Construction and validation of a risk prediction model for bacterial liver abscess-induced sepsis
10.3969/j.issn.1006-2483.2025.06.035
- VernacularTitle:细菌性肝脓肿并发脓毒症风险预测模型构建及验证
- Author:
Jiarong LI
1
;
Yongfang LIU
1
Author Information
1. Department of Infectious Diseases, The Third People's Hospital of Chengdu, Chengdu, Sichuan 610000, China
- Publication Type:Journal Article
- Keywords:
Bacterial liver abscess;
Sepsis;
Prediction model
- From:
Journal of Public Health and Preventive Medicine
2025;36(6):157-161
- CountryChina
- Language:Chinese
-
Abstract:
Objective Develop a risk prediction model for patients with bacterial liver abscess complicated by sepsis, and validate its predictive performance. Methods Clinical data were collected from 233 patients with bacterial liver abscesses admitted to our hospital between January 2019 and October 2024. Based on the occurrence of sepsis, the patients were categorized into a sepsis group (n=29) and a non-sepsis group (n=204). After conducting univariate analysis and subsequently multivariate Logistic regression analysis, the influencing factors were identified for the construction of a nomogram prediction model. The discrimination of the model was evaluated by the AUC of the ROC curve. The calibration of the model was assessed using the calibration curve and the Hosmer-Lemeshow test. The clinical utility of the model was evaluated through decision curve analysis. Results Age, history of hepatobiliary invasive procedures within three months, coexistence of malignancy, abscess location, blood culture results, and PCT levels are independent factors influencing the development of sepsis in patients with PLA (P < 0.05). The AUC of the model was 0.942, with a sensitivity of 92.6% and a specificity of 89.7%. Both calibration curves and the Hosmer-Lemeshow goodness-of-fit test for the model indicate good model calibration. The decision curves for model indicate that the model yields a favorable net benefit when applied to patients falling within the specified range of predicted probabilities. Conclusion The nomogram prediction model constructed in this study for sepsis in patients with PLA demonstrates good predictive value and can provide a reference for early identification of sepsis in PLA patients.
