risk (discrimination) and in predicting the absolute recurrence risk (calibration) was tested with survival analysis methods. The ability of the VPM to distinguish patients' risk for recurrent VTE in the validation cohort was at least as good as in the original cohort, with a calibration slope of 1.17 (95% confidence interval 0.71-1.64; P=0.456 for the hypothesis of a significant difference from 1), and a c-statistic of 0.626 (vs. 0.651 in the original derivation cohort). The VPM absolute predictions in terms of cumulative rates tended to underestimate the observed recurrence rates at 12months. By using a pooled individual patient database as a validation cohort, we confirmed the ability of the VPM to stratify patients with a first unprovoked VTE according to their risk of recurrence.