The establishment and optimization of a long-term prognosis assessment model for patients in sepsis combined with heart failure: Based on the Medical Information Mart for Intensive Care IV database

Abstract Introduction: Sepsis is a prevalent occurrence in clinical practice, with heart failure emerging as a frequent and severe sequelae of sepsis. Despite notable progress in therapeutic interventions, the protracted prognosis for individuals experiencing sepsis-associated heart failure (SHF) continues to be unfavorable. The primary aim of this study was to develop and validate a prognostic model designed to anticipate one-year mortality in patients diagnosed with SHF. Methods: The relevant data pertaining to patients diagnosed with sepsis-associated heart failure (SHF) was extracted from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database using Structured Query Language (SQL). The patient cohort was subsequently randomized into a training set and a validation set at a 7:3 ratio. The training set was utilized for model development, while the validation set served for model validation. Independent prognostic factors for SHF patients were identified through forward stepwise logistic regression, and a nomogram was constructed to predict one-year mortality in SHF patients. Validation of the nomogram included assessments using multiple indicators, such as the area under the receiver operating characteristic curve (AUC), calibration curve, integrated discrimination improvement (IDI), net reclassification index (NRI), and decision curve analysis (DCA). Results: Our study included a total of 4,228 patients diagnosed with sepsis-associated heart failure (SHF), with 2,960 patients allocated to the training cohort and 1,268 patients to the validation cohort. Through stepwise regression analysis, we identified twelve independent prognostic factors and constructed a nomogram. The AUC values for the nomogram were 0.740 (95% CI = 0.712–0.764) in the training cohort and 0.745 (95% CI = 0.726–0.763) in the validation cohort. The high AUC, Net Reclassification Index (NRI), and Integrated Discrimination Improvement (IDI) values underscored the robust discriminative capability of the nomogram. Calibration curves and results from the Hosmer-Lemeshow test indicated excellent calibration, and Decision Curve Analysis (DCA) curves demonstrated the nomogram's significant clinical utility. Conclusion: The novel prognostic nomogram designed to predict one-year mortality in patients with sepsis-associated heart failure (SHF) demonstrates superior performance compared to other existing scoring systems. The nomogram holds potential for guiding clinical practice. Nevertheless, additional external prospective validation is deemed necessary to further confirm its reliability and generalizability.