Strengths and Limitations
The major strengths of this study were its large sample size, long
follow-up period, and its population-based sample, which was
representative of Taiwanese residents. The diagnosis of AD in our study
was given by board-certified pediatricians or dermatologists, and
therefore recall bias and detection bias was likely minimal. In
addition, we were able to adjust for some covariates, including
gestational infections and maternal acetaminophen use during pregnancy,
which were not addressed in previous studies. Our analysis of the effect
of prenatal antibiotic exposure on AD diagnosed after 1 year of age and
its relationship with infant antibiotic/acetaminophen use may have
partly taken confounding by indication (increased infection or fever
that led to increases in both AD and antibiotic/acetaminophen exposure)
into account in our study, which has not been mentioned previously.
The major limitation of our study is that some covariates, which may
influence the risk of AD due to maternal antibiotic use during
pregnancy, such as maternal smoking, patients’ consulting behavior or
healthcare utilization patterns, dietary habits, environment exposures
(e.g., climate conditions, air pollution, etc.) and some unmeasured
factors shared by families,2,24,32 were not adjusted
for in our analysis. It is also possible that mild cases of AD who did
not seek medical help were not identified in our analysis. However, the
Taiwan NHI is an easily accessible, low-cost medical insurance system.
Finally, we were not able to analyze the effect of prenatal antibiotics
on AD by types of antibiotic use, which may have influenced the risk in
patients with asthma.10,11 However, to our knowledge,
this effect did not exist when investigating the influence of antibiotic
exposure during pregnancy on offspring AD.9,20,24,33
In conclusion, this cohort study revealed that maternal antibiotic use
during pregnancy was associated with a slightly increased risk of
childhood AD. This association was similar among trimesters during
pregnancy and an apparent dose-response pattern was observed, although
the association may have been partly confounded by postnatal
acetaminophen use in AD diagnosed after 1 year of age. Further research
should be conducted to investigate this confounder in a prospectively
designed study and to explore the timing before or after pregnancy in
order to elucidate whether or not this association was specific to
during pregnancy. Nevertheless, we believe that antibiotics should be
used prudently during pregnancy and that this potential effect should be
taken into account.