Development and Internal Validation of a Clinical Risk Score for Hypovitaminosis D in Italian Adults Aged ≥50 Years Attending Osteoporosis and Metabolic Bone Disease Centers

Abstract

Background/Objectives: Hypovitaminosis-D is a highly prevalent condition worldwide, associated with adverse skeletal and extra-skeletal outcomes. Increasing demand for serum 25-hydroxyvitamin D (25(OH)D) testing points toward the need for simple tools to identify individuals at risk and optimize laboratory use. We aimed to develop and validate a clinical risk score for predicting hypovitaminosis-D based on easily assessable risk factors. Methods: This cross-sectional study included 1408 adults aged ≥50 years (1286 women and 122 men) attending centers across Italy dedicated to osteoporosis and metabolic bone diseases. Demographic, clinical, lifestyle, and dietary data were collected through a standardized questionnaire. Overall, 1147 subjects (81.5%) were receiving cholecalciferol supplementation. Univariable and multivariable logistic regression analyses identified predictors of 25(OH)D < 20 ng/mL (primary outcome) and <30 ng/mL. Risk scores were derived from models and internally validated. Discriminative ability was assessed using ROC curves, and calibration was conducted by comparing predicted and observed probabilities. Results: The median age was 67 years, and 91.3% were female. Median 25(OH)D was 33.2 ng/mL; 9.8% had levels < 20 ng/mL. Independent predictors of hypovitaminosis-D (25(OH)D < 20 ng/mL) included higher body mass index, residence in Northern Italy, reduced summer sun exposure, sunscreen use, cardiovascular disease, glucocorticoid use, absence of cholecalciferol supplementation, and no prior vitamin D use. The score (range 9–18) showed good discrimination (Area under the curve; AUC: 79.1%, 95% CI: 75.3–82.9) and calibration (r = 0.98, p < 0.001). A screening cut-off (10.3–10.7) ensured high sensitivity (87.0–92.7%), while 11.9–12.0 balanced sensitivity (~62%) and specificity (~80%). A second score for 25(OH)D < 30 ng/mL showed moderate discrimination (AUC: 69.6%). Performance remained stable across seasons and in untreated subjects (AUC: 72.7%). Female predominance, widespread vitamin D use, and the absence of external validation or impact analyses limit generalizability. Conclusions: The proposed data-driven clinical risk score may help identify individuals at risk of vitamin D deficiency and support targeted screening strategies, potentially reducing unnecessary testing in routine clinical practice.