Résumé
OBJECTIVE: The aim of this study was to assess in clinical practice the accuracy of a referent d-dimer enzyme-linked immunosorbent assay for the exclusion of venous thromboembolic disease (VTED). PATIENTS AND METHODS: An observational prospective study took place in an emergency department; 205 consecutive outpatients suspected of having VTED were included. Blood samples were collected at admission for VIDAS DD measurement. Venous thromboembolic disease was confirmed by standard clinical imaging. All patients were followed up at 3 months. RESULTS: Venous thromboembolic disease was confirmed in 57 patients (28%). The sensitivity and negative predictive value of a DD assay lower than 500 ng/mL were 78% (95% confidence interval = 67%-87%) and 84% (95% confidence interval = 73%-90%), respectively. Twelve patients had a false-negative DD with one or more of the following: (a) symptoms reported for more than 15 days (n = 2), (b) prior anticoagulation (n = 3), (c) distal VTED (n = 5), or (d) high clinical probability (n = 3). CONCLUSION: In our cohort of patients, DD was less accurate than previously reported, with an upper estimate of the sensitivity of only 87%.