Sepsis is a life-threatening condition that poses a significant challenge in emergency and critical care medicine worldwide. It occurs when the body's immune response to an infection becomes dysregulated, leading to organ dysfunction. Despite ongoing efforts, sepsis continues to claim millions of lives annually, with a global incidence of over 48.9 million cases and related deaths reaching 11 million, accounting for nearly 20% of all deaths worldwide. In China alone, the annual number of hospitalized sepsis cases is estimated to be between 4.8 and 6.1 million. One of the most severe complications of sepsis is myocardial injury, known as SMCI, which affects a significant proportion of patients, with an incidence ranging from 13.8% to 79%. The occurrence of SMCI not only complicates treatment but is also strongly associated with a poor prognosis. Studies have shown that patients with SMCI have a mortality rate of 32.6% to 43.7%, significantly higher than those without myocardial injury.
Early identification of high-risk patients for SMCI and timely implementation of targeted interventions are crucial for improving the prognosis of sepsis patients. Risk prediction models, which integrate multiple factors, offer a quantitative tool to provide individualized risk assessments and assist clinicians in making early decisions. These models have proven valuable in risk stratification and prognostic evaluation for various diseases. However, existing studies on SMCI risk factors have limitations. They often fail to provide an intuitive, visual, and convenient way to calculate the risk of SMCI occurrence, thus limiting their practical value.
To address these challenges, researchers have developed a visual tool called a nomogram, which demonstrates the relationship between various clinical variables and the probability of SMCI. By creating a nomogram, healthcare professionals can rapidly assess the risk of myocardial injury in sepsis patients through user-friendly digital interfaces. This enables timely interventions, potentially reducing morbidity and mortality. While previous studies have made strides in this area, some limitations remain. For instance, certain studies have converted continuous variables into dichotomous variables, which may compromise the accuracy of SMCI probability assessment. Additionally, some research has focused on sepsis patients admitted to the intensive care unit (ICU), which may not capture the early stages of the disease.
This study aims to construct an SMCI risk prediction model by systematically collecting initial clinical data of sepsis patients upon their presentation to the emergency department. The model incorporates indicators reflecting immunity, cytokine storm, tissue perfusion, and other commonly used clinical indicators. By evaluating the model's discrimination, calibration, and clinical applicability through internal validation, the researchers hope to provide clinical evidence for the early identification of SMCI. This approach aims to facilitate early detection, diagnosis, and treatment of SMCI, ultimately improving patient outcomes.
The study analyzed the clinical data of 370 sepsis patients admitted to the Emergency Resuscitation Area of the First Affiliated Hospital of Xinjiang Medical University from September 2022 to December 2024. The inclusion criteria were patients aged 18 years or older, diagnosed with sepsis based on the Sepsis-3 diagnostic criteria, and hospitalized for more than 24 hours. Exclusion criteria included a history of certain chronic conditions and incomplete data. The patients were randomly assigned to a training cohort or internal validation cohort, with a 7:3 ratio.
The researchers calculated the variance inflation factor (VIF) to assess multicollinearity and employed LASSO regression to select optimal features and prevent overfitting. Through univariate and multivariate logistic regression analyses, they identified independent risk factors for SMCI. These factors were then used to construct a nomogram for SMCI risk prediction. The nomogram's performance was evaluated using the area under the curve (AUC), Hosmer-Lemeshow tests, calibration curves, decision curve analysis (DCA), and clinical impact curve (CIC).
The results showed that the nomogram had a moderately good performance, with an AUC of 0.856 in the training cohort and 0.853 in the internal validation cohort. Hosmer-Lemeshow tests and calibration curves indicated good calibration, and DCA and CIC analyses demonstrated good clinical applicability.
This study has several advantages. It employed LASSO regression to mitigate multicollinearity risks, and the subsequent VIF confirmed the absence of multicollinearity among the predictors. The study also collected commonly used clinical data within 24 hours of admission for sepsis patients in the emergency resuscitation area, enhancing clinical applicability and early prediction.
However, the study also has limitations. As a retrospective analysis, it lacks real-time applicability and prospective bias control. Additionally, certain factors such as fluid resuscitation and antibiotic use were not considered. The study also did not include traumatic sepsis, and the model lacks external or multicenter validation to assess its generalizability.
In conclusion, the proposed nomogram based on Log Myo, Log BNP, and Log IL-6 may serve as a practical tool for early risk assessment of myocardial injury in sepsis. However, further external validation is necessary before clinical implementation. This study contributes to the development of effective strategies for managing sepsis and improving patient outcomes.
