“…25(OH)D concentration and season of blood sampling were entered into a multivariable logistic regression model along with age, sex, NZDep, tobacco smoke exposure, breastfeeding duration and ethnicity which were identified as potential confounding variables. A second multivariable logistic regression model that included risk factors previously identified for COME (prolonged labour, cow's milk exposure before 13 months, older siblings, siblings with tympanostomy tubes, frequent colds, day care hours per week, frequent snoring and frequent nasal obstruction) was also entered into a multivariable logistic regression model with age, sex, NZDep, ethnicity, tobacco smoke exposure, breastfeeding duration, season of blood sampling and serum 25(OH)D concentration. A p value of 0.05 or less was considered to indicate statistical significance.…”