Duplex Ultrasound Criteria for De®ning the Severity of Carotid Stenosis Konstantinos A. Filis, MD, Frank R. Arko, MD, Bonnie L. Johnson, RVT, Iraklis I. Pipinos, MD, E. John Harris, MD, Cornelius Olcott IV, MD, and Christopher K. Zarins, MD, Stanford, California Duplex 1 ultrasound scan DUS) criteria for grading >50% carotid artery stenosis is typically divided into broad categories such as 50-79% stenosis, 80-99% stenosis, and occlusion. The purpose of this study is to validate DUS criteria for stratifying 50 to 100% carotid stenosis into 10% intervals using digital substraction cerebral angiography DSCA) as the standard of com- parison. Between 1996 and 2001, 163 patients were evaluated with duplex ultrasound and angiography. A total of 326 carotid arteries were studied using DUS in an accredited ICAVL vascular laboratory. Threshold velocity criteria for determining the degree of carotid stenosis was de®ned according to seven categories: <50%, 50-59%, 60-69%, 70-79%, 80-89%, 90-99%, and occlusion. Treatment decisions were based on the angiographic ®ndings. In cases where the degree of stenosis as de®ned by duplex velocity criteria did not correlate with angio- graphically de®ned stenosis, each record was reviewed to determine whether the angiographic ®ndings altered the surgeon's treatment decision. The sensitivity, speci®city, positive predictive value PPV), and negative predictive value NPV) for DUS-de®ned degree of stenosis as compared to angiographically de®ned stenosis were determined. There was a high correlation R = 0.96) between duplex scan and angiography in 93% 302/326) of the cases. Clinical management was altered in only 3% 10/326) of the cases because of the results of angio- graphy. The DUS velocity criteria to grade the severity of carotid disease in 10% intervals is reliable and accurate. Clinical management of patients with carotid stenosis can be based solely on carotid DUS in 97% of patients considered for treatment of carotid artery disease. INTRODUCTION The value of the duplex ultrasound scan DUS) as the sole diagnostic study prior to carotid endarte- rectomy EA) is well established. However, the di- agnostic accuracy of carotid ultrasound reported by centers participating in the North American Carotid Endarterectomy Trial NASCET) 1 was low, with a sensitivity of 68% and a speci®city of 67%. Differ- ences in imaging devices and variations in protocols may have accounted for this low accuracy. In the Asymptomatic Carotid Atherosclerosis Study ACAS), 2 however, the speci®city of carotid ultra- sound was above 95% as a result of a standard protocol for carotid duplex imaging at each center. Accurate characterization of the severity and local- ization of the lesion allows careful patient selection and planning for CEA. To this end, digital substrac- tion cerebral angiography DSCA) is the standard means of carotid artery stenosis assessment used by practioners in the NASCET and the ACAS. 1,2 413 Division of Vascular Surgery, Department of Surgery, Stanford University Hospital, Stanford, CA. Correspondence to: C.K. Zarins, MD, Division of Vascular Surgery, Stanford University Medical Center, 300 Pasteur Drive, H-3642, Stan- ford, CA 94305-5642, USA, E-mail: zarins@stanford.edu Ann Vasc Surg 2002; 16: 413±421 DOI: 10.1007/s10016-001-0175-8 Ó Annals of Vascular Surgery Inc. Published online: 23 July 2002