神经外科重症监护病房患者多重耐药菌感染的影响因素分析及诺谟图构建  

Influencing factors of multidrug-resistant bacterial infection among patients in neurosurgical intensive care unit and construction of a risk prediction nomogram

作　　者：周雪莲 于洪威 李阳 邓卓君 苗晓 徐艳 Zhou Xuelian;Yu Hongwei;Li Yang;Deng Zhuojun;Miao Xiao;Xu Yan(Department of Neurosurgery,the First Affiliated Hospital of Kangda College of Nanjing Medical University,Lianyungang 222000,Jiangsu Province,China;Department of Trauma and Orthopedics,the First Affiliated Hospital of Kangda College of Nanjing Medical University,Lianyungang 222000,Jiangsu Province,China;Nursing College of Nanjing Medical University,Nanjing 210000,China)

机构地区：[1]南京医科大学康达学院第一附属医院神经外科,连云港222000 [2]南京医科大学康达,学院第一附属医院创伤骨科,连云港222000 [3]南京医科大学护理学院,南京210000

出　　处：《中华临床感染病杂志》2024年第4期291-296,共6页Chinese Journal of Clinical Infectious Diseases

基　　金：2021年南京医科大学康达学院科研发展基金课题(KD2021KYJJZD098)。

摘　　要：目的:分析神经外科重症监护病房(neurosurgical intensive care unit,NSICU)患者多重耐药菌(multidrug-resistant organism,MDRO)感染的危险因素并构建诺谟图。方法:选取2021年8月至2022年10月南京医科大学康达学院第一附属医院NSICU收治的434例患者作为研究对象,按照1∶1的比例将其分为建模组(n=217)和验证组(n=217)。采用多因素Logistic回归分析NSICU患者MDRO感染的危险因素,并使用R软件构建诺谟图。采用受试者工作特征曲线(ROC)和Hosmer-Lemeshow(H-L)拟合优度评价模型的区分度和校准度。结果:多因素Logistic回归分析结果显示,基础疾病≥3个[OR=2.580,95%可信区间(CI)1.322~5.035]、联合使用抗菌药物>10 d(OR=2.336,95%CI 1.182~4.615)、低蛋白血症(OR=1.962,95%CI 1.031~3.735)、侵入性操作时间≥10 d[10~20 d:OR=2.358,95%CI 1.048~5.306;>20 d:OR=3.486,95%CI 1.643~7.395]、格拉斯哥昏迷评分≤8分(OR=2.961,95%CI 1.470~5.963)是NSICU患者MDRO感染的影响因素。诺谟图预测建模组MDRO感染的ROC曲线下面积(AUC)为0.787(95%CI 0.725~0.849),敏感度为73.3%,特异度为72.5%,H-L检验结果为χ^(2)=7.482,P=0.486,校准曲线与理想曲线较接近,平均绝对误差为0.022;预测验证组MDRO感染的AUC为0.800(95%CI 0.739~0.861),敏感度为74.7%,特异度为73.9%,H-L检验结果为χ^(2)=9.824,P=0.278,校准曲线与理想曲线较接近,平均绝对误差为0.015。结论:基于危险因素的诺谟图可以有效预测NSICU患者MDRO感染的风险,可用于临床实践。Objective To analyze the risk factors of multidrug-resistant organism(MDRO)infection among patients in neurosurgery intensive care unit(NSICU)and to construct a risk prediction nomogram.Methods A total of 434 patients admitted in the NSICU of the First Affiliated Hospital of Kangda College of Nanjing Medical University from August 2021 to October 2022 were enrolled in the study.Patients were divided into modeling group(n=217)and validation group(n=217).Multivariate Logistic regression was used to analyze the risk factors of MDRO infection in patients,and R software was used to construct a risk prediction nomogram.The prediction power of the nomogram was evaluated with receiver operating characteristic(ROC)curve,the calibration of the model was assessed with Hosmer-Lemeshow(H-L)goodness-of-fit method.Results Multivariate Logistic regression analysis showed that the underlying disease≥3(OR=2.580,95%CI 1.322-5.035),the combination of antimicrobial drugs>10 days(OR=2.336,95%CI 1.182-4.615),hypoproteinemia(OR=1.962,95%CI 1.031-3.735),invasive operation time(10-20 d:OR=2.358,95%CI 1.048-5.306;>20 d:OR=3.486,95%CI 1.643-7.395)and GCS≤8 points(OR=2.961,95%CI 1.470-5.963)were independent risk factors of MDRO infection among patients in NSICU.The area under the curve(AUC)of the nomogram in predicting the risk of MDRO infection for patients in modeling group was 0.787(95%CI 0.725-0.849)with a sensitivity of 73.3%and a specificity of 72.5%,the H-L test results wereχ^(2)=7.482,P=0.486,the calibration curve was close to the ideal curve,and the mean absolute error was 0.022.The AUC of the nomogram in predicting MDRO infection for patients in verification group was 0.800(95%CI 0.739-0.861)with a sensitivity of 74.7%and a specificity of 73.9%,the H-L test results wereχ^(2)=9.824,P=0.278,the calibration curve was close to the ideal curve,and the average absolute error was 0.015.Conclusion The nomogram constructed based on the risk factors can effectively predict the risk of MDRO infection for patients in neurosurgical ICU,wh

关 键 词：神经外科 重症监护病房 多重耐药 诺谟图 

分 类 号：R446.5[医药卫生—诊断学] R651[医药卫生—临床医学]