Objective: Colour Doppler ultrasound is operator dependent, but it is unclear how much clinical impact this might have on patient referral for carotid endarterectomy. Our aim was to quantify the interobserver variability of Doppler ultrasound.
Methods: Consecutive patients attending for carotid Doppler ultrasound underwent two examinations on the same day, in random order, by two of three observers blind to each other's results. Severity of stenosis was assessed using standard velocity criteria and lesion appearance.
Results: A total of 189 patients were scanned (378 ICAs). Of the 134 ICAs scanned by observers 1 and 2, observer 1 classified 11 as 80-99% stenosis (operable), compared with nine by observer 2. Of the 206 ICAs scanned by observers 1 and 3, observer 1 classified 11 as 80-99% stenosis, compared with only five by observer 3. Of the 38 ICAs scanned by observers 2 and 3, observer 2 classified 2 as 80-99% stenosis compared with none by observer 3. Overall, clinical management would differ in 10/378 (3%) of ICAs, but in 10/22 (45%) of those considered operable by one of the three observers.
Conclusion: There was clinically important interobserver variability in the assessment of ICA disease, which could result in serious errors if endarterectomy were performed on the basis of a single Doppler ultrasound.