尿常规结果对革兰氏阳性菌或阴性菌引起尿路感染评分预测模型的建立和验证  

Establishment and validation of urinary tract infection score prediction model caused by gram-positive or negative bacteria

作　　者：杨开洪 曾庆栈 冯斌 YANG Kaihong;ZENG Qingzhan;FENG Bin(Department of Laboratory Medicine,Yangjiang Hospital of Traditional Chinese Medicine,Guangdong Province,Yangjiang529599,China)

机构地区：[1]广东省阳江市中医医院检验科,广东阳江529599

出　　处：《中国当代医药》2022年第10期24-28,共5页China Modern Medicine

摘　　要：目的利用尿常规结果构建革兰氏阳性菌(G+)或革兰氏阴性菌(G-)引起尿路感染(UTI)的评分模型,并验证评分模型的预测效能。方法回顾性分析2019年1月至2021年5月于广东省阳江市中医医院诊疗的182例UTI患者为研究对象,按照随机数字表法分为建模组(91例,G-菌59例,G+菌32例)和验证组(91例,G-菌59例,G+菌32例),通过二元logistic回归分析筛选相关影响因素,并构建G+菌感染的风险预测模型,判断模型的预测能力。结果共检出病原菌225株,其中G-菌株138株,G+菌株75株,真菌12株。建模组细菌计数预测G-菌感染的AUC为0.793,最佳截断值为320个/μl,白细胞(WBC)计数预测G-菌感染的AUC为0.699,最佳截断值为64个/μl。二元logistic回归分析显示,亚硝酸盐(NIT)阴性、离心涂片革兰氏染色G+菌、细菌计数<320.00个/μl、WBC计数<64个/μl均为UTI患者病原菌为G+菌的影响因素,分别赋值11、4、5、4分。建模组模型AUC为0.874,验证组AUC为0.877。对患者G+菌感染进行风险分层,0~11分为低G+菌风险,13~19分为中G+菌风险,20~24分为高G+菌风险。结论构建的评分预测模型有较好的区分度和校准度,方便临床根据模型对病原菌进行预测和临床用药。Objective To establishment and validation of urinary tract infection(UTI)score prediction model caused by gram-positive(G+)or gram-negative bacteria(G-),and verify the prediction efficiency of the scoring model.Methods A total of 182 patients with UTI treated in Yangjiang Hospital of Traditional Chinese Medicine,Guangdong Province from January 2019 to May 2021 were retrospectively analyzed.According to random number table method,they were divided into modeling group(91 cases,59 cases of G-bacteria,G+bacteria 32 cases)and verification group(91 cases,59 cases of G-bacteria,G+bacteria 32 cases).The risk prediction model of G+bacteria infection was constructed by binary logistic regression analysis to screen the relevant influencing factors,and the prediction ability of the model was judged.Results A total of 225 strains were detected,including 138 G-strains,75 G+strains and 12 fungi.The AUC of G-bacterial infection predicted by bacterial count in the modeling group was 0.793,and the best cut-off value was 320/μl.The AUC of G-bacterial infection predicted by white blood cell(WBC)count was 0.699,and the best cut-off value was 64/μl.Binary logistic regression analysis showed that nitrite(NIT)negative,urinary sediment stained G+bacteria,bacterial count<320.00/μl and WBC count<64/μl were the influencing factors of G+bacteria in UTI patients,and 11,4,5 and 4 points were assigned respectively.The AUC of model subjects in the modeling group was 0.874 and that in the validation group was 0.877.The risk of G+bacteria infection was stratified,0-11 points was divided into low G+bacteria risk,13-19 points was medium G+bacteria risk,and 20-24 points was high G+bacteria risk.Conclusion The score prediction model has good discrimination and calibration,which is convenient for the prediction of pathogens and clinical medication according to the model.

关 键 词：尿常规 尿路感染 革兰氏染色 尿有形成分检测 尿干化学检测 

分 类 号：R446.12[医药卫生—诊断学]