Abstract 17169: Carotid Plaque Improves Discrimination and Reclassification of Cardiovascular Risk Prediction Models
Background: Ultrasound measured maximum intima-media thickness (IMT) of the internal carotid artery (ICA) is a measure of plaque. Its impact on performance of cardiovascular disease (CVD) risk prediction models as measured by improvement in discrimination and net reclassification is unknown.
Materials and methods: Standard CVD risk factors and maximum ICA IMT > 1.5 mm (plaque) were measured in 2946 members of the Framingham offspring cohort. CVD outcomes were evaluated for an average follow-up of 7.2 years. Two multivariate Cox-proportional hazards models were generated: one with standard risk factors and the other with standard risk factors plus plaque. Change in c-statistic and net reclassification improvement (NRI) of non-incident/incident cases with 3 and 4 risk categories were evaluated. Risk thresholds were 6% and 20% in the case of 3 categories and 5%, 10% and 20% in the case of 4 categories.
Results: On follow-up, 297 CVD events occurred. The adjusted hazard ratio (HR) was 1.915 (95% CI: 1.485, 2.469) for those with plaque compared to those without. The standard risk factor model had a c-statistic of 0.748 (95% CI: 0.719, 0.776), increasing significantly (p = 0.02) to 0.761 (95% CI: 0.734, 0.789) with plaque added. NRI was significant when either 3 (NRI = 7.4%, p = 0.011, 4.2% in events, 3.1% in non-events; see table) or 4 (NRI = 8.2%, p = 0.014, 2.7% in events, 5.5% in non-events) categories were used.
Conclusion: Plaque, measured as maximum ICA IMT >1.5 mm, improves prediction of CVD and significantly reclassifies events and non-events, whether 3 or 4 risk score categories are used. Higher value when 4 categories are used suggests that increasing number of categories may lead to inflation of NRI.
- © 2010 by American Heart Association, Inc.