Early risk stratification using clinical findings in patients with acute myocardial infarction

The value of non-invasive, clinical variables in predicting short-term outcome after acute myocardial infarction has been prospectively studied using multivariate analysis in a learning population of 1724 patients and a validation sample of 588 cases. Early risk stratification of CCU mortality could be made with 89.4% accuracy by using 7 simple clinical variables (Killip class, systolic blood pressure, heart rate, age, intraventricular conduction disturbance, localization of infarction and obesity index) recorded on admission. A discriminant function formula could also be constructed resulting in a maximal classification of 89.3% and predictive values of 91.8% and 77.9%, respectively, for 28-day survival and early death prediction in the learning population. Accuracy figures dropped only slightly (1 to 5%) in the independent test set. Predictions could be made with a high degree of accuracy, especially in the upper and lower ranges of the discriminant risk score. This investigation demonstrates that a fairly good estimation of prognosis can be made by means of discriminant analysis of clinical findings gathered routinely in most CCU's.