Abstract 11077: Overestimation of Pretest Probability of Coronary Artery Disease by Duke Clinical Score: Evaluation Using Coronary CT Angiography
Background: The Duke Clinical Score (DCS) is based on catheter angiography as a reference standard for a diagnosis of coronary artery disease (CAD). Coronary CT Angiography (CCTA) is applied to wide population and thus the prevalence of CAD is generally lower among patients who undergo CCTA when compared to the population used to develop the DCS. Because pretest probability is largely influenced by disease prevalence, the purpose of this study was to test the hypothesis that the DCS overestimates the pretest probability of CAD among patients who undergo CCTA.
Methods: The DCS was calculated for each of the 3996 consecutive CCTA studies (02/2009 to 04/2013) performed at a single institution for symptomatic patients with no known CAD. Performance of the DCS for the detection of significant coronary stenosis (≥75%) was evaluated by the Area Under the Curve (AUC) of the Receiver Operating Characteristic curve.
Results: Significant stenosis was detected by CCTA in 23.3% (930/3996) of the population. The AUC of the DCS was 0.71 when CCTA findings were used as “true” CAD. Mean DCS of the low (<30%), intermediate (30-70%), and high (>70%) risk category was 18.3±7.0% (n=993), 48.9±12.1% (n=1561), 86.0±8.2% (n=1442), respectively, while the actual prevalence of the significant stenosis was much lower, namely 8.4% (83/993), 19.4% (302/1561), and 37.8% (545/1442), respectively. The DCS and the actual prevalence were linearly fit (Figure) by the equation: prevalence = 0.418 x DCS + 0.79 (r=0.935), revealing that intermediate risk determined by DCS (30-70%) corresponded to a range of 13.3 and 30.1% of the actual CAD prevalence. The intermediate risk as defined by Appropriate Use Criteria 2010 (i.e., 10-90%) corresponded to an actual prevalence of 5.0-38.4%. Ten percent of the actual CAD prevalence corresponded to the DCS of 22%.
Conclusion: Among patients who underwent CCTA for suspected CAD, DCS overestimated the pretest probability at least 2-fold. Patients with DCS <20% should be considered to have an actual CAD prevalence of <10%. The DCS does not appear to be effective when detecting patients with high pretest probability (either >70% or >90%), and thus new criterion that can accurately calculate the pretest probability for patients who are referred to CCTA is warranted.
- © 2013 by American Heart Association, Inc.