负压封闭引流技术治疗全身软组织感染疗效的影响因素  

Influencing factors of efficacy of vacuum sealing drainage technique in the treatment of systemic soft tissue infection

作　　者：郑志学 毕敬涛[1] 蔡轩 刘亚奇[1] ZHENG Zhi-xue;BI Jing-tao;CAI Xuan;LIU Ya-qi(Department of General Surgery,Beijing Jishuitan Hospital,Beijing 100000,China)

机构地区：[1]北京积水潭医院普外科,北京100000

出　　处：《局解手术学杂志》2022年第5期443-447,共5页Journal of Regional Anatomy and Operative Surgery

摘　　要：目的探讨影响负压封闭引流(VSD)技术治疗全身软组织感染效果的因素。方法纳入我院收治的98例全身软组织感染行VSD技术治疗的患者,根据治疗效果将其分为疗效显著组(n=69)和疗效欠佳组(n=29)。单因素分析2组患者的临床资料。以临床疗效为因变量,以单因素分析中有统计学意义的指标为自变量进行多因素Logistic回归分析;通过XGBoost算法模型分析影响因素的重要度。绘制受试者工作特征(ROC)曲线分析XGBoost算法模型和多因素Logistic回归分析模型的预测效能,并验证模型的校准度。结果单因素分析显示,2组患者抗菌药物使用时间、糖尿病史、血小板计数(PLT)、血小板分布宽度(PDW)、白细胞计数(WBC)、BMI比较,差异有统计学意义(P<0.05);多因素Logistic回归分析显示,糖尿病史、WBC≥12.35×10^(9)/L、抗菌药物使用时间≥9 d、PDW≥12.38%、PLT≥132.27×10^(9)/L、BMI≥24.62 kg·m^(-2)是影响疗效的独立危险因素(P<0.05);XGBoost算法模型中影响因素排序依次为糖尿病史、WBC、抗菌药物使用时间、PDW、PLT、BMI。XGBoost算法模型和多因素Logistic回归分析模型的曲线下面积分别为0.791、0.789,灵敏度分别为79.9%、77.6%,特异度分别为78.2%、79.5%,约登指数分别为0.581、0.571。模型校准度方面,疗效欠佳组两个模型比较校准度的Hosmer and Lemeshow Testχ^(2)=0.225,P=0.367;疗效显著组两个模型比较校准度的Hosmer and Lemeshow Testχ^(2)=0.216,P=0.381。结论使用VSD技术治疗全身软组织感染时,若患者有糖尿病史、WBC≥12.35×10^(9)/L、抗菌药物使用时间≥9 d、PDW≥12.38%、PLT≥132.27×10^(9)/L、BMI≥24.62 kg·m^(-2),临床应给予重点关注,减少抗菌药物的使用,患者应控制饮食,增强体质,以提高疗效。Objective To explore the factors affecting the effect of vacuum sealing drainage(VSD)technique in the treatment of systemic soft tissue infection.Methods A total of 98 patients with systemic soft tissue infection treated by VSD technique in our hospital were included and they were divided into the significant effect group(n=69)and the poor effect group(n=29)according to their treatment effects.The univariate analysis was used to analyze the clinical data of patients between the two groups.Multivariate Logistic regression analysis was performed,with the clinical efficacy as the dependent variable and the indicators with statistically significant differences in the univariate analysis as the independent variables;the importance degree of the influencing factors was analyzed by the XGBoost algorithm model.The receiver operating characteristic(ROC)curve was drawn to analyze the prediction efficiency of the XGBoost algorithm model and the multivariate Logistic regression analysis model,and the calibration degrees of the models were verified.Results The univariate analysis showed that there were statistically significant differences in the use duration of antibacterial agents,history of diabetes mellitus,platelet count(PLT),platelet distribution width(PDW),white blood cell count(WBC),and BMI between the two groups(P<0.05);the multivariate Logistic regression analysis showed that the history of diabetes mellitus,WBC≥12.35×10^(9)/L,use duration of antibacterial agents≥9 days,PDW≥12.38%,PLT≥132.27×10^(9)/L,BMI≥24.62 kg·m^(-2) were the independent risk factors that affected the efficacy(P<0.05);the order of influencing factors in the XGBoost algorithm model was the history of diabetes mellitus,WBC,use duration of antibacterial agents,PDW,PLT,and BMI.The areas under the curve of the XGBoost algorithm model and the multivariate Logistic regression analysis model were 0.791 and 0.789,the sensitivities were 79.9%and 77.6%,the specificities were 78.2%and 79.5%,and the Youden indexes were 0.581 and 0.571,respectivel

关 键 词：负压封闭引流技术 全身软组织感染 临床疗效 影响因素 

分 类 号：R632.9[医药卫生—外科学]