Methods
Study population
The COCOA (COhort for Childhood Origin of Asthma and Allergic diseases) comprised the general Korean population and aimed to investigate the causal contribution of genetics, perinatal environment, maternal lifestyle, and psychosocial stress of the mother and child on pediatric susceptibility to allergic diseases.(14) (See “study population” in the Online Repository)
Of the 2,846 enrolled infants, 488 were lost to follow-up; children with incomplete questionnaire data about AD and antibiotic exposure until 5 years of age were excluded. The remaining 1,637 children were included in this study.
DNA Collection and SNP Genotyping
Genomic DNA was extracted from the cord blood mononuclear cells of each child and genotyped for IL-13 (rs20541) polymorphisms using the TagMan assay (ABI, Foster City, CA, USA). The endpoint fluorescent readings were measured on an ABI 7900HT Sequence Detection System (ABI, Foster City, CA, USA). The details of the methods are described in a previous study.(15) Duplicate samples and negative controls were included to ensure genotyping accuracy.
Physician’s assessment of allergic diseases and antibiotic exposure
The presence of AD was clinically diagnosed by pediatric allergists on the basis of the Hanifin and Rajka criteria.(16) Presence of AD was defined as physician-diagnosed AD in the preceding 12 months during each follow-up annually, and the incidence of AD was defined as the number of new cases of physician-diagnosed AD that developed since birth over a defined period. At consecutive visits, AD patients were assessed by pediatric allergists using the SCORing Atopic Dermatitis (SCORAD). Higher numbers indicate greater severity, and the scale ranges from 0 to 103.(17) AD severity was categorized as mild (<15) and moderate to severe (≥15) according to the objective components of the index (clinical signs and disease extent).(18)
Early antibiotic treatment was defined as exposure to antibiotics for more than 3 days within the first 6 months of life, regardless of whether the infant was hospitalized or not, and pediatric allergists obtained the information about the antibiotics at each visit after birth.
Outcomes
Our primary outcome of interest was AD incidence at the age of 6 months and 1 to 3 years according to the frequency of antibiotic exposure within 6 months as assessed by pediatric allergists. Other primary outcomes were the influence of antibiotic exposure on AD phenotypes and investigator-reported clinical signs (SCORAD).
The secondary outcomes were AD incidence at the age of 6 months and 1 to 3 years according to IL-13 genetic variations. Other secondary outcomes were combined effects of IL-13 genetic variations and antibiotic exposure on the AD incidence and phenotypes. Two transition periods were considered, namely, age from 6 months to 2 years and from 2 to 5 years, to classify AD phenotypes. We defined four different phenotypes of AD: the early-transient phenotype, with AD onset within 2-years of age and no further symptoms later; the early-persistent phenotype, with onset within 2 years of age and symptoms do not improve within less than 2 years; the late phenotype, with onset after 2 years of age; and the non-AD.(19)
Statistical analysis
Data are presented as frequencies and proportions for categorical variables. Chi-square test and Fisher’s exact test were performed to evaluate the associations between the incidence of AD at the age of 1 year and the variables. To assess effect modification according toIL-13 polymorphisms and early-life antibiotic exposure, subjects were divided into 4 groups according to environmental factors (antibiotic use within 6 months for more than 3 days) and genetic background (genotypes) and multivariable logistic regression models were performed with the data to estimate adjusted odd ratios (aOR) and the corresponding 95% confidence intervals (95% CI) for a comparison of AD occurrence risk by relevant covariates after adjusting for potential confounders: sex, maternal education level, family history of allergic diseases, history of breastfeeding, and mode of delivery. Multinomial logistic regression analysis was also used to identify the effects of the use of antibiotics within 6 months of age and the combined effect of the use of antibiotics and IL-13 genetic variations on AD phenotypes. Finally, we tested for trends regarding the effect of the risk factors (early-life environmental factors and genetic polymorphisms) on the development of AD by two-factor analysis of variance (two-way ANOVA). All statistical tests were two-sided, and significance levels of p values were set at <0.05. Statistical analyses were performed using SPSS Statistics, version 23.0 (IBM SPSS Statistics, Inc., Chicago, IL).