BACKGROUND: The risk of stroke associated with aortic arch atheroma is proportional to plaque thickness, the accurate measurement of which may influence future therapeutic decisions. Transesophageal echocardiography is the procedure of choice for plaque measurement. This study examines the reproducibility of such measurements within the context of the Aortic Arch Related Cerebral Hazard Trial of antithrombotic therapy. METHODS: Atheroma images were remeasured by 3 independent blinded observers of varying expertise. Intraobserver and interobserver variability were calculated using Lin's concordance correlation coefficient (rho-c), the Bland-Altman method, and the kappa statistic. RESULTS: A total of 160 images were obtained from 96 patients (68% male; mean age 74 +/- 9 years). Mean plaque thickness was 5.1 +/- 2.3 mm, range 1.2 to 19.3 mm. For intraobserver variability, rho-c = 0.95, mean difference = 0.01 mm, and kappa = 0.71. For interobserver variability, rho-c ranged from 0.80 to 0.91; mean difference was 0.13 to 0.48 mm; and kappa was 0.61 to 0.69. By all statistical methods, agreement was at least substantial, although lower at greater plaque thicknesses. For the clinically relevant 4 mm, agreement was 84% to 88%. CONCLUSIONS: Transesophageal echocardiography measurement of aortic plaque is substantially reproducible across a range of observer experience and, thus, is a widely applicable, adequately reliable tool for clinical and research purposes.
BACKGROUND: The risk of stroke associated with aortic arch atheroma is proportional to plaque thickness, the accurate measurement of which may influence future therapeutic decisions. Transesophageal echocardiography is the procedure of choice for plaque measurement. This study examines the reproducibility of such measurements within the context of the Aortic Arch Related Cerebral Hazard Trial of antithrombotic therapy. METHODS: Atheroma images were remeasured by 3 independent blinded observers of varying expertise. Intraobserver and interobserver variability were calculated using Lin's concordance correlation coefficient (rho-c), the Bland-Altman method, and the kappa statistic. RESULTS: A total of 160 images were obtained from 96 patients (68% male; mean age 74 +/- 9 years). Mean plaque thickness was 5.1 +/- 2.3 mm, range 1.2 to 19.3 mm. For intraobserver variability, rho-c = 0.95, mean difference = 0.01 mm, and kappa = 0.71. For interobserver variability, rho-c ranged from 0.80 to 0.91; mean difference was 0.13 to 0.48 mm; and kappa was 0.61 to 0.69. By all statistical methods, agreement was at least substantial, although lower at greater plaque thicknesses. For the clinically relevant 4 mm, agreement was 84% to 88%. CONCLUSIONS: Transesophageal echocardiography measurement of aortic plaque is substantially reproducible across a range of observer experience and, thus, is a widely applicable, adequately reliable tool for clinical and research purposes.