The risk score for in-hospital mortality in patients with ST-segment elevation myocardial infarction

Abstract

Aim    To develop a scale (score system) for predicting the individual risk of in-hospital death in patients with ST segment elevation acute myocardial infarction (STEMI) with an account of results of percutaneous coronary intervention (PCI).Material and methods    The analysis used data of 1 649 sequential patients with STEMI included into the hospital registry of PCI from 2006 through 2017. To test the model predictability, the original sample was divided into two groups: a training group consisting of 1150 (70 %) patients and a test group consisting of 499 (30 %) patients. The training sample was used for computing an individual score. To this purpose, β-coefficients of each variable obtained at the last stage of the multivariate logistic regression model were subjected to linear transformation. The scale was verified using the test sample.Results    Seven independent predictors of in-hospital death were determined: age ≥65 years, acute heart failure (Killip class III-IV), total myocardial ischemia time ≥180 min, anterior localization of myocardial infarction, failure of PCI, SYNTAX scale score ≥16, glycemia on admission ≥7.78 mmol/l for patients without a history of diabetes mellitus and ≥14.35 mmol/l for patients with a history of diabetes mellitus. The contribution of each value to the risk of in-hospital death was ranked from 0 to 7. A threshold total score of 10 was determined; a score ≥10 corresponded to a high probability of in-hospital death (18.2 %). In the training sample, the sensitivity was 81 %, the specificity was 80.6 %, and the area under the curve (AUC) was 0.902. In the test sample, the sensitivity was 96.2 %, the specificity was 83.3 %, and the AUC was 0.924.Conclusion    The developed scale has a good predictive accuracy in identifying patients with acute STEMI who have a high risk of fatal outcome at the hospital stage.