Dr. Rea's query concerns a common, clinically important question. Published data suggest that the presence of carotid bruit is a reasonably accurate indicator of carotid stenosis in symptomatic persons. Fewer data are available with which to assess its role in asymptomatic persons. In symptomatic persons, Ziegler and colleagues [1] found a sensitivity of 0.29 and a specificity of 0.88 for detecting stenosis greater than 50%; the collaborators of the North American Symptomatic Carotid Endarterectomy Trial [2] found a sensitivity of 0.62 and a specificity of 0.61 for detecting stenosis of 70% to 99%; and Hankey and Warlow [3] reported the most favorable of results, with a sensitivity of 0.76 and a specificity of 0.76 when a positive angiogram was defined as stenosis of 75% to 99%. However, the evaluation of patients with cerebrovascular symptoms will overestimate the accuracy of carotid bruit for detecting underlying carotid stenosis compared with evaluation in asymptomatic patients.

The group without neurologic symptoms that had the highest known prevalence of carotid stenosis was a Veterans Administration hospital population with cardiovascular disease [4]. This population reportedly had a 15% prevalence of carotid stenosis greater than 50% (the prevalence would be lower if only stenosis of 60% and greater was considered to indicate disease) [4]. If the test characteristics for carotid bruit reported by Hankey and Warlow are applied to a population with a 15% prevalence of carotid stenosis, the presence of carotid bruit increases the probability of carotid disease to 36%. This does not reach the threshold of 40% prevalence needed for further Doppler ultrasonographic screening to be considered cost-effective. If auscultation of carotid bruit under these conditions cannot increase the prevalence of carotid disease to the extent that it is cost-effective for further screening, then auscultation of carotid bruit is not likely to be cost-effective except in higher-risk populations that have been reported in the literature.

Although the presence of carotid bruit in patients with carotid-territory cerebrovascular symptoms increases the probability of an underlying high-grade carotid stenosis, the accuracy of using this marker is low. In asymptomatic persons, the presence of carotid bruit is even less useful and is less a reflection of underlying carotid disease than of nonspecific atherosclerotic cardiovascular disease [5].