Development and validation of a nomogram to predict risk of septic cardiomyopathy in the intensive care unit

Abstract The aim of this study was to develop a simple but effective nomogram to predict risk of septic cardiomyopathy (SCM) in the intensive care unit (ICU). We analyzed data from patients who were first admitted to the ICU for sepsis between 2008 and 2019 in the MIMIC-IV database, with no history of heart disease, and divided them into a training cohort and an internal validation cohort at a 7:3 ratio. SCM is defined as sepsis diagnosed in the absence of other cardiac diseases, with echocardiographic evidence of left (or right) ventricular systolic or diastolic dysfunction and a left ventricular ejection fraction (LVEF) of less than 50%. Variables were selected from the training cohort using the Least Absolute Shrinkage and Selection Operator (LASSO) regression to develop an early predictive model for septic cardiomyopathy. A nomogram was constructed using logistic regression analysis and its receiver operating characteristic (ROC) and calibration were evaluated in two cohorts. A total of 1562 patients participated in this study, with 1094 in the training cohort and 468 in the internal validation cohort. SCM occurred in 13.4% (147 individuals) in the training cohort, 16.0% (75 individuals) in the internal validation cohort. After adjusting for various confounding factors, we constructed a nomogram that includes SAPS II, Troponin T, CK-MB index, white blood cell count, and presence of atrial fibrillation. The area under the curve (AUC) for the training cohort was 0.804 (95% CI 0.764–0.844), and the Hosmer–Lemeshow test showed good calibration of the nomogram (P = 0.288). Our nomogram also exhibited good discriminative ability and calibration in the internal validation cohort. Our nomogram demonstrated good potential in identifying patients at increased risk of SCM in the ICU.