This large population-wide study of routine primary and secondary health-care records did not find evidence of an increased risk of asthma and eczema (including severe asthma and eczema, and other allergic and allergy-related conditions) in children aged < 5 years in relation to the change in the guidelines recommending administration of antibiotics before skin incision in CSs.
There is also no convincing evidence that pre-incision antibiotics are associated with significant risk of development of autoimmune and other conditions investigated in this study, including infections, necrotising enterocolitis, leukaemia, neurodevelopmental conditions, colic and failure to thrive. Most of these outcomes in our study were purely exploratory, as there was insufficient evidence regarding the role of microbiota in the development of these conditions. Many of these health conditions are rare in young children, resulting in large uncertainty around the estimates. Our study was not sufficiently powered to detect clinically important differences in the rarer outcomes.
We found a statistically significant reduction in neonatal sepsis similar to the relative reduction in the point estimate found in a systematic review of randomised controlled trials, although the review’s results were not statistically significant.89 The results confirm previous findings89 that antibiotics before skin incision do not increase the risk of neonatal sepsis, but the observed relative reduction in the risk associated with pre-incision antibiotics in babies born by CS, which was estimated by controlling for the risk of sepsis in babies born by VD, should be interpreted with caution in this study. The overall incidence of neonatal sepsis diagnoses recorded in routine health-care data has increased over the years. Clinically diagnosed neonatal sepsis often has no organism isolated from blood culture.90 The diagnosis of neonatal sepsis in our study is likely to be culture-confirmed sepsis, in which case the receipt of antibiotics immediately prior to birth may result in an increase in the likelihood of being culture negative, even if the baby develops signs of neonatal sepsis (i.e. it may be an artefact of the antibiotics rather than an actual decrease in the relative risk of disease).
The very small (ranging from 1% to 3%) relative increases in the number of consultations recorded in primary care, the increased likelihood of children exposed to pre-incision antibiotics being prescribed antibiotics in early childhood, the increased likelihood of hospital admission for gastroenteritis and ever being admitted to hospital for any reason that where observed in this study are all common events and, therefore, produce statistically significant differences. Residual confounding and bias may explain these differences; however, even if they are real, they are very small and unlikely to be clinically important.
In this research, we obtained risk estimates for > 50 outcomes across primary and secondary care data sets; however, because of multiple comparisons, some of the statistically significant differences we observed may have occurred by chance. Some of the health conditions, such as ADHD, may show symptoms in early childhood, but are most commonly diagnosed later in life, with the incidence being the highest in school-aged children.91 Therefore, the numbers of children born by CS with a new diagnosis of ADHD in primary care were very small in each year during our study, resulting in large variations in the incidence between the years. More evidence would be required to confirm or refute this finding before any conclusion of an association can be made. Previous research has shown conflicting results regarding exposure to antibiotics in early life and ADHD. For example, recent population-based Canadian and Danish studies92,93 did not observe a significant association between antibiotic exposure in the first years of life and ADHD, particularly in sibling-controlled analysis that adjusted for familial confounding factors. However, a large Finnish study94 suggested an association between exposure to antibiotics in early life and ADHD that may not be fully explained by familial confounding.
Our study did not identify a significant association between necrotising enterocolitis and pre-incision antibiotics for CS, including co-amoxiclav. There was some degree of uncertainty around the estimates, and they do not exclude the possibility of an increased risk. There was a slight increase in the overall incidence of necrotising enterocolitis over the study period, as recorded in the routine secondary care records. In this study, we assessed the overall association only and did not examine the risk in any specific subgroups of pregnancies and babies previously reported to be at higher risk of necrotising enterocolitis.83,95
The relative reduction in the composite outcome of maternal infectious morbidity in the post-partum period associated with pre-incision antibiotics in CS that we observed in this study is comparable to the reduced risks previously reported in randomised controlled trials.9,89 We found an overall increase in maternal sepsis recorded in health-care records, which may be due to an increase in diagnosis, prevention and treatment of sepsis over the study period. The relative increase in the CS group may, therefore, be an artefact, as any operation increases the suspicion of sepsis, leading to an increased rate of diagnosis. Evidence from randomised controlled trials shows that prophylactic antibiotics significantly reduce serious infectious complications compared with no antibiotic prophylaxis for CS.96
In this study, we investigated the impact of the timing of prophylactic antibiotics in CS deliveries. Babies born by CS have a different composition of gut microbiota after birth and are more likely to be colonised by opportunistic pathogenic bacteria than babies born by VD.13 Therefore, the impact of antibiotic prophylaxis around the time of birth in babies born by VD may be different.
The strengths of this study include using data from large population-based mother–baby linked data sets and examining the impact of pre-incision antibiotic policy on a wide range of allergy, immune system and other related outcomes. It also assessed the impact on the severity of outcomes by examining the risk of health conditions recorded in primary care and health conditions resulting in a hospital admission. The study employed a strong design, using VD as a comparison group to control for any temporal changes (e.g. changes in diagnoses and recording in medical records) over time, which could not be achieved by simple adjustment for confounding factors. We also undertook a number of sensitivity analyses to assess the robustness of our findings, including the impact of missing data, and the results were similar to those of the main analyses.
The study also has several limitations that need to be taken into account when interpreting the findings. This was an observational study that used routinely collected health-care data, and validated case definitions for different health conditions in electronic medical records, where such exist, have mostly been derived from populations that include older age groups.51,52,97 Making diagnoses for some conditions in early childhood can be difficult and rely on symptoms (e.g. asthma objective tests can be performed in children aged ≥ 5 years only).44,98 Some health conditions we investigated in this study were very rare in young children and our study was not sufficiently powered to detect clinically significant differences in risks for these outcomes. The timing of prophylactic antibiotic administration was not recorded in routine health-care records and, therefore, we were not able to ascertain exposure to pre-incision antibiotics at an individual level. It is also possible that there is residual confounding that was not fully accounted for by using VD as a comparator group. Finally, the results may not be generalisable to settings where different prophylactic antibiotic regimens are used.
Implications for practice and research recommendations
The overall findings from this study provide further evidence for the currently prevailing practice in the UK of giving prophylactic antibiotics before skin incision to help reduce the risk of maternal infections in the post-partum period.
Some of the secondary health outcomes we investigated in this study were exploratory because of a lack of or inconclusive evidence regarding an association between newborn microbiota, early exposure to antibiotics and children’s development. Many of the health conditions were also very rare in young children, resulting in large uncertainty around the estimates. We suggest undertaking a further study using routinely collected health-care data in other countries where there was no change in the timing of prophylactic antibiotic administration to assess if there have also been changes in the risk over time of developing these health conditions, particularly ADHD and gastroenteritis, irrespective of the exposure to prophylactic antibiotics around the time of birth. The current study includes health-care records until 2018, but it could also be repeated in the future to capture outcomes of interest in children aged > 5 years.
Other research is ongoing in the UK using the Born in Bradford birth cohort, allowing the ascertainment of antibiotic exposure at an individual level. The original Born in Bradford cohort included 13,776 children, and the majority of these children were born before the recommendation to give antibiotics pre incision.99 This may limit the ability of this study to assess the risk for most of the outcomes of interest investigated in this research.