Reliability and accuracy of simple visual estimation in assessment of peripheral arterial stenosis

To evaluate reliability, accuracy, and agreement of simple visual estimation (SVE) in determining the degree of peripheral arterial stenosis compared with calibrated measurements.

In 2 sessions, 23 interventionists with a wide range of experience and subspecialty training reviewed 42 angiographic images of lower extremity and carotid arteries (21 iliofemoral arteries and 21 carotid arteries). An independent physician measured all lesions using manual calipers. Intrarater and interrater reliability were assessed by intraclass correlation. A ± 5% error was considered the threshold for accuracy, and weighted κ statistics were computed to assess agreement with respect to the degree of stenosis (< 50%, nonsignificant; 50%-80%, significant; > 80%, severe).

Intrarater reliability of SVE was 0.99, and interrater reliability was 0.83. Accuracy varied from 52.8% for images of severe stenosis to 26.5% and 18.1% for significant and nonsignificant stenosis, respectively (P < .001). Agreement between SVE and caliper with regard to degree of stenosis was good (weighted κ 0.56) overall with correct classification ranging from 92.6% for severe stenosis to 53.4% and 68.2% for significant and nonsignificant stenosis, respectively (P < .001). Misclassification of nonsignificant and significant stenosis was more frequent for carotid arteries than for lower extremities.

Despite high reliability, SVE of peripheral arterial stenosis has limited accuracy in determining the exact degree of stenosis. Although severe stenosis is readily identified by SVE, arterial stenosis of < 80% is frequently overestimated, especially for carotid arteries, and should be confirmed by caliper assessment.