In two studies, the combination of physicians' low-probability clinical estimates plus negative D-dimer tests yielded acceptably low post-test probabilities of PE.
Studies suggest that anticoagulation can be withheld safely in people with low clinical probability of pulmonary embolism (PE) plus negative D-dimer tests. Results from an observational study and a randomized trial now verify earlier findings.
In a study from North Carolina, researchers assessed the performance of a bedside qualitative D-dimer test (the Simplify DD assay). The test was performed on 2302 patients evaluated for PE in the emergency department. In each case, physicians used their clinical "gestalt" to estimate pretest probability of PE as low (<15%), moderate (15%–40%), or high (>40%). Most patients with low pretest probability and negative D-dimer results did not undergo imaging; absence of venous thromboembolism (VTE) during 90 days of follow-up ruled out PE in these patients. According to clinicians' unstructured estimates, 1650 patients had low pretest probability of PE. D-dimer results were negative in 1239 of these patients, and only 9 of them (0.7%) were eventually diagnosed with PE.
Canadian researchers performed a randomized trial to compare qualitative D-dimer testing alone with D-dimer plus additional testing. Patients included in the study received no anticoagulation, had negative D-dimer test results (on the SimpliRED assay), and had either low probability of PE (based on a prediction rule) or moderate-to-high probability plus nondiagnostic ventilation-perfusion scans and negative bilateral proximal deep-vein ultrasonography. Half of the low-probability group underwent lung scans, then ultrasonography. Low-probability patients with nondiagnostic lung scans, and half of the moderate-to-high probability patients, had repeat ultrasonography at 1 and 2 weeks. By 6 months, none of 187 low-probability patients who were assigned to no additional testing had VTE diagnosed (5 were lost to follow-up); 1 of 186 low-probability patients who underwent additional testing had VTE diagnosed during follow-up and died (1 was lost to follow-up). In the moderate-to-high probability groups, 1 of 42 patients with no additional testing had VTE diagnosed, and 1 of 41 patients who underwent additional testing had subsegmental PE diagnosed.
In both of these studies, the combination of physicians’ low-probability clinical estimates plus negative D-dimer tests yielded acceptably low post-test probabilities. However, we should be cautious about generalizing these results: The North Carolina physicians previously had developed a structured decision rule for assessing clinical probability of PE, so their "unstructured" estimates reflect considerable experience in predicting PE. And the authors of the randomized trial paper point out that they enrolled too few patients with high probability of VTE to be confident about their conclusions in that group.
Reference:
Kline JA et al. Prospective study of the diagnostic accuracy of the Simplify D-dimer assay for pulmonary embolism in emergency department patients. Chest 2006 Jun; 129:1417-23.