Construction of a Predictive Model for the Development of Diabetic Nephropathy in Type 2 Diabetes Based on Clinical Data and Laboratory Indicators and Validation of Predictive Value
Objective To construct a predictive model for diabetic nephropathy (DN)in type 2 diabetes mellitus (T2DM)based on clinical data and laboratory indicators,and to validate the predictive value to provide relevant reference for clinical practice.Methods A total of 267 T2DM patients in our hospital from January, 2019 to March,2021 were selected as the research objects.According to the"Expert Consensus on the Prevention and Treatment of Diabetic Nephropathy (2014 Edition)",the DN diagnostic criteria were divided into simple T2DM group (197 cases)and DN group (70 cases).We collected two groups of clinical data and laboratory indicators,then constructed a logistic regression model of DN in T2DM,evaluated the predictive value of the model,and validated the individual value prediction.Results Univariate analysis showed that age, gender,BMI,atherosclerosis,coronary heart disease,smoking history,drinking history,family history of diabetes,FBG,2hPG,Hb,TG,Scr,and BUN were not influencing factors of DN in T2DM (P >0.05);The course of T2DM,hypertension,HbA1c,TC,HDL-C,LDL-C,and SUA were identified as influencing factors of DN in T2DM (P<0.05 );Logistic regression analysis showed that the course of T2DM, hypertension,HbA1c,TC,LDL-C and SUA were independent risk factors for DN in T2DM,while HDL-C was an independent protective factor for DN in T2DM (P<0.05);ROC curve and logistic regression model were constructed according to the above independent factors to predict the occurrence of DN in T2DM.The best cut-off value of Log(P)was 0.489,while AUC was 0.830,with 95％CI of 0.779-0.873,the sensitivity of 61.43％,and the specificity of 89.34％,which were all higher than the predictive values of each individual influencing factor.A total of 178 T2DM patients from April,2021 to January,2022 were randomly selected as the validation set,including 42 patients with DN and 136 without DN.AUC of the model for predicting DN in T2DM in the validation set was 0.922,with 95％ CI of 0.872-0.957,the sensitivity of 85.71％,and the specificity of 87.50％.Conclusion We develop a prediction model for DN in T2DM based on the duration of T2DM,hypertension,HbA1c,TC,LDL-C,HDL-C,and SUA,which has reliable predictive values and can be an important tool to predict DN risk clinically.
Type 2 diabetes mellitusDiabetic nephropathyLogistic regression analysisInfluencing factorsPredictive value
万方数据
维普
