P3101Risk stratification model for exercise training complications in patients with heart disease (EXERISK study)

Abstract

Abstract Background Exercise training is worldwide recommended to treat patients with heart disease, but sometimes, it can trigger adverse outcomes (ischemia, arrhythmias or sudden death). Several clinical tools are used to stratify patients, usually developed by expert consensus without any clinical validation during exercise sessions. Purpose The aim of this study was to develop up and validate a contemporary risk stratification calculator to identify patients at high risk of adverse outcomes during physical training. Methods A cohort of patients with heart disease that participated in a cardiac rehabilitation program were studied. A data base was constructed using several variables of interest derived from clinical records, patient evaluations, paraclinical reports along with stress testing. All patients participated in an ambulatory hospital-based cardiac rehabilitation program. Every patient exercised 5 times a week, 30 minutes continuously at a moderate-intensity, using a high-tech cyclergometer. Aerobic training was complemented with a three times a week, general gymnastics circuit. All sessions were highly supervised by trained personal, including a cardiologist. Major outcome variables were sudden death, coronary acute syndrome, syncope or stroke. Minor outcomes were any kind of arrhythmia and the presence of ischemia (angina or ST depression). Nominal, categorical or numerical variables were presented as appropriate. A bivariate analysis was performed. Those statistically significant variables (p<0.05), were used in a logistic regression multivariable analysis with a step by step forward-Wald method. The obtained linear general model (G) was incorporated into a logistic equation and results were compared with traditional risk-stratification scales. Results A total of 639 patients were included, and no major adverse outcome occurred. Seventy six percent of patients presented arrhythmias and 5% showed ischemia. Seventeen stochastically significant variables were obtained from clinical records and paraclinical tests, and were included in two multivariable models, focusing on arrhythmia (EXERISK-A) or ischemia (EXERISK-I). After analysis, three variables remained associated with arrhythmia (maximal METs, left ventricle ejection fraction and frequent premature ventricular complexes) and three with ischemia (history of angina, METs and SYNTAX score). Once logistic equations were obtained, traditional risk-stratification scales were applied to the sample. Figure 1 shows ROC curve comparison of both models with traditional stratification scales. The area under the curve was 0.730 (EXERISK-A) and 0.856 (EXERISK-I) EXERISK Conclusion Exercise-induced arrhythmias are a very common finding in patients with heart disease, few patients showed ischemia and no mayor adverse cardiovascular outcomes were demonstrated. Both risk stratification models (EXERISK-A and EXERISK-I), can accurately predict arrhythmias or ischemia in a simple and rapid fashion.