Background and purpose: To prevent strokes it is essential to correctly classify people according to their risk of stroke. The aim of the present study was to assess whether carotid ultrasound improves the stroke risk prediction in asymptomatic individuals.
Methods: The baseline visit of the Carotid Atherosclerosis Progression Study (CAPS) included assessment of conventional risk factors and carotid ultrasound. During the 10-year follow-up of 4995 subjects, strokes, transient ischaemic attacks (TIA) and deaths were recorded. We assessed the additional usefulness of carotid ultrasound compared to the Framingham Stroke Risk Score (FSRS) with reclassification statistics using four risk categories.
Results: Most risk models were not improved by carotid ultrasound. For individual stroke prediction, intima-media thickness (IMT) or plaque of the internal carotid arteries were more useful than common carotid or bifurcational IMT. The model predicting 'any stroke or death' was significantly improved when ultrasound parameters were included - 339 subjects (7.2%) were reclassified to another risk category (122 were shifted to a higher, 217 to a lower risk category); 182 (53.7%) were correctly reclassified. The net reclassification improvement (NRI) was 7.7% (p = 0.029) and the integrated discrimination improvement (IDI) was 0.73% (p = 0.023).
Conclusions: When carotid ultrasound is not restricted to the common carotid artery but includes the internal carotid segments, the inclusion of ultrasound data into stroke risk models may improve the risk classification of individuals. Further validation in primary prevention cohorts is warranted.
Keywords: Carotid arteries; atherosclerosis; individual risk prediction; intima-media thickness; plaque; reclassification; stroke.