Low accuracy of color Doppler ultrasound in the detection of proximal leg vein thrombosis in asymptomatic high-risk patients

Objective: To evaluate the ability of color Doppler ultrasound to detect proximal deep venous thrombosis (DVT) in asymptomatic high-risk patients who subsequently underwent contrast venography. Design: Prospective cohort study using blinded observers, with contrast venography as the comparison standard. Setting: Seven medical centers (university and community hospitals) participating in a clinical trial of low-molecular-weight heparin for prevention of DVT. Patients: A total of 385 consecutive patients undergoing elective unilateral hip or knee replacement. Measurements: Ten days after surgery or before hospital discharge (whichever occurred first), patients had bilateral color Doppler ultrasound examinations of the proximal veins of the lower extremities. Subsequently, a contrast venogram of the operated leg was obtained. Results: Color Doppler ultrasound studies and venograms were both evaluable in 319 of the 385 patients: Deep venous thrombosis was identified by contrast venography in 80 patients (prevalence, 25%; 95% CI, 20% to 30%) and involved the proximal veins in 21 patients (prevalence, 7%; CI, 4% to 10%). For proximal DVT, color Doppler ultrasound showed poor sensitivity (38%; CI, 18% to 62%), moderately good specificity (92%; CI, 89% to 95%), and a poor positive predictive value for this population (26%). Conclusion: Color Doppler ultrasound examinations are insensitive to proximal DVT in asymptomatic high-risk patients and should not be substituted for venography for identifying proximal DVT in such patients.