Purpose: To critically evaluate the accuracy, advantages, and drawbacks of duplex ultrasound as a diagnostic test for proximal deep-vein thrombosis.
Data identification: An English-language search using MEDLINE (1980 to 1988) and bibliographies from articles, and a hand search of pertinent radiology and ultrasound journals from 1988.
Study selection: All series comparing duplex ultrasound to the reference standard, contrast venography, were reviewed and classified into levels based on the quality of study design.
Data extraction: Results of duplex ultrasound compared with venography in the proximal deep venous system, technical problems encountered, frequency of diagnosis of other causes of leg swelling, and frequency of unsuccessful or inconclusive studies were collated.
Results of data synthesis: Four well-designed studies reported similar results. The sensitivity of duplex ultrasound in detecting proximal thrombi ranged from 92% to 95% with a combined sensitivity of 93% (CI, 88% to 98%), and the specificity ranged from 97% to 100% with a combined sensitivity of 98% (CI, 96% to 100%). Similar findings were noted in nine other studies that had minor methodologic flaws. Four studies reported that ultrasound was able to identify a nonthrombotic cause of leg swelling in 5% to 15% of cases. Four studies found that duplex ultrasound was inconclusive in 1% to 6% of cases, with a combined frequency of 2%.
Conclusions: Duplex ultrasound appears to be very accurate in the detection of acute proximal deep-vein thrombosis. This test has major advantages as well as certain limitations compared with other diagnostic methods.