Melanie B. Schernthaner MD, Presenter: Nothing to Disclose

Matthew Benenati, Abstract Co-Author: Nothing to Disclose

Reza Rajebi MD, Abstract Co-Author: Nothing to Disclose

Gail Walker PhD, Abstract Co-Author: Nothing to Disclose

Constantino Santiago Pena MD, Abstract Co-Author: Speakers Bureau, W. L. Gore & Associates, Inc Speakers Bureau, Cook Group Incorporated Speakers Bureau, Koninklijke Philips NV Advisory Board, C. R. Bard, Inc Advisory Board, Boston Scientific Corporation Advisory Board, Guerbet SA

To evaluate accuracy, inter-observer and intra-observer reliability of simple visual estimation (SVE) in grading peripheral arterial stenosis compared to calibrated measurements.

23 interventionalists with a wide range of experience (1-30 years) and subspecialty training (IR (13), Neuro-IR (2), interventional-cardiology (4) and vascular surgery (4)) reviewed 42 angiographic images of peripheral and carotid arteries in two sessions. Images where shuffled between readings. An independent team measured all lesions using manual calipers. A +/- 5% error was considered as threshold for accurate visual estimation. Lesions were categorized by clinical significance (80% severe). SVE was compared for agreement by weighted kappa statistics. Reliability was assessed by intraclass correlation.

Overall accuracy of SVE in grading stenosis was 28.3% and 27.4% for the two assessments. Errors in excess of +/- 5% occurred in 71.7% and 72.6% respectively. Agreement with respect to clinical category was fair with a weighted kappa of 0.579 in the first testing session and 0.588 in the second. 92.6% and 93.8% of severe lesions, 40.9% and 41.5% of significant lesions and 71.5% and 73.3% of insignificant lesions were correctly identified in the first and second sessions respectively. In the first session 53.0% of significant and 4.4% of insignificant lesions were categorized as severe stenosis. 49.9% of significant lesions and 4.6% of insignificant lesions were overestimated as severe in the second session. Intra-rater reliability was good (0.990) and inter-rater reliability was fair for assessment of peripheral arteries (0.823, 0.809), and carotids (0.748, 0.708). Accuracy did not differ in relation to years of experience or specialty.

Despite good intra-observer reliability, inter-observer reliability was fair. Estimation of peripheral arterial stenosis often results in overestimation of stenosis, most pronounced in the 60-80% range. There were no significant differences based on years of experience in practice or specialty. Visual estimates of stenosis potentially lead to therapeutic decisions based on inaccurate information.

Clinical decision making should be based on caliper measurements especially in non-significant stenoses.

Schernthaner, M, Benenati, M, Rajebi, R, Walker, G, Pena, C, Accuracy of Simple Visual Estimation in Grading Peripheral Arterial Stenosis—Is Eyeballing Enough?.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14009458.html