Evaluating institutional variability of duplex scanning in the detection of carotid artery stenosis

Purpose: Duplex scanning is widely used to measure carotid artery stenosis, but the issue of variability between institutions must be raised. To examine for this potential variability, we evaluated two ATL Ultramark 9 duplex machines at two hospitals within our institution. Methods: Stenosis was calculated angiographically as in the North American Symptomatic Carotid Endarterectomy Trial (NASCET). Internal carotid artery (ICA) to common carotid artery (CCA) peak systolic velocity (PSV) ratios were determined using the two ATL machines. Receiver operating characteristic (ROC) curves were generated for each hospital. Results: ROC curves varied significantly. The area under the curve from hospital 1 was 0.95 (SE = 0.021); the area under the curve from hospital 2 was 0.87 (SE = 0.034). The curves were essentially statistically different (P = 0.0516). Comments: We showed a difference in predicting diameter stenosis utilizing ICA PSV/CCA PSV ratios at two hospitals within our institution. Variability is likely due to technologists, since equipment and physician interpreters were the same. Criteria must be developed for each different machine in each laboratory. Quality assurance is critical to minimize variability.