老年糖尿病肾病预测数学模型的建立与临床价值

The Establishment and Clinical Value of a Mathematical Model for Predicting Senile Diabetic Nephropathy

目的尝试建立一个对DN具有较高诊断效能的数学评估模型,从而为DN诊断提供新的理论依据。方法收集2013年4月—2015年8月该院≥60岁确诊2型糖尿病(DM)患者212例,根据尿白蛋白排泄率(UAER)将入选对象进行分组。DN组:大量白蛋白尿组(UAER>300 mg/d)+微量白蛋白尿组(30 mg/d≤UAER<300 mg/d);正常组(UAER<30 mg/d)。收集性别、年龄、血压等一般信息,测定患者N末端前体脑钠肽(NT-proBNP)、胱抑素C(CysC)、高敏C反应蛋白(hs-CRP)、尿酸(UA)、血肌酐(Scr)等指标。采用受试者工作特征曲线(ROC曲线)下面积评价每一指标在DN中的诊断效能;选择ROC曲线下面积大的指标入模。通过ROC曲线下面积评价各评估模型在DN中的诊断效能,获得最佳预测模型,并确定最佳参考诊断点。将模型与UAER诊断DN能力进行比较,进一步验证模型在DN诊断中的可重复性和有效性。结果模型NT-ProBNP+CysC具有最大的ROC曲线下面积为0.861,敏感性为0.826,特异性为0.784,该模型对DN预测价值的最佳临界点为204.36。具有较好的预测效能,提高了实验室指标对老年DN的预测效能。结论通过建立数学评估模型可显著提升各指标在判断是否患DN的预测效能,该评估模型为NT-ProBNP+CysC,在≥60岁老年DN中具有较好的预测效能。

Objective To try to establish a mathematical evaluation model with high diagnostic efficiency for DN,so as to provide a new theoretical basis for the diagnosis of DN.Methods A total of 212 patients with diagnosed type 2 diabetes(DM)aged≥60 years in the hospital from April 2013 to August 2015 were collected.The selected subjects were divided into groups according to the urine albumin excretion rate(UAER).DN group:massive albuminuria group(UAER>300 mg/d)+microalbuminuria group(30 mg/d≤UAER<300 mg/d);normal group(UAER<30 mg/d).Collected general information such as gender,age,blood pressure,and determine the patient’s N-terminal precursor brain natriuretic peptide(NT-proBNP),cystatin C(CysC),high sensitivity C-reactive protein(hs-CRP),uric acid(UA),and blood creatinine(Scr)and other indicators.The area under the receiver operating characteristic curve(ROC curve)was used to evaluate the diagnostic efficacy of each indicator in DN;the indicator with a large area under the ROC curve was selected for the model.The area under the ROC curve was used to evaluate the diagnostic efficacy of each evaluation model in DN to obtain the best predictive model.And determine the best reference diagnosis point.The model was compared with the ability of UAER to diagnose DN to further verify the repeatability and effectiveness of the model in DN diagnosis.Results The model(NT-ProBNP+CysC)had the largest area under the ROC curve,which was 0.861,the sensitivity was 0.826,and the specificity was 0.784.The best cut-off point for the predictive value of this model for DN was 204.36.It had better predictive performance and improved the predictive performance of laboratory indicators for elderly DN.Conclusion The establishment of a mathematical evaluation model can significantly improve the predictive performance of various indicators in judging whether to have DN.The evaluation model is NT-ProBNP+CysC,which has better predictive performance in DN≥60 years old.