Background: Patients with suspected pulmonary embolism often have nondiagnostic lung scans and may present in circumstances where lung scanning is unavailable. Levels of D-dimer, a fibrin-specific product, are increased in patients with acute thrombosis; this may simplify the diagnosis of pulmonary embolism.
Objective: To determine the sensitivity and specificity of a whole-blood D-dimer assay in patients with suspected pulmonary embolism and in subgroups of patients with low pretest probability of pulmonary embolism or nondiagnostic lung scans.
Design: Prospective cohort.
Setting: Four tertiary care hospitals.
Patients: 1177 consecutive patients with suspected pulmonary embolism.
Measurements: All patients underwent an assessment of pretest probability by use of a standardized clinical model, a D-dimer assay, ventilation-perfusion lung scanning, and bilateral compression ultrasonography. Patients in whom pulmonary embolism was not initially diagnosed were followed for 3 months. Accordingly, patients were categorized as positive or negative for pulmonary embolism.
Results: Of the 1177 patients, 197 (17%) were classified as positive for pulmonary embolism. Overall, the D-dimer assay showed a sensitivity of 84.8% and a specificity of 68.4%. In 703 patients (3.4%) with a low pretest probability of pulmonary embolism, the likelihood ratio associated with a negative D-dimer test result was 0.27, resulting in a posterior probability of 1.0% (95% CI, 0.3% to 2.2%). In 698 patients with nondiagnostic lung scans (previous probability, 7.4%), the likelihood ratio associated with a negative D-dimer test result was 0.36, resulting in a posterior probability of 2.8% (CI, 1.4% to 4.8%).
Conclusions: A normal D-dimer test result is useful in excluding pulmonary embolism in patients with a low pretest probability of pulmonary embolism or a nondiagnostic lung scan.