Keywords
antibiotic prophylaxis - cesarean delivery - asthma - childhood obesity - microbiota
- neonatal microbiome
The prevalence of cesarean delivery in the United States has been increasing since
the 1990s.[1] Surgical site infections occur after approximately 5 to 12% of cesarean deliveries
and are among the most common complications after the procedure.[2]
[3]
[4]
[5] There have been many well-designed studies that have assessed a variety of antibiotic
prophylaxis regimens designed to reduce this complication. For example, a meta-analysis
of three randomized clinical trials demonstrated that antibiotic prophylaxis given
prior to skin incision rather than after umbilical cord clamping significantly reduced
the incidence of surgical site infections.[2] Recently, a randomized clinical trial of azithromycin added to the usual antibiotic
prophylaxis regimen prior to nonelective cesarean deliveries demonstrated a significant
reduction in infectious morbidity from 12 to 6.1%.[6] Another meta-analysis of 16 trials demonstrated that women who had vaginal preparation
prior to unscheduled cesarean delivery also had a significant reduction in endometritis
compared with women who did not receive vaginal preparation.[7] However, a subsequent secondary analysis of the adjunctive azithromycin trial demonstrated
that vaginal preparation did not make a difference in the incidence of surgical site
infections.[8] Finally, Harper et al performed a cost analysis and demonstrated that adding azithromycin
to the usual antibiotic prophylaxis for both elective and nonelective cesarean deliveries
would be cost saving.[9] However, among these critical works, little attention is paid to the long-term effects
of antibiotic prophylaxis exposure on the fetus. Costantine et al's meta-analysis
of three randomized trials demonstrated no difference in neonatal sepsis or neonatal
intensive care unit admissions, but did not include any longer neonatal follow-up.[2] The randomized trial of the adjunctive azithromycin trial also had short neonatal
follow-up that also showed no differences in readmissions at 3 months of age between
the groups.[6] We believe the failure to consider neonatal consequences represents an important
knowledge gap.
Our argument is not that the hypothetical risks to neonates of antibiotic exposure
should, a priori, outweigh the demonstrated benefits of antibiotics to women. Rather,
we believe that nascent evidence pointing to potential risks to children from in utero
exposure to antibiotics—described later—deserves the attention of researchers and
academicians before they expose ever increasing numbers of children to these agents.
Other disciplines of medicine have also raised concerns regarding the long-term effect
of antibiotics on neonatal microbiome and development of chronic diseases such as
obesity.[10]
[11] Evidence of these concerns have recently been expressed by several professional
organizations. For example, the U.S. Preventive Services Task Force updated their
guideline of routine urine culture testing among pregnant women from a Grade A to
a Grade B recommendation based on recent understanding of the influence of antibiotics
on the microbiome.[12] In addition, the updated guideline on group B streptococcal (GBS) infections by
the American Academy of Pediatrics highlighted the risks of disruption of the infant
microbiome from prenatal antibiotic exposure and noted that the secondary effects
of this intervention are unknown and an area of active investigation.[13] Finally, one of the reasons for the postdelivery timing of intervention in the ANODE
trial was concern about antibiotics on the infant microbiome.[14] Thus, investigators should recognize the potential for risk, and include plans for
follow-up of children, and should explore ways to mitigate possible adverse consequences.
Up until now potential, and potentially serious, risks have been largely ignored in
the design and interpretation of studies, and in their translation into treatment
guidelines. Cost/benefit or risk/benefit analyses that underpin recommendations of
therapeutic regimens may not be as useful if neonatal consequences are not part of
the calculus.
Evidence
Culture-based studies, which have shown the presence of microbes in the placenta,
amniotic fluid, fetal membrane, umbilical cord blood, and meconium, suggest that neonatal
microbiome development may begin at birth.[15] One published hypothesis is that the maternal gut microbiota may be transferred
to the fetus via the bloodstream.[16] Labeled bacterial species that were given to pregnant mice orally were found in
the meconium of their offspring.[16] Another exposure to microbes in the newborn is during the delivery, either by vaginal
flora after vaginal delivery or by common skin flora after cesarean delivery.[15] What is important for this discussion is that the neonatal microbiome can be disrupted
by antibiotic exposure. Cox et al performed a series of experiments using low-dose
penicillin that was given to pregnant mice and their offspring at different times
(e.g., pregnant mice were given low-dose penicillin right before birth) and the microbiota
was tested in the offspring. They found that antibiotics changed the neonatal microbiome
from that seen among offspring that were not exposed to antibiotics and, importantly,
the microbiota was altered even with short exposure to antibiotics, that is, not continued
exposure throughout early life.[17] Several studies have looked at the effect of intrapartum antibiotics on the neonatal
microbiome. Corvaglia et al collected fecal samples of 84 infants on days 7 and 30
of life and found that there was significant difference in certain colonies among
infants born from GBS-positive women who had received intrapartum antibiotics (N = 35) compared with infants born from GBS-negative women who did not receive intrapartum
antibiotics (N = 49) at day 7, but the difference diminished by day 30.[18] The limitation of this study was that they only evaluated for three bacteria that
may play an important role in the development of the microbiome. Another work, evaluating
52 newborns by fecal sampling on day 7, had a similar finding, but did not repeat
fecal sampling at a later age.[19] Finally, Mazzola et al evaluated the entire microbiome diversity on fecal samples
using whole genome sequencing and polymerase chain reaction from 26 infants on days
7 and 30 of life. They found that infants who had exposure to intrapartum antibiotics
in GBS-positive women had significantly lower diversity of bacteria compared with
GBS-negative women on day 7 with only partial recovery on day 30.[20] Whether the difference in diversity of the neonatal microbiome persists is currently
unknown.[21]
Several other consequences of antibiotic exposure have been studied. Ahmadizar et
al, in a meta-analysis, demonstrated an increased risk of asthma with antibiotic use
in early life. They analyzed two large population-based cohorts from the Netherlands
and Scotland, including 7,393 and 891 children, respectively, and found an increased
risk of asthma with an odds ratio of 2.18 in children who had antibiotic exposure
in the first 3 years of life.[22] Another recent retrospective cohort revealed an association of antibiotic exposure
in the first year of life and the development of asthma. Yoshida et al used health
insurance claim data in Japan and found that antibiotic exposure during pregnancy
was also associated with asthma development (hazards ratio: 1.18, 95% confidence interval
[CI]: 1.08–1.30), though this association was only seen up to 3 years of age.[23] In addition, a case–control study of 134 children with asthma demonstrated that
antibiotic exposure during pregnancy was a significant risk factor for development
of asthma (adjusted odds ratio: 3.19, 95% CI: 1.52–6.67).[24]
The association of eczema and prenatal antibiotic exposure has also been explored.
Dom et al retrospectively analyzed a prospective cohort study of pregnant women who
completed a questionnaire on antibiotic use during pregnancy, and assessed the subsequent
occurrence of allergies, asthma, and eczema in their children up to 4 years of age.
In the 773 children analyzed, they found an association of prenatal antibiotic exposure
with development of eczema (adjusted odds ratio: 1.82, 95% CI: 1.14–2.92).[25] Sariachvili et al analyzed 976 children from the same cohort to determine if breastfeeding
had a protective effect on development of eczema within the first year of life, but
they did not find a significant association. They did demonstrate an increased risk
of eczema after antibiotic use in pregnancy (adjusted odds ratio: 1.8, 95% CI: 1.2–2.7).[26] However, a recent meta-analysis on the development of eczema and prenatal antibiotic
exposure that included four observational studies found no significant difference
in the risk of developing eczema with prenatal antibiotic exposure (odds ratio: 1.30,
95% CI: 0.86–1.95).[27]
There is also literature on the relationship of early antibiotic exposure and development
of childhood obesity. Using longitudinal data, Mueller et al demonstrated an increased
risk of childhood obesity at 7 years after prenatal exposure to antibiotics in the
third trimester. They analyzed 436 mother–child dyads after collecting data on prenatal
antibiotic use by using a questionnaire that was administered in the third trimester
and analyzing data on actual body weight and measurement of the child at age 7. They
found an increased risk of obesity associated with prenatal antibiotic exposure with
an adjusted odds ratio: 1.77 (95% CI: 1.25–2.51). They also found an association with
cesarean delivery (adjusted odds ratio: 1.41, 95% CI: 1.01–1.96).[28] It has been suggested that intestinal microbiota plays an essential role in these
findings, and differences in the microbiome may lead to obesity. However, it is important
to note that although there are statistical significance among the stated studies,
the CIs are narrow and close to one, and may signal noise rather than harm.[29] Nonetheless, it is imperative for us to be aware and prospectively assess and confirm
this association. Antibiotic exposure is one of three key factors that may influence
the neonate's microbiota, the other two being breastfeeding and mode of delivery.[15] Currently, there are no direct studies evaluating the effect of intrapartum antibiotic
exposure on childhood obesity; however, further information about this link should
be forthcoming from ongoing research. For example, a group in Canada is currently
recruiting low-risk women from a midwife practice, planning for vaginal birth after
37 weeks, with a planned study population of 240. They will prospectively follow this
cohort and obtain stool samples up to 3 years of age to describe the intestinal microbiome
of infants who were breastfed, and to determine if infants born to women who receive
intrapartum antibiotics for GBS or antibiotic prophylaxis have intestinal microbiota
at the first year of life that differs significantly in type from those not exposed
to antibiotics.[30] This issue is concerning not only because of the direct medical harms from obesity
but also because of the greater health care costs that will be incurred. One group
of authors estimated that $190 billion per year of health care spending is due to
treating obesity and obesity-related conditions.[31]
Future Considerations and Solutions
It must be noted that these studies assessed exposures that were usually much greater
in magnitude than a single dose of antibiotics around the time of birth, and the risks
associated with the lesser doses may be de minimis. On the contrary, there is some
worrisome data associated with GBS prophylaxis, and it has been shown that it does
not take prolonged exposure to antibiotics to alter the microbiome. If the studies
on the microbiome, as well as the cited work on asthma and obesity, are borne out,
then future investigations of antibiotics in pregnancy will need to weigh the benefits
to the mother against potential harms to the child. This would be particularly important
when the maternal benefits are marginal. A hypothetical example would be a study that
showed that the number needed to treat to prevent one case of endometritis was 100.
While that one woman might be spared a prolonged course of antibiotics, and rarely,
a more serious consequence, then it would be reasonable to ask, what biologic cost
is borne by the 100 exposed neonates? To address this conundrum, we suggest two possible
solutions. First, future studies on antibiotic prophylaxis should look at long-term
effects on the neonates, until at least 3 years of age, to evaluate for outcomes such
as obesity and asthma. In one study, bacterial species were characterized from fecal
samples over the first year of life in 110 healthy children and were found to be similar
to adult populations by the time the children reached the age of 3.[32] Thus, a straightforward outcome to consider will be the body mass index of children
at the age of 3. However, this outcome (as well as others such as asthma and eczema)
may have confounding variables that were not addressed in the previous studies, including
social determinant factors, such as type of housing and food security. It is important
that these variables are accounted for when evaluating these outcomes in future studies.
Though these studies will be difficult and logistically challenging, they are essential,
if we are to be able to consider the development of chronic diseases when crafting
recommendations for antibiotic use in pregnancy.
Second, there should be more studies looking at antibiotic prophylaxis after cord
clamping to see if they can be made to have an efficacy similar to that seen with
antibiotics given before clamping. A recent single site cohort study analyzed this
issue, and found no difference in surgical site infections after cesarean sections
after a policy change from antibiotic prophylaxis administered after cord clamping
to preincision.[33] In addition, Valent et al demonstrated in a randomized controlled trial that 48 hours
of cephalexin and metronidazole in addition to the usual practice of preincision cefazolin
administration reduced surgical site infections from 15 to 6% in obese women, which
was a similar risk reduction from 12 to 6.1% seen in the trial by Tita et al.[6]
[34] If antibiotic prophylaxis after cord clamping in conjunction with postoperative
antibiotic regimen were as effective as preincision antibiotic prophylaxis, this would
dramatically reduce the exposure of antibiotics to the fetus and reduce their long-term
effect.
In conclusion, we are not arguing that appropriate antibiotic prophylaxis should no
longer be a standard of care in obstetrics. However, there is a need for investigators
to recognize the current evidence based on observational studies. Therefore, the long-term
effects of antibiotic exposure in the fetus/neonate (e.g., potential risk of asthma
and childhood obesity related to changes in neonatal microbiota) should be considered
when planning large-scale trials that will involve antibiotic prophylaxis.
