某医院呼吸道感染多重耐药菌感染风险Nomogram模型建立和验证  

Establishment and validation of a Nomogram model for the risk of multidrug-resistant bacteria infection in a hospital with respiratory tract infection

作　　者：李燕[1] 朱金玉[1] 张羽[1] 孙洋 闫小燕 徐真[1] LI Yan;ZHU Jinyu;ZHANG Yu;SUN Yang;YAN Xiaoyan;XU Zhen(Department of Infection Prevention and Control,Nanyang Central Hospital,Nanyan,Henan,China,473000)

机构地区：[1]南阳市中心医院感染防控科,河南南阳473000

出　　处：《分子诊断与治疗杂志》2023年第4期673-676,681,共5页Journal of Molecular Diagnostics and Therapy

基　　金：河南省医学科技攻关计划联合共建项目(LHGJ20191457)。

摘　　要：目的 构建及验证呼吸道感染住院病例多重耐药菌(MDRB)感染风险Nomogram模型。方法 选取2017年至2021年在某综合医院呼吸道感染住院病例的9 488例,并随机抽取2 800例为验证组,其余6 688例为建模组,收集患者的临床资料分析MDRB感染的影响因素,并构建及验证MDRB感染的Nomogram模型。结果 9 488例呼吸道感染住院病例中检出病原菌有1 398例,共培养分离病原菌数1 478株,其中革兰阴性菌1 109株占75.03%,革兰阳性菌346株占23.41%,真菌23株占1.56%。MDRB感染有144例,占10.30%(144/1 398)。多因素Logistic回归分析,年龄、抗菌药物应用种类、住院时间、机械通气及留置导管是呼吸道感染住院病例MDRB感染的独立危险因素(χ^(2)=4.703、9.552、39.500、37.662、19.817,P<0.05)。基于独立危险因素建立呼吸道感染住院病例MDRB感染的Nomogram风险预测模型,结果显示建模组、验证组AUC值为0.758、0.756,提示诊断效能较好;Hosmer-Lemeshow检验显示(建模组P=0.989,验证组P=0.341),提示拟合优度良好。结论 呼吸道感染住院病例检出病原菌主要以革兰阴性菌为主,且存在较高的多药耐药菌感染,建立相关Nomogram风险预测模型有利于医护人员筛查相关危险因素并采取对应干预措施,从而降低MDRB感染情况,进而减轻患者的治疗难度。Objective To construct and validate the risk Nomogram model of multidrug-resistant bacteria(MDRB)infection in hospitalized patients with respiratory tract infection.Methods A total of 9 488cases of respiratory infections hospitalized in a general hospital from 2017 to 2021 were selected,and randomly selected 2800 cases as the validation group and the remaining 6 688 cases as the modeling group.The clinical data of patients were collected to analyze the factors influencing multi-drug resistant bacterial infections and to construct and validate the Nomogram model of MDRB.Results Of the 9 488 hospitalized cases of respiratory infection,1 398 pathogenic bacteria were detected,and 1 478 strains of pathogenic bacteria were isolated from co-culture,of which 1 109 strains of Gram-negative bacteria accounted for 75.03%,346 strains of Grampositive bacteria accounted for 23.41%,and 23 strains of fungi Accounting for 1.56%.There were 144 cases of MDRB infection,accounting for 10.30%(144/1 398).Multivariate Logistic regression analysis showed that age,type of antibiotics used,length of hospital stay,mechanical ventilation and indwelling catheter were independent risk factors for MDRB infection in hospitalized patients with respiratory tract infection(χ^(2)=4.703、9.552、39.500、37.662、19.817,P<0.05).A Nomogram risk prediction model for MDRB in hospitalized cases of respiratory tract infections based on the above independent risk factors was established.The results showed the AUC values of the modeling group and the verification group were 0.758 and 0.756,indicating that the diagnostic efficiency was better;The Hosmer-Leme test show test showed(P=0.989 for the modeling group and P=0.341 for the validation group),suggesting a good goodness of fit.Conclusion The pathogenic bacteria detected in hospitalized cases of respiratory tract infections are mainly Gram-negative bacteria,and there is a high level of multi-drug resistant bacterial infections.The establishment of a relevant Nomogram risk prediction model will help heal

关 键 词：呼吸道感染 病原学 多重耐药菌感染 风险预测模型 

分 类 号：R446.5[医药卫生—诊断学]