维持性血液透析医院感染风险评估模型建立及其应用价值

Establishment and application value of nosocomial infection risk assessment model for the prediction of infection in maintenance hemodialysis patients

目的建立维持性血液透析(MHD)患者医院感染的预测模型,并对其应用价值进行评价。方法选择温州医科大学附属第二医院肾内科2017年3月-2020年3月进行维持性血液透析的慢性肾病患者180例作为研究对象,根据其透析期间是否发生医院感染分为感染组(n=81)和非感染组(n=99)。统计患者呼吸道感染、尿路感染、消化道感染、皮肤感染及导管感染情况,归纳患者发生医院感染危险因素并以此建立模型,采用受试者工作特征(ROC)曲线对建立模型与单一指标对感染的诊断效能及建立模型对不同类型感染的诊断效能进行评价,采用拟合优度检测法评价所建立模型与实际观测值间的差异。结果81例感染组患者以呼吸道感染、尿路感染和消化道感染为主;初次通路类型、合并糖尿病、合并低蛋白血症、透析龄、营养控制状态(CONUT)评分、C-反应蛋白、血红蛋白是维持性血液透析患者发生医院感染的影响因素(P<0.05);ROC分析结果显示,与透析龄、CONUT评分、C-反应蛋白、血红蛋白单一指标相比,本研究建立的医院感染模型对维持性血液透析患者医院感染的诊断效能更高(曲线下面积为0.941);ROC分析结果显示,本研究建立的医院感染模型对于呼吸道感染、尿路感染、消化道感染、皮肤感染、导管感染不同类型的感染均具有较好的诊断效能,其AUC分别为0.916、0.934、0.908、0.957,0.895,拟合优度检验结果显示,本研究建立的维持性血液透析医院感染预测模型与实际观测值之间差异无统计学意义(P=0.663),R^(2)=0.803,提示该模型的预测能力较为准确。结论基于医院患者临床资料所建立的预测模型对维持性血液透析患者发生医院感染具有良好的效能,可作为临床指导MHD患者管理及治疗的参考。

OBJECTIVE To establish a predictive model of nosocomial infection in maintenance hemodialysis patients,and to evaluate the application value of this model.METHODS A total of 180 patients with chronic kidney disease who underwent maintenance hemodialysisin the department of Nephrology,The Second Affiliated Hospital of Wenzhou Medical University from Mar.2017 to Mar.2020 were recruited as the research subjects.According to whether nosocomial infection occurred during dialysis,they were divided into the infection group(n=81)and non-infected group(n=99).Patients with respiratory tract infection,urinary tract infection,digestive tract infection,skin infection and catheter infection were counted and risk factors of nosocomial infection were summarized.Based on this,a model was established.Receiver operating characteristic curves(ROCs)of single indicator and combined indicators were drawn to evaluate the diagnostic efficiency in different types of infection.The difference between the established model and the actual observations was evaluated by the goodness of fit test.RESULTS Eighty-one patients in the infection group were mainly with respiratory tract infection,urinary tract infection and digestive tract infection.Primary access type,diabetes mellitus,hypoalbuminemia,higher age on dialysis,higher CONUT score,elevated C-reactive protein,hemoglobin Elevated were independent risk factors for nosocomial infection in maintenance hemodialysis patients(P<0.05).ROC analysis showed that,compared with the single indicators such as dialysis age,CONUT score,C-reactive protein and hemoglobin,the model established in this study was more effective in diagnosing nosocomial infection in maintenance hemodialysis patients(AUC=0.941).ROC analysis showed that the nosocomial infection model established in this study was effective in the diagnosis of respiratory tract infection,urinary tract infection,digestive tract infection,skin infection and catheter infection,with the areas under the curves(AUCs)were 0.916,0.934,0.908,0.957,0.895,respectively.The goodness-of-fit test results showed that the difference of values between the maintenance hemodialysis nosocomial infection prediction model established in this study and actual observations were not significant(P=0.663,R^(2)=0.803),which indicated that the predictive ability of the model was more precise.CONCLUSION The prediction model established based on the clinical data of hospital patients has a good effect on the occurrence of nosocomial infection in maintenance hemodialysis patients,which can be used as an important reference for clinical guidance of the management and treatment of MHD patients.