Modelling and prediction of 25-hydroxyvitamin D levels in Norwegian relapsing-remitting multiple sclerosis patients.

BACKGROUND/AIM 25-Hydroxyvitamin D (25(OH)D) levels are suggested to influence the susceptibility and risk of disease progression in multiple sclerosis (MS). Seasonal fluctuation of 25(OH)D levels may differ in magnitude between individuals. The purpose of this paper was to model the seasonal fluctuation of vitamin D in Norwegian MS patients and to examine to which extent one single 25(OH)D measurement predicts the level at other time points throughout the year. METHODS During December 2004 and July 2008, 762 serum samples were collected from 92 Norwegian relapsing-remitting MS patients. Time series analysis and multivariate modelling techniques were used to model seasonal fluctuations and intra- and inter-individual variations in 25(OH)D values. RESULTS Most patients reached their lowest 25(OH)D level in March/April and the highest in July/August. There were substantial intra-individual variations in the extent of seasonal fluctuation, with 36.6% of explainable variation in seasonally adjusted 25(OH)D levels (on a logarithmic scale) attributable to the patient level. The remaining 63.4% could be accounted for by sources of inter-individual variation. Both the total and inter-individual variabilities were lowest in February, and the prediction interval in this month was up to 26% narrower compared to other months. The prediction intervals would be at least 21% wider with only one observation available per patient. CONCLUSIONS The seasonal fluctuations of 25(OH)D levels in Norwegian relapsing-remitting MS patients are subject to pronounced intra- and inter-individual variation. The most representative measurements of 25(OH)D levels are taken in February.