Construction of a nomogram for predicting serum vitamin D deficiency in children/adolescents with new-onset type 1 diabetes: A single-center study in China

Yang Xin^1,2Jiang Hui^1,2Liao Min^1,2Lin Meng^1,2*Wu Jin^1,3

• ¹Key Laboratory of Birth Defects and Related Diseases of Women and Children, West China Second University Hospital, Sichuan University, Chengdu, Sichuan Province, China
• ²Department of Pediatric Genetics, Metabolism and Endocrinology Nursing, West China Second University Hospital, Sichuan University, Cheng Du, China
• ³Department of Pediatric Genetic Metabolism and Endocrinology, West China Second University Hospital, Sichuan University, Chengdu, Sichuan Province, China

Construction of a nomogram for predicting serum vitamin D deficiency in children/adolescents with new-onset type 1 diabetes: A single-center study in China Objective: The vitamin D-type 1 diabetes (T1D) association has been debated in public health. The purpose of this study was to develop a vitamin D deficiency prediction model and investigate vitamin D deficiency risk factors in children and adolescents with new-onset T1D. Methods: A single-centre, retrospective analysis of paediatric patients (1-18 years) with new-onset T1D and initial 25-hydroxyvitamin D assessments was performed at a tertiary hospital in China between January 2020 and July 2024 (n=353). The patients were divided into two groups according to whether their vitamin D deficiency exceeded 12 ng/mL. After identifying vitamin D deficiency risk factors in children/adolescents with new-onset T1D, a receiver operating characteristic (ROC) curve model was developed to predict the probability of vitamin D deficiency in these individuals. That model was represented with a nomogram. Calibration and clinical decision analysis curves were used to evaluate the model's effectiveness after internal validation via bootstrapping. Results: The prevalence rate of serum vitamin D deficiency among patients with new-onset T1D was 26.35% (93/353). Multivariate logistic regression analysis revealed that minority status (X1), weight (X2), diabetic ketoacidosis severity (X3), serum vitamin D testing season (X4), free triiodothyronine (X5), and high-density lipoprotein (X6) were closely associated with serum vitamin D deficiency development in children/adolescents with new-onset T1D (P<0.05). The model was logit (P) =e x /(1+ e x ), X=4.626-1.878*X1-0.038*X2-0.821*X3-0.88*X4 + 0.351*X5 + 0.532*X6. The area under the curve (AUC) of the serum vitamin D deficiency predictive model among patients with new-onset T1D was 0.769 (95% CI=0.711-0.826). The predicted probability's best cut-off value was 0.671. Conclusions: The established risk prediction model has good efficacy, providing a reference for screening high-risk vitamin D deficiency groups among children/adolescents with new-onset T1D and taking preventive and protective measures. The nomogram was developed based on a single-center cohort in China, and its generalizability needs further validation in more extensive populations.