Is venous duplex imaging an appropriate initial screening test for patients with suspected pulmonary embolism?

Noninvasive tests for deep venous thrombosis (DVT) are helpful in evaluating patients with suspected pulmonary embolism (PE) who have non-high-probability ventilation/perfusion (V/Q) lung scans. Based on the enthusiasm for these noninvasive tests, venous duplex imaging (VDI) has evolved as the initial screening test for patients with clinically suspected PE in some centers. This study evaluates the utility of VDI as the initial test in a diagnostic algorithm for patients with suspected PE. A total of 306 consecutive patients who underwent VDI as the initial screening test for clinically suspected PE during the past 24 months were reviewed; 121 patients were subsequently evaluated with V/Q scans and 20 underwent pulmonary arteriography. VDI demonstrated DVT in 10% (23/216), with 7% (22/306) having proximal DVT and 3% (9/306) having isolated calf DVT. In 25 patients with unilateral leg symptoms, DVT was found in 40% (10/25); however, among the 281 without unilateral leg symptoms, results of VDI were abnormal in only 5% (15/281). V/Q scans were obtained in 40% (121/306), with only 16% (19/121) of scans showing a high probability of PE. DVT was found in 25% (5/19) of patients with high-probability V/Q scans and in 25% (26/102) with non-high-probability scans. In patients with clinically suspected PE the incidence of detectable infrainguinal DVT is low. VDI appears to be a reasonable initial screening test in patients with clinically suspected PE and unilateral leg symptoms. However, in patients without unilateral leg symptoms, the diagnostic yield is low and an alternative diagnostic approach appears justified.