糖尿病肾脏疾病患者高尿酸血症风险模型的构建及验证

Construction and validation of hyperuricemia risk model in diabetic kidney disease patients

目的调查糖尿病肾脏疾病(DKD)患者高尿酸血症(HUA)的发生率和危险因素,并构建列线图预测模型以指导临床实践。方法采用目的抽样法选取2019年1月至2022年6月南京市高淳人民医院诊断DKD的患者219例为训练集,入院后根据HUA的定义将其分为HUA组(n=102)和非HUA组(n=117)。按照相同标准选取2022年7月至2023年7月105例DKD患者为验证集。记录患者的一般临床特征和近3个月的血生化平均值,分析DN患者发生HUA的危险因素,并建立列线图预测模型。结果体质指数(BMI)(OR=1.782)、糖化血红蛋白(HbA1c)(OR=2.601)和高脂血症(OR=1.669)是发生HUA的危险因素,而估计肾小球滤过率(eGFR)(OR=0.606)是其保护因素。建立列线图后计算训练集与验证集的一致性指数分别为0.854和0.802,校正曲线与理想曲线走势基本一致,受试者操作特征(ROC)曲线计算曲线下面积(AUC)分别为0.867(95%CI=0.802~0.923)和0.811(95%CI=0.745~0.872),而且临床净获益值较高,表明该模型具有良好的预测能力。结论DKD患者通常有较高的HUA患病率,高BMI、高Hb A1c和高脂血症是HUA的危险因素,高e GFR是保护因素。构建列线图模型对指导临床早期准确筛查HUA高风险群体具有较好的指导价值。

Objective To investigate the incidence and risk factors of hyperuricemia(HUA)in patients with diabetic kidney disease(DKD),then construct a nomogram predictive model to guide clinical practice.Method A retrospective summary was conducted on 219DKD patients diagnosed in our hospital from January 2019 to June 2022 as training set.After admission,they were divided into HUA group(n=102)and non-HUA group(n=117)according to the definition of HUA.Additionally,105 DKD patients from July 2022 to July 2023 were selected as validation set.The general clinical characteristics of patients and average blood biochemical values during the past 3 months were recorded,then univariate and Logistic regression analysis was to analyze the risk factors to HUA in DKD patients,and a nomogram predictive model was established.Result It showed that BMI(OR=1.782),Hb A1c(OR=2.601),and hyperlipidemia(OR=1.669)were risk factors to HUA,while e GFR(OR=0.606)was a protective factor.After establishing the nomogram,the consistency indices of training set and validation set were calculated to be 0.854 and 0.802,respectively.The correction curve and ideal curve trends were basically consistent,and the AUC calculated by ROC were 0.867(95%CI=0.802-0.923)and 0.811(95%CI=0.745-0.872),the clinical net benefit value was relatively high,which all indicated that the model had good predictive ability.Conclusion DKD patients usually have a higher prevalence of HUA,with high BMI,high Hb A1c,and hyperlipidemia being risk factors,while high e GFR as a protective factor.The construction of nomogram model has good value for guiding early and accurate screen of HUA high-risk groups in clinical practice.