Prognostic significance of residual venous obstruction in patients with treated unprovoked deep vein thrombosis: a patient-level meta-analysis.

Residual venous obstruction (RVO) could improve the stratification of the risk of recurrence after unprovoked deep vein thrombosis (DVT), but results from clinical studies and study-level meta-analyses are conflicting. It was the objective of this analysis to determine if RVO is a valid predictor of recurrent venous thromboembolism (VTE) in patients with a first unprovoked DVT who had received at least three months of anticoagulant therapy. Individual patient data were obtained from the datasets of original studies, after a systematic search of electronic databases (Medline, Embase, Cochrane Library), supplemented by manual reviewing of the reference lists and contacting content experts. A multivariate, shared-frailty Cox model was used to calculate hazard ratios (HRs) for recurrent VTE, including, as covariates: RVO; age; sex; anticoagulation duration before RVO assessment; and anticoagulation continuation after RVO assessment. A total of 2,527 patients from 10 prospective studies were included. RVO was found in 1,380 patients (55.1%) after a median of six months from a first unprovoked DVT. Recurrent VTE occurred in 399 patients (15.8%) during a median follow-up of 23.3 months. After multivariate Cox analysis, RVO was independently associated with recurrent VTE (HR = 1.32, 95% confidence interval [CI]: 1.06-1.65). The association was stronger if RVO was detected early, i.e. at three months after DVT (HR = 2.17; 95% CI: 1.11-4.25), but non-significant if detected later, i.e. >6 months (HR = 1.19; 95% CI: 0.87-1.61). In conclusion, after a first unprovoked DVT, RVO is a weak overall predictor of recurrent VTE. The association is stronger if RVO is detected at an earlier time (3 months) after thrombosis.