A study of 10 rapid d-dimer tests for the exclusion of acute venous thromboembolism

Abstract number: P1477a

Gardiner^* C., Mackie^* I. J., PENNANEAC'H† C., Abe† G., Walford† C., Machin‡ S. J.

†University College Hospital, UK; ^*University College London, UK; ‡Univesrity College London, UK

A large number of sensitive rapid d-dimer assays have recently been developed for the purpose of the exclusion of venous thromboembolism (VTE). We studied the performance of 10 rapid d-dimer methods frequently used in the UK for the exclusion of VTE: Amax Auto D-Dimer, Biopool AutoDimer, and MiniQuant d-dimer (Trinity Biosciences); VIDAS and MDA D-Dimer (bioMerieux); D-Dimer Plus (Dade Behring); Auto d-dimer 540 and Auto d-dimer 700 (Helena Biosciences), IL Test (Instrumentation Laboratory), NycoCard (Axis Shield). D-dimer assays were performed on 119 patients presenting with suspected acute VTE. Pre-test probability was assessed using a modified version of the Wells scoring system and VTE was confirmed using objective imaging techniques (Doppler US, VQ scan, CT scan and contrast venography where appropriate). ROC analysis was used to establish cut-off values corresponding to 100, 95 and 90% sensitivity for VTE and these were compared with any cut-off values provided by the manufacturers. 25/119 patients were diagnosed with VTE (24 DVT, 1 PE). There was wide variation in cut-off levels (21–229 and 485–804 ng mL^-1 FEU) and specificity (19–66%) at the 100% sensitivity level. However, when the patients were divided into two groups on the basis of pretest clinical scoring of DVT likely (score >=2) and DVT unlikely (score <=1, and 90% sensitivity cut-off levels were applied to the d-dimer assays, only one patient with demonstrable VTE was not correctly identified. We conclude that care must be taken in assigning d-dimer cut-off values for the exclusion of VTE and that even then, performance in terms of positive predictive value and specificity are highly variable. Our data suggests that d-dimer levels may be useful in the exclusion of VTE but should always be used in conjunction with standardized assessment of pretest probability.