血培养中鉴别凝固酶阴性葡萄球菌为致病菌回归模型的建立与临床评价

Establishment and clinical evaluation on a regression model of coagulase-negative Staphylococcus as pathogenic bacterium in blood culturing

目的建立以血培养中凝固酶阴性葡萄球菌(CNS)为血流感染致病菌的鉴别模型,并对其临床价值进行评价。方法回顾性分析2013年1月—2015年6月南京医科大学第一附属医院的住院患者178例,血培养报告为CNS,根据相应的判断标准,将其分为致病菌组(104例)和污染菌组(74例),记录血培养采集前患者的体温(T)、白细胞(WBC)计数、中性粒细胞(NEU)比值、常规C反应蛋白(CRP)、降钙素原(PCT)、红细胞沉降率(ESR)、血糖(GLU)、血培养首瓶报阳时间(TTP)及血培养报阳瓶数(N)9项指标,并进行统计学分析。结果致病菌组T、WBC计数、NEU比值、CRP、PCT、ESR和N均高于污染菌组,而GLU和TTP均低于污染菌组(P<0.05),各单项指标诊断CNS为致病菌的受试者工作特征(ROC)曲线的曲线下面积(AUC)排前3位的是TTP(0.870)、PCT(0.869)和NEU比值(0.833);用前进法构建的致病菌鉴别回归模型由T、PCT、TTP和NEU比值4项指标组成,回归模型诊断CNS为致病菌的AUC为0.959。当回归模型的诊断界值定为0.50时,其敏感性为0.918,特异性为0.900,约登指数达到最大值(0.820),此时回归模型的诊断正确率为92.0%,回归模型诊断与临床诊断一致性分析的Kappa系数达到了0.827。结论由T、PCT、TTP和NEU比值构建的回归模型对临床快速判断CNS是否为血流感染的致病菌以及指导抗菌药物的合理使用具有重要价值。

Objective To establish and evaluate a discriminant model of coagulase-negative Staphylococcus(CNS) as pathogenic bacterium in blood culturing. Methods A total of 178 inpatients with blood culturing report of CNS were analyzed retrospectively from January 2013 to June 2015 in the First Affiliated Hospital with Nanjing Medical University. According to the related criterion,these inpatients were classified into 2 groups,pathogenic bacterium group (104 cases) and contaminating bacterium group (74 cases). Before blood culturing,9 indices,including temperature(T),white blood cell (WBC)count,neutrophil (NEU)ratio,C-reactive protein(CRP),procalcitonin (PCT),erythrocyte sedimentation rate (ESR),glucose (GLU),time of the first bottle reported to be positive (TTP) and the number of bottles reported to be positive (N),were recorded.Results T,WBC count,NEU ratio,CRP,PCT,ESR and N in pathogenic bacterium group were higher than those in contaminating bacterium group,while GLU and TTP were lower (P<0.05). For the areas under receiver operating characteristic (ROC) curve (AUC) for the diagnosis of CNS,those of TTP(0.870),PCT(0.869)and NEU regression ratio(0.833) ranked the top 3. The regression model of CNS as pathogenic bacterium established through forward method consisted of T,PCT,TTP and NEU ratio,and AUC of the regression model was 0.959. When the cut-off value was 0.50,the sensitivity was 0.918,the specificity was 0.900,Youden index reached its maximum of 0.820,the diagnosis accuracy of the regression model was 92.0%,and the Kappa value in consistency analysis between the regression model diagnosis and clinical diagnosis was 0.827. Conclusions The regression model established with T,PCT,TTP and NEU ratio plays a role for diagnosing rapidly whether CNS is a pathogenic bacterium of blood stream infection or not and guiding the rational use of antibiotics.