头颈部多间隙感染患者继发下行性坏死性纵隔炎的危险因素及其预测模型构建

Establishment of prediction model and risk factors for secondary descending necrotizing mediastinitis in patients with multiple space infection of head and neck region

目的探究头颈部多间隙感染患者继发下行性坏死性纵隔炎(DNM)的危险因素及预测模型构建。方法回顾性选取2018年10月-2022年10月贵州省人民医院收治的331例头颈部多间隙感染患者作为研究对象,根据是否继发DNM分为继发组40例、非继发组291例,分析DNM现况及病原菌分布特点,Logistic回归分析法分析DNM的危险因素,构建预测模型,并分析其对头颈部多间隙感染患者DNM发生的预测价值。结果331例头颈部多间隙感染患者中DNM为40例,继发率为12.08%,继发组40例患者共分离出40株病原菌,其中以革兰阳性菌为主(27株,67.50%);多因素Logistic分析结果显示,有糖尿病、积气、咽旁间隙、C-反应蛋白(CRP)水平高为头颈部多间隙感染患者继发DNM的危险因素(P<0.05);构建的回归方程模型:Logit(P)=-30.879+合并糖尿病×2.193+CRP水平×0.236+积气×4.387+咽旁间隙×3.211,受试者工作特征(ROC)曲线结果显示,当Logit(P)>0.082时,曲线下面积(AUC)值为0.995,诊断敏感度为97.50%,特异度为94.85%。结论头颈部多间隙感染患者DNM的发生率较高,危险因素较多,感染以革兰阳性菌为主,构建的模型对DNM的预测价值较高,临床可针对此及时制定措施以预防DNM的发生。

OBJECTIVE To explore the risk factors for secondary descending necrotizing mediastinitis(DNM)in the patients with multiple space infection of head and neck region and establish the prediction model.METHODS A total of 331 patients with multiple space infection of head and neck region who were treated in Guizhou Provincial People′s Hospital from Oct 2018 to Oct 2022 were recruited as the research subjects and were divided into the secondary group with 40 cases and the non-secondary group with 291 cases according to the status of secondary DNM.The prevalence of DNM and distribution of pathogens were observed,logistic regression analysis was performed for the risk factors for the DNM,the prediction model was established,and the value of the model in prediction of the DNM was analyzed.RESULTS Among the 331 patients with multiple space infection of head and neck region,40 had DNM,with the incidence of secondary DNM 12.08%.Totally 40 strains of pathogens were isolated from the 40 patients of the secondary group,27(67.50%)of which were gram-positive bacteria.The result of multivariate logistic analysis showed that diabetes mellitus,pneumatosis,parapharyngeal space and high level of C-reactive protein(CRP)were the risk factors for the secondary DNM in the patients with multiple space infection of head and neck region(P<0.05).The established regression equation model was as follows:Logit(P)=-30.879+complication with diabetes mellitus×2.193+CRP level×0.236+pneumatosis×4.387+parapharyngeal space×3.211.The result of receiver operating characteristic(ROC)curve analysis showed that when Logit(P)was more than 0.082,the area under curve(AUC)was 0.995,with the sensitivity 97.50%for diagnosis,the specificity 94.85%.CONCLUSION The incidence of DNM is high among the patients with multiple space infection of head and neck region,and there are a variety of risk factors.The gram-positive bacteria are dominant.The established model has high value in prediction of DNM,and it is necessary for the hospital to formulate prevention measures so as to prevent the DNM.