Background: Current literature examining associations between vitamin D and chronic disease generally use a single assessment of 25-hydroxyvitamin D [25(OH)D], assuming the 25(OH)D concentration of an individual is consistent over time.
Methods: We investigated the intraindividual variability between two measures of plasma 25(OH)D concentrations collected approximately five years apart (1997-2000 to 2002-2005) in 672 postmenopausal women participating in the Women's Health Initiative. Plasma 25(OH)D was assessed using the DiaSorin LIAISON® chemiluminescence immunoassay. The within-pair coefficient of variation (CV) was 4.9% using blinded quality control samples. Mean and SDs of 25(OH)D at the two time points were compared using a paired t test. An intraindividual CV and intraclass correlation coefficient (ICC) were used to assess intraindividual variability. A Spearman correlation coefficient (r) assessed the strength of the association between the two measures, and concordance in vitamin D status at two time points was compared.
Results: Mean 25(OH)D concentrations (nmol/L) significantly increased over time from 60.0 (SD = 22.2) to 67.8 (SD = 22.2; P < 0.05). The CV was 24.6%, the ICC [95% confidence interval (CI)] was 0.59 (0.54-0.64), and the Spearman r was 0.61 (95% CI = 0.56-0.66). Greater concordance over five years was observed in participants with sufficient compared with deficient or inadequate baseline 25(OH)D concentrations (weighted kappa = 0.39). Reliability measures were moderately influenced by season of blood draw and vitamin D supplement use.
Conclusion: There is moderate intraindividual variation in 25(OH)D concentrations over approximately five years.
Impact: These data support the use of a one-time measure of blood 25(OH)D in prospective studies with ≤ five years of follow-up.