Use of cardiovascular events prediction models for the prognosis of a severe coronary artery lesion

Objectives: To determine the best cut-off point and the level of agreement of the Framingham, PROCAM, and Reynolds risk scales, for the diagnosis of a severe atherosclerotic coronary artery lesion by comparing the areas under the receiver operator characteristics (ROC) curves. Methods: A cross-sectional study was carried out on adults that were subjected to coronary angiography. The Framingham, PROCAM, and Reynolds were applied, with the first two adjusted to a Colombian calibration study. Data were collected concurrently in the institutions of reference in Cardiology and haemodynamics in Bogota. Results: Of the 200 patients study, 66% were female, and 37.5% greater than 70 years-old. Hypertension was recorded in 53.2%, and overweight and obesity in 52.7%. Elevated levels of high sensitivity C-Reactive Protein (hsCRP) were observed in 61.5% of cases and 50% had at least one major coronary lesion greater than 70%. The best cut-off points according to each of the ROC curves: Adjusted Framingham, 5.8% (sensitivity, 80%, specificity, 41%), Adjusted PROCAM, 1.7% (sensitivity, 78%, specificity, 45%) and Reynolds, 3.8% (sensitivity, 68%, specificity, 45%). Additionally, the three areas under the curve (AUC) were 0.59, 0.59, and 0.57, respectively.