Establishment and validation of a critical care echocardiography-based predictive model for sepsis-induced...

Integrating echocardiographic parameters for a comprehensive and precise evaluation of sepsis-induced cardiomyopathy (SIC) presents a significant challenge.

To develop a nomogram for the echocardiographic diagnosis of SIC.

A cohort of 181 septic patients was prospectively enrolled for critical care echocardiography assessments. The cohort was randomly divided into a training dataset (70 %, n = 126) and a validation dataset (30 %, n = 55). The LASSO regression analysis was used to identify key echocardiographic predictors, which were then analyzed using multivariate logistic regression to determine the final diagnostic predictors and establish an echocardiographic model for SIC. A nomogram was developed based on the model, which was evaluated and verified for discrimination, calibration, and clinical utility.

Three key predictors, including left ventricular global longitudinal strain (GLS), early diastolic mitral flow velocity (E), and tricuspid annular plane systolic motion amplitude (TAPSE), were selected from 14 variables to develop a SIC echocardiographic predictive model. The model exhibited a strong discrimination with an area under the curve (AUC) value of 0.879 in the training dataset and 0.888 in the validation dataset. The results of the Hosmer-Lemeshow test further validated the consistency between predicted probabilities and actual outcomes in both datasets. Decision curve analysis (DCA) indicated a substantial net clinical benefit within risk threshold ranges of 5-100 % in the training dataset and 21-100 % in the validation dataset.

The nomogram, incorporating GLS, E, and TAPSE, emerged as a reliable non-invasive tool for assessing the risk of SIC.

The study protocol was registered in the ChiCTR database (Registration No. ChiCTR2200066966; Date of Registration: December 22, 2022).