Ultrasonography of Leg Veins in Patients Suspected of Having Pulmonary Embolism

TO THE EDITOR:

Turkstra and colleagues [1] found that the value of compression ultrasonography for detection of deep venous thrombosis in patients suspected of having pulmonary embolism is limited because of a sensitivity of 29%. We disagree with the authors' methods and their conclusion. In their article, the sensitivity of ultrasonography was estimated from ventilation-perfusion scans, which have never been proven to be the gold standard for diagnosing deep venous thrombosis and pulmonary embolism. For example, all high-probability lung scans in the study were considered to show a pulmonary embolism. It is well established that this is true only when there is also high clinical probability; when clinical probability is low, pulmonary embolism is confirmed by angiography in only 66% to 88% of cases [2, 3]. Thus, Turkstra and colleagues overestimated the diagnosis of pulmonary embolism; this clearly decreases the chance that ultrasonography would be useful for diagnosing pulmonary embolism.

In patients with pulmonary embolism proved by angiography, the sensitivity of ultrasonography remains low; however, the total population consisted of only 33 patients. In the same way, of the 178 patients in whom pulmonary embolism was excluded by lung scanning (n = 86) or angiography (n = 92), 5 patients with abnormal ultrasonography results were considered to have a false-positive ultrasonography result. It has also been shown that 6% of patients with normal lung scans and high or uncertain clinical probability have pulmonary embolism proven by angiography. Thus, it is possible that among the 86 patients in whom pulmonary embolism was excluded by lung scanning, some may actually have had a pulmonary embolism (especially in the case of an abnormal ultrasonography result).

Before Turkstra and colleagues' 5 patients with abnormal ultrasonography results (3 of whom had strong risk factors for thrombosis, such as surgery and history of pulmonary embolism) can be considered to have false-positive results, venography at least should be done. Of the 30 patients with uncertain lung scans (in whom pulmonary angiography was not done), 8 were subsequently treated with anticoagulants and 4 had an abnormal ultrasonography result. In this situation, in which the authors considered these 8 patients as having pulmonary embolism, the sensitivity of ultrasonography is 50%. How can the authors claim that the sensitivity of ultrasonography is higher with uncertain lung scans than with high-probability scans? Otherwise, how do the authors explain the difference in the prevalence of pulmonary embolism between the group in which ultrasonography was not done (maximum, 8 of 40 [20%]) and the group in which ultrasonography was done ([149 + 8]/357 = 44%; P = 0.003)? Finally, because Turkstra and colleagues examined only common femoral and popliteal veins with ultrasonography, the title of their article should be “Diagnostic Utility of Compression Ultrasonography of Popliteal and Common Femoral Veins … .”

• Copyright ©2004 by the American College of Physicians