Diabetes, Metabolic Syndrome and Obesity (Jan 2024)
Construction of a Nomogram-Based Prediction Model for the Risk of Diabetic Kidney Disease in T2DM
Abstract
Xian Wang,1 Xiaming Liu,1 Jun Zhao,1 Manyu Chen,1 Lidong Wang2 1Graduate School of Chengde Medical College, Chengde, Hebei, People’s Republic of China; 2Department of Endocrinology and Immunology, Chengde Central Hospital Affiliated to Chengde Medical College, Chengde, Hebei, People’s Republic of ChinaCorrespondence: Lidong Wang, Department of Endocrinology and Immunology, Chengde Central Hospital Affiliated to Chengde Medical College, Chengde, Hebei, People’s Republic of China, Tel +86 13703142117, Email [email protected]: To investigate the predictors of diabetic kidney disease (DKD) in type 2 diabetes mellitus (T2DM) patients and establish a nomogram model for predicting the risk of DKD.Methods: The clinical data of T2DM patients, admitted to the Endocrinology Department of Chengde Central Hospital from October 2019 to September 2020 and divided into a case group or a control group based on whether they had DKD, were collected. The predictive factors of DKD were screened by univariate and multivariate analysis, and a nomogram prediction model was constructed for the risk of DKD in T2DM. Bootstrapping was used for model validation, receiver operating characteristic (ROC) curve and GiViTI calibration curve were used for evaluating the discrimination and calibration of prediction model, and decision analysis curve (DCA) was used for evaluating the practicality of model.Results: Predictors for DKD are diabetic retinopathy (DR), hypertension, history of gout, smoking history, using insulin, elevation of body mass index (BMI), triglyceride (TG), cystatin C (Cys-C), and reduction of 25 (OH) D. The nomogram prediction model based on the above nine predictors had good representativeness (Bootstrap method: precision: 0.866, Kappa: 0.334), differentiation [the area under curve (AUC) value: 0.868], and accuracy (GiViTI-corrected curved bands, P = 0.836); the DAC curve analysis showed that the prediction model, whose threshold probability was in the range of 0.10 to 0.70, had clinical practical value.Conclusion: The risk of DKD in T2DM could be predicted accurately by DR, hypertension, history of gout, smoking history, using insulin, elevation of BMI, TG, Cys-C, and reduction of 25 (OH) D.Keywords: type 2 diabetes mellitus, diabetic kidney disease, predictive model, 25-hydroxyvitamin D, diabetic retinopathy
