An Individualized Nomogram for Predicting Risk of Sepsis in Patients With Pyogenic Liver Abscesses: A ten Years Retrospective Analysis

Tiezhao ZhangXidong CaoLiyong ZhangJinHua CuiJian LiZiyu BaiAijun Yu^*

• Affiliated Hospital of Chengde Medical University, Chengde, China

The incidence of pyogenic liver abscess (PLA) has been increasing. It is crucial to identify patients at high risk of sepsis early and develop personalized treatment plans to reduce the disease burden of patients with liver abscesses. A retrospective study involving 490 patients with PLA was conducted. In chronological order, patients treated from August 2014 to September 2021 were employed as the training cohort (n = 341), and patients treated from October 2021 to July 2024 were employed as the validation cohort (n = 149). The occurrence of sepsis in patients with liver abscesses was defined as the outcome. The Chi-square test or Fisher's exact test was used to test categorical variables, whereas the Student's t-test was employed to evaluate continuous variables. Independent risk factors associated with sepsis in the training cohort were identified using multivariate logistic regression analysis. A nomogram was developed and validated using an independent cohort. Model performance was systematically evaluated through three analytical approaches. Receiver operating characteristic (ROC) curves were generated for both the training and validation cohorts to assess discrimination accuracy. Calibration curves were constructed to quantify the agreement between predicted and observed outcomes. Decision curve analysis (DCA) was conducted to determine the clinical utility threshold where the nomogram's net benefit surpassed empirical treatment strategies across both cohorts. A total of 108 (22 %) patients with PLA were complicated with sepsis. In patients with liver abscesses, independent risk factors for sepsis, including white blood cell count, international normalized ratio (INR), presence of gas, and SOFA score, were identified through multivariate logistic regression analysis. For the training and validation cohorts, the AUC values of the nomogram were 0.880 (95 % CI: 0.832-0.929) and 0.901 (95 % CI: 0.839-0.964), respectively. showed that the newly established nomogram exhibited superior predictive performance and clinical utility. The Hosmer-Lemeshow test (χ²=8.60, P=0.377) suggests good fit. The calibration curve showed good consistency, and the DCA decision curve showed that the model was clinically effective.A model with four clinical features was developed to predict the risk of sepsis in patients with liver abscesses. The model exhibited good predictive ability during time verification.