Statistical analysis
The chi-squared and Fisher’s exact tests were performed to compare categorical variables of smoke exposure and cotinine levels between children with and without AD. Multivariate logistic regression analysis was used after adjusting for potential confounding factors to calculate adjusted odds ratios (ORs) and 95% confidence intervals (CIs). The confounding factors were maternal education level, sex of the children, the type of milk given during the first 6 months of life, SHS exposure during the first year of life, parental history of allergic diseases, the presence of a pet during the first year of life, daycare attendance during the first year of life, and mode of delivery. To analyze the effect of smoke exposure after birth, children in the school age (7–9 years) group were further adjusted for SHS exposure during their 4th to 6th year of age. The paramed command in STATA version 16.1 (STATA Corp. College Station Texas, USA) was used to perform causal mediation analysis using parametric regression models with SHS as exposure, AD as the outcome, and IgE level as mediators. All statistical analyses were performed using SPSS version 21.0 (SPSS Inc., Chicago, IL, USA), and a p-value of less than 0.05 was considered statistically significant.