Variance components analysis of carotid and femoral intima-media thickness measurements

Abstract

B-mode ultrasound intima-media thickness (IMT) measurements of carotid and femoral arterial walls are used in atherosclerosis studies. In this study, the components contributing to IMT measurement variability in males with coronary artery disease were investigated by means of repeated B-mode ultrasound scans and repeated off-line video image analyses. For statistical analysis, a mixed-model analysis of variance was used. From sonographer data, it was shown that human subjects and their arterial wall segments contributed 75% of the total IMT measurement variability in this population. Inter-sonographer variance contributed 25%. The intra-sonographer variance was negligible (<1%). In off-line image analysis, variance components due to subjects and segments, inter-analyst variance, and residual fluctuation were 88%, <1% and 11%, respectively. Intra-analyst variance was negligible (<1%). The major source of B-mode ultrasound IMT measurement variability finds its origin in subjects and their arterial walls. Although sonographers proved a lesser source of variability, in comparative studies they should enter a study well trained and should be randomly assigned to subjects. Follow-up examinations should preferably be done by the same sonographer. Off-line image analysis contributed little to IMT measurement variability.