Abstract 19907: Clinical Prediction Risk Scores Systematically Underestimate the Presence and Extent of Subclinical Atherosclerosis in Brazilian Women: The Brazilian Study of Longitudinal Health (ELSA-Brasil)
Introduction: Different scores were developed to predict cardiovascular events in different populations. Due to genetic, social and economical differences, their performances vary according to the population to which they are applied. While there is concern that some scores may underestimate risk in women, few data exist for the Brazilian population.
Hypothesis: Clinical risk scores will underestimate the coronary artery calcium (CAC) in Brazilian women.
Methods: In a substudy from the ELSA-Brasil, a multicenter prospective study that enrolled civil servants in Brazil, 4546 participants underwent a CAC score. We calculated the Framingham risk score, Reynolds score, ASSIGN, PROCAM and both the low and the high European SCORE. To test the ability of each score to predict CAC values, we used a zero inflated model, which accommodates better CAC distribution. This is a two part model; the discrete part evaluates the association of the predictor with the probability of a CAC=0, and the continuous part of the model evaluates the association of the score with the CAC value.
Results: The population included 1868 (55%) women with a mean age of 48±8 years. The median CAC was 0 (range 0 - 5363). In figure 1 we present the association of each score with the CAC values stratified by gender. All score are significantly associated with the discrete part of the model (probability of a CAC=0) (p>0.001). However, while all scores were associated with the continuous part of the model for men (p<0.0001); only the Reynolds (p=0.035) and the high SCORE (p<0.0001) were associated with the continuous part of the model in women, whereas the Framingham score (p=0.96), ASSIGN (p=0.33) and PROCAM (p=0.78) were not. Figure 1 shows the association of each score with the CAC results stratified by gender.
Conclusions: In a large Brazilian cohort, all clinical risk scores were significantly associated with the presence of CAC in men, whereas only the SCORE and Reynolds were associated with CAC values in women.
Author Disclosures: M.S. Bittencourt: None. A.C. Pereira: None. H.L. Staniak: None. R. Sharovsky: None. L. Gomez: None. V. Mendes: None. R. Blankstein: None. K. Nasir: None. I. Bensenor: None. P.A. Lotufo: None.
- © 2014 by American Heart Association, Inc.