Variability in vitamin D assays impairs clinical assessment of vitamin D status

Background: Measuring serum 25(OH)D concentration is common in clinical practice despite the questionable reliability of assays. Aims: The aim of the present study was to examine agreement in 25(OH)D concentrations measured by different assays and laboratories, and consider related clinical implications. Methods: Serum samples from 813 participants in the Australian Multicentre Study of Environment and Immune Function (the Ausimmune Study) were assayed for 25(OH)D concentration. Duplicate samples from subsets of subjects were sent to different laboratories, two using DiaSorin Liaison (Laboratory A and B) and one using Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS - selected here as the nominal gold standard). Pairwise within-assay (both within-laboratory and between-laboratories) and between-assay agreement was examined using Deming regression and Bland-Altman plots. Common 25(OH)D cut-points for classification of vitamin D deficiency were used to compare the different assays. Results: 25(OH)D concentrations measured using Liaison were substantially lower at Laboratory A than at Laboratory B (mean bias -11.60 nmol/L, 95% limits of agreement -46.39, 23.18). Both Liaison assays returned much lower 25(OH)D concentrations than LC-MS/MS (mean bias up to -26.05 nmol/L, 95% limits of agreement of -13.21, 65.31). For Laboratory A participants, 46% (355/765) were classified as vitamin D deficient (25(OH)D <50 nmol/L) using Liaison compared with 17% (128/765) using LC-MS/MS. For Laboratory B participants, the respective figures were 36% (76/209) and 20% (41/209). Hence, between 1-in-5 and 1-in-3 participants were misclassified as 'deficient'. Conclusion: Bias and variability in 25(OH)D measurements sufficient to affect significantly clinical decision-making were found both between-laboratories and between assays. The adoption of common standards to allow assay calibration is required urgently.