Redefining vitamin D status: Establishing population-based indirect reference intervals through big data analysis.

Objectives: To establish accurate population-based reference intervals (RIs) for serum 25-hydroxyvitamin D [25(OH)D] using the refineR indirect method and real-world data (RWD), accounting for demographic, methodological, and seasonal factors.

Methods: A retrospective analysis of 130,030 serum 25(OH)D samples collected from 2018 to 2022 at a tertiary hospital in Barcelona was performed. Samples were measured using VDSP-certified Liaison and Atellica immunoassays. The refineR algorithm was employed to establish RIs, utilizing Box-Cox transformations to adjust for data distribution. Demographic variables (sex and age), assay differences, medical department and seasonal variations were analysed. RIs were verified using a subset of healthy individuals.

Results: The median serum 25(OH)D level was 25 ng/mL (62.5 nmol/L). Vitamin D deficiency (≤20 ng/mL, ≤50 nmol/L) was observed in 34.2% of samples, and severe deficiency (≤12 ng/mL, ≤30 nmol/L) in 12.6%. The default Box-Cox transformation estimated RIs of 11.5-64.5 ng/mL (28.7-161.2 nmol/L), while the modified Box-Cox transformation yielded RIs of 14.2-65.9 ng/mL (35.5-164.7 nmol/L). Women exhibited wider RIs (14.5-68.6 ng/mL, 36.2-171.5 nmol/L) compared to men (11.6-57.3 ng/mL, 29-143.2 nmol/L). Method-specific RIs were 10.2-58.6 ng/mL (25.5-146.5 nmol/L) for the Liaison assay and 9.9-59.3 ng/mL (24.7-133.2 nmol/L) for the Atellica assay. The lowest RIs were observed in outpatients (4.3-46.4 ng/mL, 10.7-116 nmol/L) and endocrinology patients (5.5-43.9 ng/mL, 13.7-109.7 nmol/L). Seasonal variation significantly impacted RIs, with higher levels during summer months.

Conclusions: The refineR algorithm effectively established population-based RIs for serum 25(OH)D in Barcelona, revealing significant demographic and seasonal variations. Redefining 25-hydroxyvitamin D thresholds based on population-specific data may reduce unnecessary screening and supplementation, minimizing associated risks. This study highlights the need for population-, seasonal-, and method-specific RIs to improve vitamin D assessment and management.