Methods
Study design and data collection
A schematic illustration of study design is presented in Figure 1. Ethical approval was obtained from Centralized Institutional Review Board (CIRB) of SingHealth (reference 2009/280/D) and Domain Specific Review Board (DSRB) of Singapore National Healthcare Group (reference D/09/021). The Growing Up in Singapore Towards healthy Outcomes (GUSTO) cohort is a mother-offspring birth cohort where only Singapore citizens or permanent residents of Chinese, Malay and Indian ethnicity with homogenous ethnic background were approached for participation 13. A total of 1247 healthy pregnant women, aged 18 years and above, were recruited during their first trimester (<14 weeks’ gestation) at two major public maternity hospitals in Singapore. Interviewer-administered questionnaires were used at 26-28 weeks of gestation to collect data on a) demographic/socioeconomic characteristics (ethnicity, mother’s educational status, monthly household income, child’s gender), and b) history of prenatal tobacco smoke exposure (active and/or passive). Information on current feeding practices was periodically collected at 3-month intervals using a separate interviewer-administered questionnaire provided to the mothers and cumulatively used to derive “duration of total breastfeeding”. Additionally, child’s frequencies of daily toothbrushing and sweet snack intake were recorded using a separate oral health questionnaire provided to primary caregivers at the 24-month dental visit.
Skin and oral examination
Skin examinations for AD diagnosis was performed by physicians/clinical residents at the clinic visits using Hanifin and Rajka criteria14. Of the children diagnosed with AD, the disease severity was assessed using the SCORAD (SCORing Atopic Dermatitis) index15. Trained clinical residents performed the SCORAD assessments in children at 18- and 36-month clinic visits. If a child had SCORAD scores at both the 18- and 36-month clinic visits, the higher of the two SCORAD scores was used in the analysis. Additionally, caregiver-reported history of doctor-diagnosed AD in the first year and findings from skin prick test at 18 months were recorded as described previously 8.
Dental examinations in children were carried out at 3-year clinic visit by three dentists, who were trained and calibrated to standardize ECC scoring. Examinations were conducted with the knee-to-knee position, using plane surface mouth mirrors aided by tactile inspection, when deemed necessary. Caries detection was performed using modified International Caries Detection and Assessment System (ICDAS) diagnostic criteria 16, with ICDAS code 1 not recorded due to logistical constraints. No additional detection methods or radiographs were used. Inter- and intra-examiner reliability were assessed during the training phase and quantified using the Intraclass Correlation Coefficient (ICC).
Statistical analyses
Statistical analyses were performed using STATA (Version 12). Continuous variables are presented as mean (standard deviation) and median (interquartile range), while categorical variables are presented as N (%). The outcome of interest was number of decayed, missing, or filled surfaces (dmfs ). The SCORAD criteria was used to categorize AD severity into: i) moderate-to-severe AD (SCORAD≥25) and ii) mild AD (SCORAD<25) cases. A third group comprising children without AD diagnosis at both clinic visits was used as the reference group (non-AD). Comparison of caries rates across the three groups was done using Kruskal-Wallis analysis, followed by Mann-Whitney test with Bonferroni correction. The distribution of caries was skewed with overdispersion, making negative binomial regression the appropriate statistical technique for multivariable analysis. The estimates were exponentiated to obtain adjusted incidence risk ratios (aIRR) to estimate caries risk in 3-year old children. Potential confounding factors, including ethnicity, maternal education, household income, child’s gender, and prenatal tobacco smoke exposure (active/passive) were adjusted for in the analysis. Additionally, robustness of association was further assessed by controlling for postnatal diet (such as duration of breastfeeding, child’s daily frequency of sweet snack intake) and oral hygiene factors (such as child’s daily frequency of toothbrushing).