Identification of internal carotid artery stenosis in coronary artery bypass candidates.

Identification of stenotic internal carotid arteries in neurologically asymptomatic coronary artery bypass candidates is difficult. We correlated carotid arteriograms with carotid bruits in 422 vessels and demonstrated a sensitivity of 79% and a specificity of 65% with regard to the bruit's identification of vessels with a 50% or more reduction in angiographic diameter. The noninvasive duplex scan, used in conjunction with Fast Fourier Transform spectral analysis, had a sensitivity and a specificity of 90% in this same regard. Since there is only a 6% incidence of significant internal carotid stenosis in bypass candidates, Bayes' theorem used in conjunction with our findings showed that a carotid bruit was an incorrect predictor of internal carotid stenosis 87% of the time. The duplex scan was somewhat better, but still inaccurate 64% of the time. Thus neither parameter alone has a high enough predictive value to determine whether preoperative carotid angiography is necessary. However, if one limits duplex scanning to patients with a carotid bruit, the predictive value is raised to 92%. At the same time, there is minimal increase in the number of undiagnosed patients. Thus far this combined approach is recommended for preoperative evaluation of patients scheduled for coronary artery bypass.