Does the combination of rapid D-dimer tests with a pre-test clinical probability model improve the exclusion rate of deep venous thrombosis in symptomatic outpatients?
Combining rapid D-dimer tests with a pre-test clinical probability model improves the exclusion rate of DVT without compromising sensitivity or negative predictive value, potentially reducing unnecessary objective testing.
We assessed the performance of three rapid D-dimer tests (Auto Dimertest, VIDAS and Tinaquant) in combination with a pretest clinical probability model for deep venous thrombosis (DVT) in 106 consecutive outpatients with suspected DVT. Contrast venography or colour-coded duplex ultrasonography demonstrated the presence of DVT in 47 patients (14 distal DVT and 33 proximal DVT). First, we assessed the accuracy indices for different cut-off levels of the rapid D-dimer tests. Sensitivity was found to be 97.9-100%, negative predictive value (NPV) was 96.3-100%, and the exclusion rate was 24.5-31.1%. Next, the patients were grouped according to the pre-test clinical probability model in categories with low, moderate or high probability. In patients with a low pre-test probability, DVT would have been directly ruled out and the patients would not have undergone further investigations. In patients with a moderate probability, D-dimer testing and, in the case of a positive result, objective testing would have been performed and, in the case of a negative result, they would have been ruled out of having DVT. Patients with high probability would directly have undergone objective tests for DVT. The combination with the pre-test clinical probability model improved the exclusion rate (43.5-44.6%), whereas sensitivity (97.5-100%) and NPV (97.6-100%) remained roughly unchanged. The combination of rapid D-dimer tests with a pre-test clinical probability model may help to reduce unnecessary work-up in patients with suspected DVT.
Fünfsinn et al. (2001) studied this question.