Construction and validation of a risk prediction model for vitamin D deficiency in patients with type 2 diabetes mellitus.

OBJECTIVES: Patients with type 2 diabetes mellitus (T2DM) have a high prevalence of vitamin D deficiency, but convenient and efficient screening tools are lacking in clinical practice. This study aimed to construct and validate a predictive model for vitamin D deficiency risk in T2DM patients based on routine clinical indicators. METHODS: Clinical data were retrospectively collected from 618 T2DM patients hospitalized in the Department of Endocrinology of a tertiary general hospital between January 2024 and December 2024. Patients were randomly divided into a training cohort (n = 432) and a validation cohort (n = 186) at a ratio of 7:3. LASSO regression was used to screen predictors, and a logistic regression model was constructed to generate a nomogram. The discrimination, calibration, and clinical utility of the model were evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA), respectively. RESULTS: The prevalence of vitamin D deficiency in T2DM patients was 71.8%. LASSO regression combined with multivariate logistic regression showed that female (OR = 3.53, 95%CI: 1.99-6.25, P < 0.001), elevated triglycerides (TG) (OR = 1.56, 95%CI: 1.15-2.12, P = 0.004), elevated glycated hemoglobin (HbA1c) (OR = 1.17, 95%CI: 1.03-1.32, P = 0.048), and urinary albumin-to-creatinine ratio (UACR) ≥ 300 mg/g (OR = 9.68, 95%CI: 2.58-29.24, P < 0.001) were independent risk factors for vitamin D deficiency, whereas age ≥ 65 years (OR = 0.34, 95%CI: 0.19-0.59, P < 0.001) was a potential protective factor. The nomogram model based on these five variables achieved an AUC of 0.7468 (95%CI: 0.6994-0.7942) in the training cohort and 0.7557 (95%CI: 0.6753-0.8362) in the validation cohort. Calibration curves revealed favorable consistency between predicted and actual probabilities (Hosmer-Lemeshow test: training cohort P = 0.436, validation cohort P = 0.672). DCA indicated net benefit within the clinically defined threshold range. CONCLUSION: We developed a nomogram using Gender, Age, TG, HbA1c, and UACR for predicting vitamin D deficiency risk in T2DM inpatients. Internal validation indicated promising performance. While the model may assist in identifying high-risk individuals in hospital settings, its clinical utility and generalizability remain to be confirmed through external validation.