Analytical Bias in the Measurement of Plasma 25-Hydroxyvitamin D Concentrations in Infants

Hypovitaminosis D is prevalent worldwide; however, analytical bias in the measurement of circulating 25-hydroxyvitamin D (25(OH)D) concentrations may affect clinical treatment decisions and research. We performed parallel plasma 25(OH)D analyses using the Abbott Architect i2000 chemiluminescent immunoassay (CIA) and liquid chromatography-tandem mass spectrometry (LC-MS/MS) for paired samples from the same infants at 3 (n = 69), 6 (n = 79) and 12 months (n = 73) of age. To test agreement, we used Lin's concordance correlation coefficient and corresponding 95% confidence interval, Bland-Altman's limits of agreement, and Bradley-Blackwood (BB) test. Agreement was high at 3 months (coefficient between difference and mean -0.076; BB F = 0.825; p = 0.440), good at 12 months (-0.25; BB F = 2.41; p = 0.097) but missing at 6 months of age (-0.39; BB F = 12.30; p < 0.001). Overall, 18 infants had disparate results based on the cut-off point for vitamin D deficiency (25(OH)D < 50 nmol/L), particularly at three months, with seven (10%) infants deficient according to CIA but not LC-MS/MS, and four (6%) deficient by LC-MS/MS but not CIA. To our knowledge, this is the first study to show that the reported 25(OH)D concentration may be influenced by both age and assay type. Physicians and researchers should be aware of these pitfalls when measuring circulating 25(OH)D concentrations in infants and when developing treatment plans based on measured vitamin D status.