Use of electron beam tomography data to develop models for prediction of hard coronary events

Background Prediction of hard cardiac events (myocardial infarction and cor onary death) remains difficult in spite of the identification of several re levant risk factors for the development of coronary artery disease (CAD). N ew indicators of risk might add to our predictive ability. We used measures of coronary artery calcification (CAC) found by electron beam tomography ( EBT) imaging to develop prediction models for hard cardiac events alone and in association with traditional risk factors for CAD. Methods Two groups of patients were studied: group A, 676 asymptomatic pati ents (mean age 52 years, 51% men) prospectively followed up For 32 +/- 7 mo nths after being referred by primary care physicians for a screening EBT, a nd group B, 10,122 asymptomatic patients screened by EBT at one center and used as controls for calculation of calcium score nomograms. Results The occurrence of hard events in group A patients was related to tr aditional risk factors for CAD, presence of CAC (score >0), Ln (1 + absolut e calcium score [CS]), and age- and sex-specific CS percentiles (CS%). Univ ariate analyses showed that age, smoking, diabetes mellitus, presence of CA C, in (1 + absolute CS), and CS% were predictive of hard events (all P < .0 5). Multiple logistic regression analyses demonstrated that CS% was the onl y significant predictor of events and provided incremental prognostic value when added to traditional risk factors for CAD (chi-square, P <.001). In a comparison of receiver-operator characteristic curves for prediction of ha rd events, the area under the curve for CS% plus conventional risk factors and age was significantly larger than that obtained by use of traditional r isk factors and age separately as predictors (0.84 vs 0.71, respectively, P < .001). Furthermore, the area under the curve of CS% alone was significan tly larger than that of traditional risk factors and age combined (0.82 vs 0.71, P = .028). Conclusions Patients are usually selected For EBT screening on the basis of the presence of conventional risk Factors for CAD. However, an age- and se x-specific calcium score provides the best predictive model for the occurre nce of hard coronary events and adds incremental prognostic information to conventional risk factors for CAD.