A step-by-step guide for validation of carotid duplex studies

Recent trials have demonstrated remarkable efficacy of carotid endarterectomy in stroke prophylaxis for patients identified by degree of internal carotid stenosis. Ironically, a recently published multicenter validation study comparing carotid Doppler ultrasonography with conventional contrast arteriography showed marked interlaboratory variability in sensitivity. It is therefore necessary for accuracy of carotid ultrasound to be established locally as a matter of quality assurance. We present here a step-by-step guide for data collection, interpretation, and presentation of results using model data from 39 patients (77 internal carotid arteries) who underwent both carotid duplex scanning and contrast arteriography at our institution. Calculations for deriving accuracy, sensitivity, specificity, as well as positive and negative predictive values are demonstrated. In addition, careful analysis of 'missed' ultrasound cases provides insights about problem circumstances such as: (1) overestimation of internal carotid artery (ICA) stenosis due to the presence of severe contralateral obstructive disease, (2) underestimation of ICA stenosis due to the presence of severe common carotid artery or innominate artery obstructive disease, (3) misidentification of the external carotid artery as the ICA, (4) erroneous identification of occlusion ('pseudoocclusion'), and (5) physician interpretation errors. In conclusion, determination of accuracy of carotid ultrasound and analysis of ultrasound 'misses' is an essential quality assurance activity for all laboratories.