Ultrasound for the diagnosis of deep vein thrombosis: where to now?

BMJ 1998; 316 doi: https://doi.org/10.1136/bmj.316.7124.2 (Published 03 January 1998) Cite this as: BMJ 1998;316:2

A new protocol for diagnosis and treatment

  1. Bruce L Davidson, Associate professor of medicinea (brucedavidson{at}pobox.com),
  2. Eric J Deppert, Director, walk in clinica
  1. a Department of Medicine, Allegheny University Hospitals-Graduate, 301 South 19th Street, Philadelphia, PA 19103, USA

    In this issue Cogo and others from Italy, Canada, and the Netherlands describe using two ultrasound examinations a week apart in 1702 outpatients with suspected deep vein thrombosis of the leg to determine whom to treat (p 17).1 Four hundred and twelve were diagnosed with thrombosis by this procedure; the rest were observed without treatment for six months. Nine of the untreated patients developed proof of thromboembolic disease during follow up: two had pulmonary embolism, fatal in one. Overall, fewer than 1% of patients presenting with possible deep vein thrombosis were missed with the authors' approach and only one (<0.1% of patients enrolled) paid with his life. Is the authors' protocol now ready for routine use? Is this the requiem for the venogram?

    The protocol was straightforward. The authors used the least expensive type of ultrasound imaging: grey scale real time ultrasound. Doppler signals, manoeuvres to change venous flow, and colour were not used. While Doppler and colour technology are essential for measuring valvular regurgitation echocardiographically and certain other applications, they appear unnecessary in the authors' hands for identifying deep vein thrombosis requiring treatment.

    The authors used full compressibility of the vein—that is, complete obliteration of the vein lumen as visualised on the ultrasound screen after pressure with the transducer probe—as the sole criterion for clot. They have previously validated this technique.2 They compressed each leg at just three sites: the common femoral vein next to the artery at the femoral ligament, the popliteal vein next to the artery behind the knee, and the same vein about eight centimetres distal, where it trifurcates into smaller calf veins. These manoeuvres can be easily and quickly learnt, although interpretation of the result at the distal popliteal vein is tricky. The first two sites are readily identifiable but inclusion of the third site was a mixed blessing: it allowed detection of more thromboses (thereby reducing the need for follow up studies from two to one3), but caused false positive readings, reducing the specificity of the result and the positive predictive value of clot detection at this distal site. Failure of compressibility at the first two sites was proved to be thrombus by contrast venography 99% of the time but at the third site only 79% of the time. While some doctors may follow symptomatic distal popliteal vein clots for propagation before treating, many will anticoagulate. If these results are generalisable, about 20% of patients with abnormal compressibility limited to the distal popliteal vein will be falsely positive and unnecessarily subjected to anticoagulation. Fortunately, a compression ultrasound examination abnormal solely at this distal site was rare, occurring in only 23 (1.4%) patients.

    Only 12 (3%) of the 412 patients with abnormal ultrasound examinations were picked up at the second test a week after presentation; by then, 400 patients with abnormal tests had already been identified. Is it worth bringing 1300 patients back to find thrombi in 12 (roughly 1%)? We think it is. A follow up visit a week after deep vein thrombosis is suspected is reasonable, and the consequences of missing those 12 patients' disease are worrisome. The cost of each ultrasound examination should not be high, far less than that of venography or empirical anticoagulation.

    The study results appear valid. Only 0.7% of patients studied slipped through the two ultrasound examinations to present with thrombosis or embolism. Even if we assume that the eight protocol violators (one patient with venous thrombosis found at elective venography and seven given anticoagulation without objective evidence of thrombosis) all truly had thrombosis and had been missed by the second examination, the failure rate of the protocol would still be only 1.4% over six months.

    These results may not be generalisable to patients excluded from the study, especially pregnant patients and those with previous leg thromboses. Recurrent leg thromboses are not uncommon.4 5 Nevertheless, were low cost ultrasound devices,6 the limited expertise required for these simple examinations, and low molecular weight heparin for treatment7 8 9 10 all readily available to practitioners, this new paradigm for diagnosis and treatment would represent a profound change in care, to the advantage of many thousands of patients worldwide. Diagnosis would be simple and office based, and treatment home based, with great convenience, savings, and no apparent loss in safety.

    All this good news does not represent the requiem for the venogram. Contrast venography will remain essential for the foreseeable future for diagnosing symptomatic calf thromboses that do not extend proximally, previously thrombosed legs again symptomatic, or asymptomatic leg thromboses in patients with or at risk of pulmonary embolism11 and for evaluating the efficacy of antithrombotic and thrombolytic drugs and of new imaging and patient management techniques (like the Cogo study). Promising uses of ultrasound contrast agents are being investigated in organ imaging, but there is little firm progress to report on from contrast ultrasound imaging of the deep veins.12 This study indeed does represent a highly useful and generalisable advance for a large segment of our patients, but we still have a long way to go.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    View Abstract