Comparison with other studies
Our results are consistent with a cross-sectional study by McKeever et al., who used a birth cohort of 24690 children derived from the General Practice Research Database in the United Kingdom, and found a positive association between prenatal antibiotic use and doctor-diagnosed eczema in a dose-related manner.20 In line with our results, their study also revealed that infections during pregnancy increased the risk of allergic diseases and adjusting for them did not notably affect the estimates. Dom et al. conducted a prospective birth cohort study of 773 children in Belgium. A questionnaire was used to investigate parent-reported eczema, and the results revealed that prenatal antibiotic exposure was strongly positively associated with eczema in children up to 4 years old after adjusting for post-natal antibiotics.21 Conversely, two Danish cohort studies that used a prospective design were conducted by Stensballe et al. and the results showed that prenatal antibiotic use did not influence the risk of eczema in 411 and 30675 children, respectively.22 However, their follow-up period was shorter and did not take exposure in all trimesters into account. In addition, a study by Sasaki et al. who used a Japanese nationwide dataset to conduct a prospective birth cohort study showed no association between prenatal antibiotic exposure and diagnosed or observed AD at 1 year old in 70408 children (aOR 1.01, 95% CI 0.97-1.06).23 Similar to the study by Stensballe et al., their follow-up period was also shorter and excluded children born through cesarean delivery, which may influence the risk of prenatal antibiotic exposure in AD.
A recent prospective cohort study performed by Mubanga et al. investigated 722767 singleton children and 74663 discordant siblings in Sweden. In line with our results, they found the risk of AD was greater in children exposed to prenatal antibiotics than in those who were not exposed (aHR 1.10, 95% CI 1.09-1.12), whereas no such increased risk was found in the sibling-control analysis, suggesting some familial confounders.24 Based on the covariates they adjusted for, we further adjusted the maternal comorbidities, maternal AD, maternal allergic rhinitis, gestational infections, and maternal acetaminophen use during pregnancy, which may be potential risk factors according to previous studies.12,25,26 In addition, their study revealed an apparent dose-response relationship, and the category of infection did not modify the association of maternal antibiotic use and child AD, which was consistent with our results. In terms of the trimester-specific association between antibiotic exposure during pregnancy and childhood AD, their results were comparable with our results in that the risk increased in all trimesters; however, our data showed a slightly higher risk in the first and second trimesters compared with the third trimester, which might be explained by antibiotics-related dysbiosis or immunomodulatory effects, phenomena that are known to have a greater impact during early pregnancy.
Several studies and meta-analyses revealed that post-natal antibiotics and acetaminophen exposure were associated with an increased risk of AD, especially within the first year of life.27-29Antibiotics and acetaminophen are frequently used to treat infection. However, reverse causation must be taken into consideration, so we performed a subgroup analysis by independently examining the effects of these two medications on AD. After excluding AD diagnosed before 1 year of age, the positive association was still present in the prenatal antibiotic exposure group regardless of postnatal antibiotic use, but the adjusted hazards reduced to null in children who were not exposed to postnatal acetaminophen, suggesting that some confounders common to both prenatal antibiotics and postnatal acetaminophen use existed (e.g., infection, fever, or AD itself). However, the majority (87%) of AD patients in our cohort were diagnosed before 1 year of age, so this negative association could only represent the minority (13%) of AD patients diagnosed after 1 year of age. Our results are consistent with a study by Dom et al. that found adjusting for post-natal antibiotics did not significantly change the result showing prenatal antibiotic use increased the risk of childhood AD.21
Even though our results showed a positive association between maternal use of antibiotics during pregnancy and childhood AD, this does not necessarily imply a causal link between the two. A registry-based cohort study conducted by Stokholm et al. investigated the temporal association between maternal antibiotic use and childhood asthma and showed that the risk of asthma was similar among 80 weeks before pregnancy, during pregnancy, and 80 weeks after pregnancy.30 The authors suspected that maternal antibiotic use is a surrogate marker of maternal propensity for infection, which is the true cause of childhood asthma. Loewen et al. replicated their results in a retrospective cohort study and proposed that some unmeasured factors, such as healthcare utilization patterns, shared familial factors, and maternal deficiency of vitamin D or other immunomodulatory nutrients may explain both the increased risk of maternal antibiotic use/infection and the increased risk of asthma in offspring.31 While we could not address the temporal association between maternal antibiotic use before and after pregnancy, our results were unchanged after adjusting for three common infections during pregnancy. Furthermore, maternal antibiotic use may also be a surrogate marker of infant antibiotic use, which is associated with asthma and AD.31 Our data showed that infants born to mothers who received antibiotics during pregnancy were indeed more likely to be exposed to postnatal antibiotics; however, the positive association between maternal antibiotic use during pregnancy and childhood AD remained significant in the subgroup analysis independent of infant antibiotic use.