Background: The standard diagnostic approach in patients suspected of having pulmonary embolism starts with perfusion-ventilation lung scanning. If the resulting scan is not diagnostic, pulmonary angiography should be done. The use of tests for deep venous thrombosis has been advocated as an adjunct to establishing the diagnosis of pulmonary embolism, but no prospective studies have provided adequate information about the value of these tests.
Objective: To determine the accuracy and potential clinical utility of compression ultrasonography in the diagnosis of pulmonary embolism.
Design: Prospective cohort study with blinded assessment of ultrasonographic results.
Setting: Teaching hospital.
Patients: 397 consecutive inpatients and outpatients in whom pulmonary embolism was clinically suspected.
Measurements: Sensitivity and specificity of compression ultrasonography. Perfusion-ventilation scanning and angiography were the conjoint gold standard for determining the presence or absence of pulmonary embolism. Also calculated were the number of angiograms and lung scans avoided and the number of patients unnecessarily treated when compression ultrasonography was included in the diagnostic strategy.
Results: The overall sensitivity of compression ultrasonography for deep venous thrombosis in patients with pulmonary embolism was 29% (95% CI 22% to 37%); the specificity was 97% (CI, 94% to 99%). Adding ultrasonography to the diagnostic approach before lung scanning would avoid approximately 14% of lung scans and 9% of angiograms but would lead to unnecessary treatment of 13% of patients who have an abnormal ultrasonographic result (2% to 4% of all those receiving anticoagulation). When compression ultrasonography is done only in patients with a nondiagnostic lung scan, 9% of angiographies are prevented at the cost of unnecessarily treating 26% of patients who have an abnormal ultrasonographic result (2% of all patients receiving anticoagulation).
Conclusion: The diagnostic value of compression ultrasonography for the detection of deep venous thrombosis in patients suspected of having pulmonary embolism is limited; the gain in diagnostic efficiency obtained through the use of ultrasonography may be offset by a loss in diagnostic accuracy.