Risk Factors and Predictive Model Construction of Klebsiella pneumoniae Liver Abscess Invasion Syndrome

目的：回顾性分析肺炎克雷伯菌肝脓肿综合征患者发生侵袭综合征的危险因素，构建和验证列线图预测模型，并进行相关验证。方法：检索收集青岛大学附属医院2017.12~2023.12期间第一诊断为肝脓肿且引流液培养为单一肺炎克雷伯杆菌的389例患者的病例资料，应用多因素logistic回归筛选肝外侵袭的独立预测因素构建列线图，绘制受试者工作特征(ROC)曲线对模型的预测效能进行评价，校准曲线对模型的校准度进行评价，临床决策曲线(DCA)评价模型的临床效益，Bootstrap法进行内部验证评估模型的稳定性和泛化能力。结果：本项研究总共纳入389例患者作为研究对象，根据肝外表现分为未侵袭组和侵袭组两组，其中未侵袭组332例，侵袭组57例，两组在年龄、性别之间均没有统计学差异。最终筛选出sofa评分(OR = 1.306, 95% CI: 1.136~1.507, P 0.05，校准曲线图示校准曲线与理想曲线贴合良好，临床决策曲线(DCA)图显示当阈值概率在0.01~0.99时模型均能产生更好的临床效益。结论：sofa评分、中性粒细胞百分率、空腹血糖、白蛋白是肺炎克雷伯菌肝脓肿患者发生侵袭综合征的独立危险因素，根据此指标可构建有效的肺炎克雷伯肝脓肿侵袭综合征的预测模型，具有一定的临床应用价值。Objective: To retrospectively analyze the risk factors of invasive syndrome in patients with Klebsiella pneumoniae liver abscess syndrome, construct and verify the nomogram prediction model, and carry out relevant verification. Method: The case data of 389 patients who were first diagnosed with liver abscess and whose drainage fluid was cultured as a single KP in the Affiliated Hospital of Qingdao University from 2017.12 to 2023.12 were collected. Multivariate logistic regression was used to screen the independent predictors of extrahepatic invasion to construct a nomogram. The receiver operating characteristic (ROC) curve was drawn to evaluate the predictive efficacy of the model, the calibration curve was used to evaluate the calibration of the model, the clinical decision curve (DCA) was used to evaluate the clinical benefits of the model, and the Bootstrap method was used to evaluate the stability and generalization ability of the model. Results: A total of 389 patients were included in this study. According to extrahepatic manifestations, they were divided into non-invasive group and invasive group, including 332 cases in non-invasive group and 57 cases in invasive group. There was no significant difference in age and sex between the two groups. Finally, sofa score (OR = 1.306, 95% CI: 1.136~1.507, P 0.05, the calibration curve diagram shows that the calibration curve fits well with the ideal curve, the visual calibration curve diagram shows that the predicted risk fits well with the actual risk represented by the standard line. The clinical decision curve (DCA) diagram shows that the model can produce better clinical benefits when the threshold probability is 0.01~0.99. Conclusion: Sofa score, neutrophil percentage, fasting blood glucose and albumin are independent risk factors for invasive syndrome in patients with Klebsiella pneumoniae liver abscess. According to this index, an effective predictive model for invasive syndrome of Klebsiella pneumoniae liver abscess can be constructed, which has certain clinical application value.