Carotid stenosis by digital subtraction angiography: reproducibility of the European Carotid Surgery Trial and the North American Symptomatic Carotid Endarterectomy Trial measurement methods and visual interpretation.

PURPOSE: To evaluate different carotid stenosis estimation methods with digital subtraction angiography.
METHODS: We assessed the intraobserver reproducibilty and interobserver variability of visual interpretation and the measurement methods used by the European Carotid Surgery Trial and the North American Symptomatic Carotid Endarterectomy Trial. Angiographic stenosis measurements according to both criteria were performed twice by a radiologist, a neurologist, and a vascular surgeon. Eighty bifurcations of consecutive symptomatic patients underwent 480 pairs of measurements. In addition, four radiologists estimated the stenoses visually.
RESULTS: Intraobserver consistency was slightly better by the European (kappa, 0.86 to 0.94) than by the North American (kappa, 0.68 to 0.91) trial criteria or by visual interpretation (kappa, 0.79 to 0.81). No significant interobserver variability was found, except in the subgroup of mild stenoses by the North American Trial criteria. By kappa statistic, the interobserver agreement was excellent by the European trial method (kappa, 0.72 to 0.86), good by the North American trial method (kappa, 0.59 to 0.77), and good to excellent by visual evaluation (kappa, 0.68 to 0.88). The visual estimation agreed more closely with the European (kappa, 0.73 to 0.92) than with the North American trial (kappa, 0.55 to 0.74) criteria measurements.
CONCLUSIONS: All three methods have good reproducibility in digital subtraction angiography. Interobserver differences become more important in the estimation of mild stenosis.