Objective: To determine whether cervical bruits, alone or combined with other clinical characteristics, can distinguish high-grade (70% to 99%) carotid artery stenoses from less severe stenoses in patients with symptoms of cerebrovascular disease.
Design: Cross-sectional comparison of clinical observations with contemporaneous angiography.
Setting: The North American Symptomatic Carotid Endarterectomy Trial (NASCET), a multicenter randomized controlled trial of carotid endarterectomy.
Patients: All patients enrolled in the NASCET from its inception in 1988 to November 1991.
Results: A focal ipsilateral carotid bruit had a sensitivity of 63% and a specificity of 61% for high-grade stenosis and, when absent, only lowered the probability for high-grade stenosis from a pretest value of 52% to a post-test probability of 40%. When combined with four other clinical characteristics (an infarction on computed tomography of the head, a carotid ultrasound scan suggesting more than 90% stenosis, a transient ischemic attack rather than a minor stroke as a qualifying event, and a retinal rather than a hemispheric qualifying event), the predicted probabilities of high-grade stenosis ranged from a low of 18% (when none of the features was present) to a high of 94% (when all the features were present.
Conclusions: Cervical bruits alone were not sufficiently predictive of high-grade symptomatic carotid stenosis to be useful in selecting patients for angiography; they were absent in over one third of patients with high-grade stenosis. When combined with other clinical variables, patients with high or low probabilities of 70% to 99% stenoses could be identified, but this prediction model still missed many individuals with high-grade stenosis, even in this training set of selected patients.