Abstract 9523: Comparison of Guideline Based Pretest Probability Scores Applied to Patients Undergoing Coronary Computed Angiography - The Partners Registry
Introduction: According to current guidelines, the initial test for the evaluation of coronary artery disease (CAD) in individuals with no known disease should be chosen according to the pre-test probability of obstructive CAD. While U.S. guidelines use the Diamond and Forrester (DF) classification, the recent European guidelines have proposed the use of other methods to estimate the pretest probability.
Hypothesis: The performance of newer CAD prediction scores will be superior to the DF classification in patients referred to a coronary CTA.
Methods: Exam findings were classified as non-obstructive (no CAD or <50% stenosis) or obstructive (≥50%). For each individual we calculated a continuous version of the Diamond and Forrester score, the first version of the CAD consortium score (CAD1), and the two versions of the CAD consortium 2, the basic version (CAD2b), which includes the same variables as the DF classification, and a complete version (CAD2c), which also includes risk factors. We compared the calibration (predicted vs. observed prevalence of obstructive CAD) and discrimination (area under the ROC curves).
Results: We included 2580 individuals (mean age 55±13 years, 58% males) with complete information on symptoms and risk factors needed to calculate the scores. 545 (21%) individuals had obstructive CAD on coronary CTA. The estimated probability of CAD by DF was higher than observed, especially among those with an intermediate-high pretest probability (Figure). When evaluating the discrimination, the c-statistic for the DF=0.718 (95%CI: 0.695 - 0.742), for CAD1= 0.7265 (0.703 - 0.750), for CAD2b =0.753 (0.730 - 0.776) and for the CAD2c = 0.791 (0.771 - 0.811). The CAD2c had a significantly higher c-statistic than the other models (p<0.0001)
Conclusions: The CAD2c provides better discrimination and calibration to predict obstructive CAD. Thus, the CAD consortium complete score should be preferred against current risk prediction strategies.
Author Disclosures: M.S. Bittencourt: None. E. Hulten: None. K. Nasir: None. M. Cheezum: None. S. Abbara: None. B. Gohshhajra: None. U. Hoffmann: None. M. Di Carli: None. R. Blankstein: None.
- © 2014 by American Heart Association, Inc.