Abstract

The value of a coronary artery disease prediction algorithm, the Framingham risk score (score), for detecting coronary artery calcium (CAC) was examined in 385 men and 472 women, aged 29 to 43 years. Scores were compared in subjects with and without CAC and were also used to predict presence of CAC. Receiver-operating characteristic curves were computed to compare different prediction models. The score model was compared with age only, natural logarithm of body mass index (lnBMI) only, and score plus lnBMI models. CAC was detected in 30% of men and 16% of women. The mean score was significantly higher in men and women with CAC. For every 2-point increase in the score, the odds of CAC increased by 30% in women and 20% in men. Significant associations between CAC status and risk factors were observed for age in women, and high- density lipoprotein cholesterol and blood pressure in men and women. The area under the receiver-operating characteristic curve for the score was 0.67 and 0.57 for women and men, respectively. When lnBMI was added to the score model, the area increased to 0.76 in women (lnBMI p <0.0001, score p <0.005). For men, the area increased from 0.57 to 0.67, and the score was no longer significant (p >0.60) in the model with lnBMI (p <0.0001). Score predicts CAC in asymptomatic young adults. Inclusion of lnBMI in the score model adds significantly to the prediction of CAC in women and men. The lnBMI model has a greater predictive value than the score in this young population.