Development And Evaluation Of A Mortality Risk Prediction Model For Sepsis-induced Myocardial Dysfunction Patients

Objective To explore the factors affecting the mortality risk in patients with sepsis-induced myocardial dysfunction patients（SIMD）and to develop and validate a practical nomogram to predict the risk of 90-day mortality in SIMD patients,in order to identify patients with poor prognosis earlier,intervene and reduce mortality.Methods Based on the Medical Information Mart for Intensive Care Ⅲ（MIMIC Ⅲ,version1.4）,a structured query language（SQL）was used to extract patients who met the diagnosis of sepsis.According to the inclusion and exclusion criteria,basic information,echocardiography data,vital signs and laboratory tests on the first day of admission to the ICU were collected from the final enrolled patients,and the 90-day mortality was used as the primary clinical outcome,and the patients eligible for SIMD diagnosis were randomly divided into the development and validation groups in the ratio of 7:3.The development group used logistic regression to graph nomogram.The predictive model was evaluated by calculating the area under ROC curve（AUC）to compare the prediction performance between the nomogram and existing clinical scores.Besides,the prediction accuracy of the model was evaluated by calibration curve and Hosmer-Lemeshow goodness of fit test,the clinical applicability of the prediction model was evaluated by Decision Curve Analysis（DCA）.Results A total of 5185 patients with sepsis from the MIMIC database were included in this study,of whom 1559（30.1%）were diagnosed with SIMD at first ICU admission,with a 90-day mortality rate of 36.0%（568/1559）,and SIMD was independently associated with mortality in ICU patients.Kaplan-Meier survival analysis showed that regardless of propensity score matching（PSM）or not,and whether the SIMD group was balanced or not,there was a significant difference in 90-day survival（logrank test: P < 0.05）.For nomogram,a total of 11 variables from 1559 SIMD patients were included.When predicting the risk of 90-day mortality,the model showed good discrimination in the development and validation sets（AUC: 0.76 and0.76）.In contrast,the area under the curve for the sequential organ failure score（SOFA）,the logistic organ dysfunction score（LODS）,the Oxford Acute Severity of Illness Score（OASIS）,and the Simplified Acute Physiology Score,second edition（SAPS II）were0.65,0.65,0.65,and 0.71.Respectively,in the development set,the slope of calibration curve was 1.00 and the Hosmer-Lemeshow test showed the P-value was 0.301.From the decision curve analysis,it was clear that the nomogram acquired more net benefit compared to other severity scores.Conclusions Sepsis-induced myocardial dysfunction was independently associated with 90-day mortality in ICU patients.We developed a 90-day mortality risk predictive model for SIMD patients and achieved the best prediction of 90-day mortality in SIMD patients by outperforming existing clinically common scores（SOFA）and ICU severity of illness scores（LODS,OASIS,and SAPS II）,and the nomogram model was internally validated to have better discrimination,calibration,and clinical applicability,and thus has potential clinical application.