A Nomogram for Predicting Hospital Mortality in Intensive Care Unit Patients with Acute Myocardial Infarction.

backgroundThis study aims to construct and validate an early-stage nomogram for predicting hospital mortality of ICU patients with acute myocardial infarction (AMI), to help clinicians determine the appropriate intervention.

methodsThe primary cohort of 2704 patients diagnosed with acute myocardial infarction in admission records from eICU-Collaborative Research Database (eICU-CRD) v2.0. Univariate logistic regression analysis and multivariate logistic regression analysis were enrolled for the construction of the predictive nomogram. Demographic factors, history of clinical cardiovascular disease, vital signs, the use of vasopressors, urine output, and serum variables in the first 24 hours were included in this analysis. The nomogram was evaluated by performance traits including Harrell's concordance index (C-index) and area under the receiver operating characteristic (AUC) analysis, calibration curve, and decision curve analysis (DCA). The nomogram was validated in a different cohort containing 1026 subjects collected from MIMIC-III Database v1.4. Finally, in order to compare the performance with other classic prediction models, AUC analysis, calibration curve, DCA and accuracy analysis (net reclassification improvement (NRI)) were conducted for three ICU scores in validated cohort.

resultsThe nomogram revealed 14 predictors of the first 24 hours derived from univariate and multivariable analyses, including age, history of peripheral vascular disease, atrial fibrillation, cardiogenic shock and cardiac arrest, the use of norepinephrine, urine output, white blood cell (WBC), hemoglobin (Hb), red blood cell (RBC), red cell distribution width (RDW), glucose, bicarbonate and magnesium. The C-index of this nomogram was 0.834 (95% CI 0.812 to 0.856). Then, the result of AUC analysis, the DCA and calibration curve indicated that our nomogram was feasible for clinical prediction. The predictive ability and clinical use of the nomogram were verified in the validated cohort. The AUC analysis of ICU scores showed that the AUC of these score systems was ranged from 0.811 to 0.860 (the AUC of nomogram: 0.885). Moreover, our nomogram also showed a better performance in calibration curve and DCA NRI.

conclusionThe study presents a prediction nomogram incorporating 14 variables that could help identify AMI patients admitted in ICU who might have a high risk of hospital mortality in the first hospitalized 24 hours. This nomogram showed a better performance than normal ICU score systems.