Intersonographer Reproducibility and Intermethod Variability of Ultrasound Measurements of Carotid Artery Stenosis: The Tromsø Study

• Published in: Cerebrovascular Diseases
• Main Authors: Mathiesen, Ellisiv B., Joakimsen, Oddmund, Bønaa, Kaare H.
• Format: Article in Journal/Newspaper
• Language: English
• Published: S. Karger AG 2000
• Subjects: Tromsø
• Online Access: http://dx.doi.org/10.1159/000016058; https://www.karger.com/Article/Pdf/16058

Background and Purpose: Knowledge of the reproducibility of a diagnostic method is important in order to evaluate its usefulness. Few studies have examined interobserver and intermethod agreement on ultrasound measurements of carotid stenosis. Methods: Intersonographer agreement on ultrasound measurements of carotid plaque morphology and the estimated degree of stenosis by three ultrasound methods were assessed in a random sample of 51 participants with stenotic carotid arteries selected from a population health survey. The degree of stenosis was assessed by measurements of velocity, lumen diameter reduction and cross-sectional lumen area. Intermethod agreement on the degree of carotid stenosis was also assessed. Results: Agreement on plaque echogenicity and heterogeneity was moderate (κ = 0.56 and κ = 0.60, respectively). The mean degree of stenosis and median absolute difference between observers of the estimated degree of stenosis by the velocity method were 46.3 and 10.8%, respectively. The corresponding values were 51.0 and 5.8% for the diameter method, and 57.1 and 7.2%, for the cross-sectional lumen method. The limits of agreement for intersonographer reproducibility varied between ±19.7 and 26.5%. For all methods, reproducibility increased with increasing degree of stenosis. Differences between the methods were large in low-grade stenosis but were acceptable in high-grade stenosis. Conclusions: Considerable differences in ultrasound measurement of stenosis, which could lead to different clinical conclusions, were regularly encountered no matter what ultrasound method was used.