Abstract:  Objective To compare the performance of updated Diamond–Forrester method (UDFM) and Duke clinical score (DCS) in a Chinese population referred to coronary computed tomography angiography (CCTA). Methods A total of 5 743 consecutive patients with suspected coronary artery disease (CAD) who underwent CCTA for stable chest pain were collected in Tianjin Chest Hospital from December 2015 to December 2017. CAD was defined as stenosis ≥50% in diameter at least one segment of the coronary artery by CCTA. For each patient, pretest probability (PTP) of CAD was estimated according to UDFM and DCS, respectively. Area under the receiver operating characteristic curve (AUC), net reclassification improvement (NRI) and Hosmer–Lemeshow goodness-of-fit statistic (H-Lχ2 ) were assessed to validate and compare the two models. Results Overall, 1 872 (32%) patients were diagnosed as CAD by coronary CTA examination. There were no significant differences between DCS (AUC=0.772, 95%CI: 0.759-0.786) and UDFM (AUC=0.765,95%CI:0.751-0.779, P= 0.068). DCS demonstrated the improved classification over UDFM through the positive NRI (26.20%, P＜0.001). The calibration of UDFM (H-L χ2 =137.823) and DCS (H-L χ2 =156.704) was both unsatisfactory (P＜0.01), and both overestimated the actual prevalence of CAD. Conclusion Compared to UDFM, DCS is superior for estimating the PTP of CAD in Chinese population referred for CCTA, but both overestimate the actual prevalence of CAD.

Key words:  coronary disease, coronary angiography, ROC curve, area under curve, stable chest pain, pretest probability, UDFM, DCS