Predicting hospital mortality in patients with acute myocardial infarction.

BACKGROUND Patients who have a myocardial infarction are a heterogeneous group. If those at risk for early mortality could be readily identified, it would provide a more solid basis for management decisions. Although past research has explored factors associated with mortality, findings are inconsistent. Variables have also been combined into prognostic indices, but these tools have yet to be evaluated adequately. OBJECTIVES To determine factors predictive of hospital mortality in patients with acute myocardial infarction, and to examine the usefulness of two severity-of-illness indices. METHODS The medical records of 392 patients diagnosed with acute myocardial infarction who had undergone coronary angiography during 1989 at a university medical center were reviewed. RESULTS Overall mortality was 9.4% (n = 37). Logistic regression analysis demonstrated that history of myocardial infarction, cardiogenic shock, age, left ventricular ejection fraction, and the number of occluded coronary vessels were significantly associated with hospital mortality in patients with acute myocardial infarction. The two severity-of-illness indices were significant predictors of mortality, although sensitivity, specificity, and predictive values varied. A formula for determining the probability of mortality, based on logistic regression analysis, is also presented. CONCLUSIONS Five factors were found to predict hospital mortality. The two severity-of-illness indices were moderately useful in predicting mortality. Unlike previous indices that did not incorporate currently available diagnostic data, the new formula included data from coronary angiography and nuclear scans. Although this formula requires validation on independent samples of patients with myocardial infarction, the findings of this study advance clinicians' ability to predict patient outcome.