肝癌术后感染性并发症的危险因素分析及构建预测模型

Analysis of the Risk Factors for Postoperative Infectious Complications in Hepatocellular Carcinoma after Hepatectomy and Construction of a Predictive Model

摘要: 目的：分析肝癌切除术后感染性并发症的危险因素，并构建预测模型。方法：回顾性收集江苏省人民医院宿迁医院2017年6月~2023年6月行肝癌切除术患者的临床资料，采用单因素分析及多因素Logistic回归分析筛选出肝癌切除术后感染性并发症的独立因素，基于独立危险因素建立列线图预测模型，并检验其预测效能。结果：共纳入215例患者，42例患者术后发生感染性并发症，发生率为19.5%。共培养出72株病原菌。Logistic回归结果显示，年龄(≥65岁)、合并糖尿病、腹腔引流管留置时间(≥7 d)及Alb (<35 g/L)为肝癌切除术后感染性并发症发生的危险因素(P < 0.05)。基于以上结果构建列线图预测模型，校准曲线显示该列线图模型的预测发生概率和实际发生概率之间一致性较好，Bootstrap法计算出C-index值为0.678，区分度良好，受试者工作特征曲线下面积(AUC)为0.768 (95% CI: 0.688~0.848)，决策分析曲线显示价值性高。结论：年龄(≥65岁)、合并糖尿病、腹腔引流管留置时间(≥7 d)及Alb (<35 g/L)为肝癌切除术后感染性并发症发生的危险因素，基于以上危险因素构建的预测模型具有一定预测价值。

Abstract: Objective: To analyze the risk factors of postoperative infectious complications (PIC) in hepatocellular carcinoma (HCC) after hepatectomy and construct a predictive model. Methods: Retrospective collection of clinical data from patients undergoing HCC resection in Jiangsu Province (Suqian) Hospital from June 2017 to June 2023. Single factor analysis and multivariate logistic regression analysis were used to screen for independent risk factors of infectious complications after hepatectomy. Establishing a column chart prediction model and testing its predictive effectiveness based on independent risk factors. Results: A total of 215 patients were included, and 42 patients developed PIC after hepatectomy, with an incidence rate of 19.5%. A total of 72 pathogenic bacteria were cultivated. Logistic regression results showed that age (≥65 years old), diabetes, retention time of peritoneal drainage tube (≥7 days) and Alb (<35 g/L) were risk factors for HCC resection of PIC (P < 0.05). Establishing a column chart prediction model based on the above results, the calibration curve shows good consistency between the predicted probability of occurrence and the actual probability of occurrence of the column graph model. The Bootstrap method calculated a C-index value of 0.678, indicating good discrimination. The area under the Receiver Operating Characteristic was 0.768 (95% CI: 0.688~0.848). The decision analysis curve shows high value. Conclusions: Age (≥65 years old), diabetes, retention time of peritoneal drainage tube (≥7 days) and Alb (<35 g/L) were risk factors for PIC after HCC resection. The prediction model constructed based on the above risk factors has certain predictive value.