Statistical analysis
Participant characteristics are described separately for mothers who received 1000 IU cholecalciferol and those who received placebo using frequency and percentage distribution for categorical variables, mean (SD) for normally distributed continuous variables and median (interquartile range) for non-normally distributed continuous variables.
We used logistic regression to examine associations between being randomised to the active group and developing eczema at ages 12, 24 and 48 months, expressing results as odds ratios (ORs) and 95% confidence intervals (CIs). Models were adjusted for duration of breastfeeding as descriptive analyses showed differences in breastfeeding duration between the groups. Sensitivity analyses were undertaken to examine whether the effect of the intervention differed in mothers breastfeeding for <1 and ≥1 month because of its reported association with eczema and the influence of gestational supplementation on breast milk vitamin D content, and to take account of season of birth. We additionally used logistic regression to examine the associations between the offspring’s risk of atopic eczema at ages 12, 24 and 48 months and maternal late pregnancy serum 25(OH)D concentration and single-nucleotide polymorphisms (SNPs) in or near key vitamin D metabolism genes, specifically rs12785878 (encoding 7-dehydrocholesterol reductase (DHCR7) in the epidermal vitamin D biosynthesis pathway), rs10741657 and rs6013897 (encoding 25-hydroxylase (CYP2R1) and 24-hydroxylase (CYP24A1) , respectively).10 Analyses were performed using Stata (Stata version 15.1, Statacorp LP, TX).