Abstract
Seasonal variation in serum concentration of the vitamin D metabolite 25(OH) vitamin D [25(OH)D], which contributes to host immune function, has been hypothesized to be the underlying source of observed influenza seasonality in temperate regions. The objective of this study was to determine whether observed 25(OH)D levels could be used to simulate observed influenza infection rates. Data of mean and variance in 25(OH)D serum levels by month were obtained from the Health Professionals Follow-up Study and used to parameterize an individual-based model of influenza transmission dynamics in two regions of the United States. Simulations were compared with observed daily influenza excess mortality data. Best-fitting simulations could reproduce the observed seasonal cycle of influenza; however, these best-fit simulations were shown to be highly sensitive to stochastic processes within the model and were unable consistently to reproduce observed seasonal patterns. In this respect the simulations with the vitamin D forced model were inferior to similar modeling efforts using absolute humidity and the school calendar as seasonal forcing variables. These model results indicate it is unlikely that seasonal variations in vitamin D levels principally determine the seasonality of influenza in temperate regions.