基于血清25(OH)D水平构建2型糖尿病并发尿路感染风险的列线图预测模型  

Construction and evaluation of nomogram prediction model on risk of urinary tract infection in type 2 diabetes patients based on serum 25(OH)D levels

作　　者：周晶晶 李东风[1] 李时慧 郑丽丽 刘桂 丁丞章 袁幸运 ZHOU Jingjing;LI Dongfeng;LI Shihui;ZHENG Lili;LIU Gui;DING Chengzhang;YUAN Xingyun(Department of Endocrinology,the Affiliated Hospital of West Anhui Health Vocational College(Lu'an Second People’s Hospital),Lu'an 237000,China)

机构地区：[1]皖西卫生职业学院附属医院(六安市第二人民医院)内分泌科,安徽六安皖237000

出　　处：《安徽医学》2025年第2期141-147,共7页Anhui Medical Journal

基　　金：安徽省教育厅高校自然科学研究项目(编号:KJ2021B005);安徽省高校优秀青年人才支持计划项目(编号:GXYQ2022218)。

摘　　要：目的 检测2型糖尿病(T2DM)患者25羟维生素D[25(OH)D]水平,探讨T2DM患者尿路感染的危险因素,并构建T2DM并发尿路感染的预测模型。方法 选取2023年1月至2024年2月在皖西卫生职业学院附属医院(六安市第二人民医院)内分泌科就诊的607例T2DM患者为研究对象,收集研究对象的临床资料,电化学发光法检测血清25(OH)D水平,对尿路感染患者完善中段尿培养检查并收集培养结果。将研究对象依据入院时间,按照7∶3比例分为建模组(n=425)及验证组(n=182)。建模组采用logistic回归明确T2DM患者尿路感染的独立危险因素,应用R软件构建T2DM并发尿路感染的列线图预测模型,并应用内部(建模组)及外部(验证组)验证来评价模型的预测效能。结果 607例T2DM患者中,共有99例(16.31%)发生尿路感染,检出病原菌70株,其中以革兰阴性菌(64.29%)为主,尿路感染患者25(OH)D水平明显低于非感染患者,差异有统计学意义(P<0.05)。Logistic回归分析结果显示,女性、病程长、清蛋白偏低、高C反应蛋白(CRP)及低25(OH)D水平是建模组T2DM并发尿路感染的独立危险因素(P<0.05)。根据所筛选的危险因素构建T2DM并发尿路感染的列线图预测模型。模型内部验证的校正曲线显示观察值与预测值基本一致,HosmerLemeshow拟合优度检验显示拟合度良好(P>0.05),受试者工作特征(ROC)曲线下面积为0.909(95%CI:0.875~0.943),决策曲线分析显示当患者的阈值概率在10%~75%时,使用列线图预测尿路感染风险的净获益率更高。对验证组进行外部验证,结果与内部验证结果相一致。结论 T2DM合并尿路感染患者血清25(OH)D水平减低,基于25(OH)D及其它危险因素构建的列线图预测模型,可为T2DM患者尿路感染甄别及早期防治提供参考。Objective To detect the level of 25 hydroxyvitamin D[25(OH)D]and collect clinical data of patients with type 2 diabetes(T2DM),explore the risk factors of urinary tract infection in T2DM patients,and construct a nomogram predictive model of T2DM complicated with urinary tract infection.Methods A total of 607 T2DM patients admitted to the Department of Endocrinology of the Affiliated Hospital of West Anhui Health Vocational College(Lu’an Second People’s Hospital)from Jan 2023 to Feb 2024 were enrolled in the study.The clinical data of the patients were collected,and serum 25(OH)D levels were detected by electrochemiluminescence assay.For patients with urinary tract infection,mid stage urine culture was performed and the culture results were collected.According to the admission time,the subjects were divided into the training group(425 cases)and validation group(182 cases)based on a 7∶3 ratio.The independent risk factors for urinary tract infection in training group were identified by Logistic regression analysis,and a nomogram prediction model was constructed by R software.In-ternal(the training group)and external validation(the validation group)were applied to evaluate the predictive performance of the model.Re-sults Among 607 T2DM patients,99 cases(16.31%)complicated with urinary tract infection.A total of 70 strains of pathogenic bacteria were detected,and the majority(64.29%)of bacteria were gram-negative bacteria.The level of 25(OH)D in patients with urinary tract infection was significantly lower than that in non-infected patients(P<0.05).Logistic regression showed that women,long course of disease,low albumin,high C-reactive protein(CRP),and low 25(OH)D were independent factors for urinary tract infections in T2DM patients of training group(P<0.05).Based on the risk factors,a nomogram prediction model was constructed for T2DM complicated with urinary tract infection.Internal vali-dation of the model showed that the observed values were consistent with the predicted values in calibration curve.The Hosme

关 键 词：2型糖尿病 25羟维生素D 尿路感染 危险因素 列线图 预测模型 

分 类 号：R58[医药卫生—内分泌]