Variability in measurement of specific parameters for carotid duplex examination

The variability of four carotid artery frequency parameters used for classifying disease with duplex scanning was prospectively studied. Forty-eight patients (94 patent carotid arteries) were each examined by two technologists. Measured parameters were the peak systolic frequency (PSF) and the first zero slope from the common carotid artery, and the PSF and end diastolic frequency (EDF) from the internal carotid artery. Measurements from all the examinations were made twice by each technologist. Interobserver, intraobserver, and interpatient variability in measurement of the first zero slope was so great that we have abandoned its use. Measurement of variability for PSF and EDF was much less (correlation coefficients 0.68 to 0.92). These parameters were measured with sufficient precision to warrant their continued use for important decision steps in classifying carotid artery disease. Interpatient differences in PSF sufficient to cause disagreement regarding the hemodynamic significance of carotid disease occurred in only three instances. In each of these cases the differences were due to examination technique (failure to identify a very distal internal carotid artery stenosis, difficulty distinguishing between a kink and a stenosis, and failure to recognize an improper Doppler angle). We conclude that the variability of PSF and EDF is within clinically acceptable levels and is mainly due to examination technique rather than measurement of waveform parameters or changes in patient hemodynamics.