Predicting prognosis of sepsis in patients based on right ventricular strain imaging development and validation of a nomogram model

TL;DR

This study develops a nomogram model using right ventricular strain and clinical parameters to predict 28-day mortality in sepsis patients.

Contribution

A novel nomogram model incorporating right ventricular strain and clinical biomarkers is developed and validated for sepsis prognosis.

Findings

Non-survivors had significantly lower right ventricular free wall and global strain compared to survivors.

The nomogram model achieved an AUC of 0.907 for predicting 28-day mortality in sepsis patients.

Age, SOFA score, procalcitonin, lactate, and RV-FWS were identified as independent risk factors for poor outcomes.

Abstract

The right ventricle (RV) plays a significant role in septic myocardial injury and associated organ dysfunction. Hence, identifying right ventricular systolic dysfunction (RVSD) early is crucial for improving outcomes in septic patients, yet current research on RVSD in sepsis remains limited.

The study aims to identify risk factors for adverse outcomes in septic patients and construct a nomogram prediction model incorporating right ventricular strain and right ventricle–pulmonary artery coupling parameters.

This single-center prospective study included 156 sepsis patients admitted from September 2021 to October 2024. General clinical, laboratory, and echocardiographic data were collected within 72 h of sepsis diagnosis. Prognosis was used to divide patients into two groups. Lasso regression was used to examine the baseline features of both groups. Multivariable logistic regression analysis and a nomogram were used to predict sepsis prognosis. The relationship between RVSD and 28-day mortality was examined.

Within 28 days, 52 of 141 sepsis patients died. Univariate analysis showed that the non-survivor cohort was older and had higher APACHE II and Sequential Organ Failure Assessment (SOFA) ratings and procalcitonin, B-type natriuretic peptide, cTnI, and lactate. RV-free wall strain (−18.9% ± 1.6% vs. −20.1% ± 1.5%, p < 0.001) and RV global strain (−18.6% ± 1.4% vs. −17.6% ± 1.0%, p < 0.001) were lower in the non-survivor group compared to the survivor cohort. PASP and RV-GS/PASP ratio significantly differed between the two groups (p < 0.05). Multivariable logistic regression analysis identified age >67 years, SOFA score ≥7.5, procalcitonin ≥5.7 ng/ml, lactate ≥3.5 mmol/L, RV-FWS ≥−19.4%, and RV-GS/PASP ≥−0.55 as independent risk factors for poor sepsis outcomes. The prognostic model using these six risk factors had an area under the curve (AUC) of 0.907 (95% CI: 0.858–0.954). Internal validation showed strong nomogram calibration with a C-index of 0.88.

The RV-GS/PASP ratio demonstrated significant prognostic utility for predicting clinical outcomes in sepsis patients. Furthermore, the nomogram model incorporating age, SOFA score, procalcitonin, lactate, and RV-FWS exhibited excellent discriminative ability, with an AUC of 0.907.

Figures