Measurement variability in duplex scan assessment of carotid atherosclerosis.

BACKGROUND AND PURPOSE: The reproducibility of duplex scan measures of carotid atherosclerosis was evaluated as part of a study assessing the prevalence of carotid disease in elderly adults. METHODS: Doppler measures of blood flow velocity were used to evaluate disease severity, and extent of carotid plaque was scored from the B-mode image. A reader assigned a grade from 0 to 3 to each of seven segments in the carotid system, based on the number and size of lesions present. Reproducibility data were obtained from 30 study participants who underwent a repeat scan by a second sonographer. Each scan was then scored by two readers. RESULTS: Doppler measures of blood flow velocity were found to be highly reproducible, with intraclass correlation coefficients of 0.81 for the common carotid artery, 0.84 for the internal carotid artery, and 0.77 for the internal carotid artery velocity to common carotid artery velocity ratio. Reproducibility of plaque grade was evaluated using segment as the unit of analysis, and both sonographer and reader variation were analyzed. When readers differed perfect agreement was achieved in 84% of the segments (K = 0.67), and when sonographers differed perfect agreement was obtained in 78% of the segments (K = 0.56). When both sonographer and reader differed, perfect agreement was obtained in 77% of the segments (K = 0.53). The plaque index, created by summing plaque grades from selected segments, was highly reproducible, with an intraclass correlation coefficient of 0.86. CONCLUSIONS: The duplex scan protocol described here provides reliable measures of both extent and severity of carotid disease that are appropriate for use in cross-sectional studies.