Article type
Year
Abstract
Background: vascular ultrasound (VU) is a non-invasive test for extracranial carotid stenosis (ECS) imaging with better risk-cost benefit when compared to magnetic resonance angiography (MRA), computed tomography angiography (CTA) and digital subtraction angiography (DSA). However, VU accuracy is not well established.
Objectives: we aimed to establish the accuracy of VU (index test) for ECS imaging compared with CTA as a reference test.
Methods: we performed a retrospective cohort study comparing all carotid VU performed in a university centre, during 2016 and 2017. Inclusion criteria were patients that had undergone VU and CTA within three months of the interval. Exclusion criteria were anything that could confuse the analysis. Four credentialed expert physicians (two examiners for VU and two for CTA) evaluated the images. They were blinded to the original report of the test, results of the other examiners, and results of the other method. We stored data in an electronic database, with restricted access, but which may be provided on request, for instance in future systematic reviews. The criteria used for CTA analysis was the NASCET method and for VU analysis was Radiology Consensus on duplex ultrasound (Grant et al. 2003). We compared results according to Grant et al. 2003's six levels of stenosis. Cut-offs for accuracy calculation was stenosis of 50% or more. The unit of analysis was the patient but when necessary we considered arteries separately.
Results: we included 50 patients. We excluded four patients for non-atherosclerotic disease and excluded five arteries due to previous stenting or carotid dissection. General accuracy of VU was 77%. There was 3% discordance between VU evaluators and 13% between CTA evaluators. VU and CTA had general discordance of 23%. We present the profile of sample accuracy table 1 and the epidemiological data in table 2.
Conclusion: the accuracy of VU in diagnosing ECS in our service (77%) was lower than high-quality international institutions, which usually is over 80%. The main limitation of our results is that physicians in training performed all VU. However, even when performed by a doctor in training, VU has good accuracy and it is the modality of choice for initial evaluation of carotid artery disease. Besides that, the CTA examination was more dependent on the examiner than VU.
Patient or healthcare consumer involvement: VU is largely used worldwide but its accuracy is not well established. The diagnostic value regarding VU for ECS could aid to define if some patients should undergo a carotid endarterectomy based on VU scanning alone. It will also contribute to the best decision making when a patient cannot undergo CTA, MRA or DSA, but would benefit from carotid revascularization.
Objectives: we aimed to establish the accuracy of VU (index test) for ECS imaging compared with CTA as a reference test.
Methods: we performed a retrospective cohort study comparing all carotid VU performed in a university centre, during 2016 and 2017. Inclusion criteria were patients that had undergone VU and CTA within three months of the interval. Exclusion criteria were anything that could confuse the analysis. Four credentialed expert physicians (two examiners for VU and two for CTA) evaluated the images. They were blinded to the original report of the test, results of the other examiners, and results of the other method. We stored data in an electronic database, with restricted access, but which may be provided on request, for instance in future systematic reviews. The criteria used for CTA analysis was the NASCET method and for VU analysis was Radiology Consensus on duplex ultrasound (Grant et al. 2003). We compared results according to Grant et al. 2003's six levels of stenosis. Cut-offs for accuracy calculation was stenosis of 50% or more. The unit of analysis was the patient but when necessary we considered arteries separately.
Results: we included 50 patients. We excluded four patients for non-atherosclerotic disease and excluded five arteries due to previous stenting or carotid dissection. General accuracy of VU was 77%. There was 3% discordance between VU evaluators and 13% between CTA evaluators. VU and CTA had general discordance of 23%. We present the profile of sample accuracy table 1 and the epidemiological data in table 2.
Conclusion: the accuracy of VU in diagnosing ECS in our service (77%) was lower than high-quality international institutions, which usually is over 80%. The main limitation of our results is that physicians in training performed all VU. However, even when performed by a doctor in training, VU has good accuracy and it is the modality of choice for initial evaluation of carotid artery disease. Besides that, the CTA examination was more dependent on the examiner than VU.
Patient or healthcare consumer involvement: VU is largely used worldwide but its accuracy is not well established. The diagnostic value regarding VU for ECS could aid to define if some patients should undergo a carotid endarterectomy based on VU scanning alone. It will also contribute to the best decision making when a patient cannot undergo CTA, MRA or DSA, but would benefit from carotid revascularization.
PDF