完全植入式输液港术后1周感染预测模型的构建与验证

Construction and validation of a prediction model for postoperative infection within one week of totally implantable venous access port surgery

摘要:

目的 探讨完全植入式输液港(TIVAP)术后1周内发生感染的危险因素, 并建立预测模型。 方法 纳入行TIVAP术患者650例, 其中, 2017年1月—2022年1月480例作为建模集, 2022年2月—2023年3月170例作为验证集。建模集按是否发生术后感染分为感染组17例和非感染组463例。使用t检验和χ²检验比较两组患者的一般情况及临床特征, 二分类logistic回归分析术后感染的危险因素。通过列线图对预测模型进行可视化表达。 结果 建模集感染组17例患者中6例为切口感染, 10例为囊袋感染, 仅1例为血流感染。Logistic回归分析显示, 年龄≥65岁[OR=3.093, 95%CI: 1.048~9.128, P=0.041]、手术时间≥2 h[OR=3.286, 95%CI: 1.008~10.716, P=0.049], 以及囊袋血肿[OR=20.200, 95%CI: 6.750~60.455, P＜0.001]是TIVAP术后感染的独立危险因素。受试者工作特征(ROC)曲线显示曲线下面积为0.833, 灵敏度为70.6%, 特异度为88.1%。校准曲线分析显示预测模型与实际发生风险一致性较好。决策曲线分析(DCA)显示此模型在风险阈概率1%~60%具有临床获益。 结论 年龄≥65岁、手术时间≥2 h、囊袋血肿是腋静脉TIVAP术后1周感染的独立危险因素, 以此建立的预测模型准确度及校准度较好。医务人员应牢记感染高危因素和术后感染护理流程, 加强对高危患者的术后观察。

Abstract:

Objective To explore the risk factors for the infection within one week after totally implantable venous access port(TIVAP)surgery, and establish a prediction model. Methods 650 patients undergoing TIVAP surgery were included in the analysis. 480 cases from January 2017 to January 2022 were as the modeling set, and 170 cases from February 2022 to March 2023 were as the validation set. The modeling set was further divided into the infection group (n=17) and the non-infection group (n=463) according to the occurrence of postoperative infection. General condition and clinical characteristics of patients in two groups were compared using t-test and χ² test, and risk factors for postoperative infection were analyzed with binary logistic regression. Visualization was realized through nomogram. Results Among the 17 patients in the infection group of the modeling set, 6 had incisional infection, 10 had pocket infection, and only 1 had blood stream infection. Logistic regression analysis showed that age ≥65 years old (OR=3.093, 95%CI: 1.048-9.128, P=0.041), duration of surgery ≥2 hours (OR=3.286, 95%CI: 1.008-10.716, P=0.049), and pocket hematoma (OR=20.200, 95%CI: 6.750-60.455, P < 0.001) were independent risk factors for TIVAP postoperative infection. The receiver operating characteristic (ROC) curve showed an area under the curve of 0.833, sensitivity of 70.6%, and specificity of 88.1%, respectively. Calibration curve analysis showed a good consistency of the predictive model with the actual risk occurrence. Decision curve analysis (DCA) showed that the model had clinical benefits with risk threshold probability between 1% and 60%. Conclusion Age ≥65 years old, duration of surgery ≥2 hours, and pocket hematoma are independent risk factors for the infection within one week after TIVAP surgery. Prediction model constructed based on these factors has a good accuracy and calibration. Healthcare workers should keep in mind the high risk factors of infection and postope- rative infection care procedures, and strengthen postoperative observation on high risk patients.