Objectives: A high degree of variability has been reported regarding the ultrasound-based assessment of flow-mediated dilatation. We wanted to investigate the variability and find out how it might be reduced most efficiently.
Design: Brachial artery flow-mediated dilatation was measured by high-resolution ultrasound in 22 healthy adults on two consecutive days. Two observers analysed all images twice. The total variance was split into variance components and estimated hierarchically using the method of restricted maximum likelihood.
Results: The relative proportional contributions from intraobserver (residual), interobserver, interpatient and interday variance components, with percentage dilatation as outcome variable, were 0.41, 0.18, 0.25, and 0.15, respectively.
Conclusions: The major source of variability when assessing flow-mediated dilatation was found to be intraobserver variability. The simplest way to reduce total variability is for the observer to average results from repeated image analyses. We suggest that three repetitions are sufficient. This will reduce the total variance by 30%.