Intended for healthcare professionals

  1. Research
  2. Association between...
  3. Association between exposure to antibiotics during pregnancy or early infancy and risk of autism spectrum disorder, intellectual disorder, language disorder, and epilepsy in children: population based cohort study
CCBYNC Open access
Research

Association between exposure to antibiotics during pregnancy or early infancy and risk of autism spectrum disorder, intellectual disorder, language disorder, and epilepsy in children: population based cohort study

BMJ 2024; 385 doi: https://doi.org/10.1136/bmj-2023-076885 (Published 22 May 2024) Cite this as: BMJ 2024;385:e076885
  1. Ahhyung Choi, postdoctoral research fellow1 2,
  2. Hyesung Lee, research professor1 3,
  3. Han Eol Jeong, postdoctoral research fellow1 3,
  4. Seo-Young Lee, professor4 5,
  5. Jun Soo Kwon, professor6 7 8,
  6. Jung Yeol Han, professor9,
  7. Young June Choe, associate professor10,
  8. Ju-Young Shin, associate professor1 3 11
  1. 1School of Pharmacy, Sungkyunkwan University, Suwon, South Korea
  2. 2Harvard-MIT Center for Regulatory Science, Harvard Medical School, Boston, MA, USA
  3. 3Department of Biohealth Regulatory Science, Sungkyunkwan University, Suwon, South Korea
  4. 4Department of Neurology, College of Medicine, Kangwon National University, Chuncheon, South Korea
  5. 5Interdisciplinary Graduate Program in Medical Bigdata Convergence, Kangwon National University, Chuncheon, South Korea
  6. 6Department of Psychiatry, Seoul National University College of Medicine, Seoul, South Korea
  7. 7Department of Brain and Cognitive Sciences, Seoul National University College of Natural Sciences, Seoul, South Korea
  8. 8Institute of Human Behavioral Medicine, Seoul National University Medical Research Center, Seoul, South Korea
  9. 9Korean Mothersafe Counselling Center, Department of Obstetrics and Gynecology, Inje University Ilsan Paik Hospital, Goyang, South Korea
  10. 10Department of Pediatrics, Korea University Anam Hospital, Korea University College of Medicine, Seoul, South Korea
  11. 11Department of Clinical Research Design & Evaluation, Samsung Advanced Institute for Health Sciences and Technology, Sungkyunkwan University, Seoul, South Korea
  1. Correspondence to: J-Y Shin shin.jy{at}skku.edu
  • Accepted 11 April 2024

Abstract

Objective To evaluate the association between antibiotic use during pregnancy or early infancy and the risk of neurodevelopmental disorders in children.

Design Nationwide population based cohort study and sibling analysis.

Setting Korea’s National Health Insurance Service mother-child linked database, 2008-21.

Participants All children live born between 2009 and 2020, followed up until 2021 to compare those with and without antibiotic exposure during pregnancy or early infancy (first six months of life).

Main outcomes measures Autism spectrum disorder, intellectual disorder, language disorder, and epilepsy in children. After 1:1 propensity score matching based on many potential confounders, hazard ratios with 95% confidence interval were estimated using Cox proportional hazard models. A sibling analysis additionally accounted for unmeasured familial factors.

Results After propensity score matching, 1 961 744 children were identified for the pregnancy analysis and 1 609 774 children were identified for the early infancy analysis. Although antibiotic exposure during pregnancy was associated with increased risks of all four neurodevelopmental disorders in the overall cohort, these estimates were attenuated towards the null in the sibling analyses (hazard ratio for autism spectrum disorder 1.06, 95% confidence interval 1.01 to 1.12; intellectual disorder 1.00, 0.93 to 1.07; language disorder 1.05, 1.02 to 1.09; and epilepsy 1.03, 0.98 to 1.08). Likewise, no association was observed between antibiotic exposure during early infancy and autism spectrum disorder (hazard ratio 1.00, 0.96 to 1.03), intellectual disorder (1.07, 0.98 to 1.15), and language disorder (1.04, 1.00 to 1.08) in the sibling analyses; however, a small increased risk of epilepsy was observed (1.13, 1.09 to 1.18). The results generally remained consistent across several subgroup and sensitivity analyses, except for slightly elevated risks observed among children who used antibiotics during very early life and those who used antibiotics for more than 15 days.

Conclusions In this large cohort study, antibiotic exposure during pregnancy or early infancy was not associated with an increased risk of autism spectrum disorder, intellectual disorder, or language disorder in children. However, elevated risks were observed in several subgroups such as children using antibiotics during very early life and those with long term antibiotic use, which warrants attention and further investigation. Moreover, antibiotic use during infancy was modestly associated with epilepsy, even after control for indications and familial factors. When prescribing antibiotics to pregnant women and infants, clinicians should carefully balance the benefits of use against potential risks.

Introduction

Neurodevelopmental disorders are emerging as a critical public health problem among children, given the long lasting effect of these disorders on individuals’ lives and society.12 The global prevalence of neurodevelopmental disorders, including autism spectrum disorders, has been steadily increasing.345 Although the causes of neurodevelopmental disorders are not yet fully understood, several potential risk factors include advanced maternal age, preterm birth, and environmental factors.16 In recent years, a growing body of evidence is also highlighting that alterations in the microbiome may play a significant role in the development of neurodevelopmental disorders.67

Antibiotics, which are known to disturb the composition of the microbiome, are commonly used during pregnancy and infancy to treat infections.8910 Given that fetal and early life is the critical period for the extensive development of the gut microbiome, concern is growing about antibiotic use during these periods.11 Few epidemiological studies have investigated an association between prenatal or infant antibiotic use and neurodevelopmental disorders such as autism spectrum disorder and epilepsy.12131415 However, the evidence remains limited and inconclusive, possibly owing to insufficient control for confounding in some studies. Because infection, in both severe and less severe forms, has been linked with neurodevelopmental consequences, confounding by indication is of particular concern in investigating the role of antibiotics in neurodevelopmental disorders.161718 In addition, familial confounding presents another potential source of bias, as the pathophysiology of neurodevelopmental disorders involves both genetic and environmental factors. Therefore, comprehensive investigation on this topic based on real world data is warranted, while controlling for these potential confounding, as pregnant women and infants are generally excluded from randomised trials.

In this study, we aimed to evaluate whether exposure to antibiotics during pregnancy or early infancy is associated with subsequent development of autism spectrum disorder, intellectual disorder, language disorder, and epilepsy in children by using a large nationwide database in South Korea. Although intellectual disorder and language disorder are recognised as another common type of neurodevelopmental disorder, no study to date has thoroughly evaluated whether these disorders are associated with antibiotic use. To account for confounding by indication and unmeasured familial factors, we implemented two designs: a propensity score matched cohort study and a sibling analysis.

Methods

Data source and study design

We conducted a nationwide retrospective cohort study using the National Health Insurance Service (NHIS) mother-child linked database from 2008 to 2021.19 The NHIS database contains claims data of the entire population (>50 million) in South Korea, and construction of the mother-child linkage has been described previously.2021 This database comprises comprehensive information on socioeconomics, healthcare utilisation (for example, diagnosis, drug prescription, and medical procedures) from both inpatient and outpatient settings, health examination records for mothers and infants, and vital statistics data. We estimated the start of pregnancy on the basis of a previously validated algorithm developed using administrative data.22 This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.23

Study cohorts

We constructed two separate cohorts among all children born between 1 April 2009 and 31 December 2020 (fig 1). The first cohort (hereafter referred to as the pregnancy cohort), for analysing antibiotic exposure during pregnancy, consisted of all children after exclusion of children with chromosomal abnormalities and those whose mothers took antibiotics one month before the start of pregnancy but not during pregnancy to avoid classifying those who actually took their antibiotics or had available antibiotics after the start of pregnancy in the unexposed group. The second cohort (hereafter referred to as the infant cohort), for analysing antibiotic exposure during early infancy, comprised all children after exclusion of those who died during the first six months of life, children with chromosomal abnormalities, and those who received a diagnosis of one of the study outcomes during the first six months of life.

Fig 1
Fig 1

Flowchart of study cohort identification. NHIS=National Health Insurance Service

Antibiotic exposure

We defined exposure as the presence of one or more prescriptions for a systemic antibiotic (Anatomical Therapeutic Chemical classification system code J01). For the pregnancy cohort, we defined the exposed group as children whose mothers had at least one prescription for antibiotics at any time during pregnancy and the unexposed group as children whose mothers had no history of prescriptions for antibiotics from 30 days before pregnancy to the end of the pregnancy. For the infant cohort, we defined the exposed group as children who had at least one prescription for antibiotics during the first six months of life and the unexposed group as children with no history of prescriptions for antibiotics during this period.

Outcomes

Study outcomes were autism spectrum disorder (ICD-10 (international classification of diseases, 10th revision) code F84), intellectual disorder (F70-F73), language disorder (F80), and epilepsy (G40, G41, F803, R56, and prescription for antiepileptic medication) (supplementary table A). Autism spectrum disorder, intellectual disorder, and language disorder are typically diagnosed by psychiatrists or paediatricians in South Korea on the basis of clinical evaluation and standardised tests and recorded with ICD-10 codes. A previous study found a high positive predictive value of diagnostic codes in Korea’s claims data,24 and our definition of epilepsy has been validated with high positive predictive values especially among children (1-4 years old: 84.7%; 5-9 years old: 92.9%).25 We followed up all children from birth (pregnancy cohort) or six months after birth (infant cohort) until the occurrence of an outcome, death, or the end of the study period (31 December 2021), whichever came first.

Covariates

For the analyses of both pregnancy and infant cohorts, we considered a broad range of potential confounders including demographics (for example, maternal age, insurance type, and income level at delivery), indications for antibiotic use (for example, respiratory infection, urinary infection), infection related healthcare utilisation as a proxy for severity of infection (for example, number of hospital visits for infection diagnoses, number of distinct infection diagnoses), maternal conditions (for example, attention deficit/hyperactivity disorder, autoimmune diseases, diabetes mellitus), medication use (for example, paracetamol, antidepressants, antipsychotics), obstetric conditions (for example, nulliparity, multiple gestations), measures of healthcare utilisation (for example, obstetric comorbidity index),2627 smoking status, and body mass index. For the analyses of the infant cohort, we additionally considered maternal use of antibiotics, sex of child, preterm birth, caesarean section, birth weight, and type of feeding. We considered maternal antibiotic use during pregnancy as a covariate in the infant cohort as it may influence the infant’s exposure and the study outcomes, whereas we did not consider maternal antibiotic use during early infancy as a covariate in the pregnancy cohort as the infant’s exposure occurs after the prenatal exposure. Details of the covariates and their assessment windows are presented in supplementary table A.

Statistical analyses

To control for potential confounding, we did one-to-one propensity score matching using the greedy nearest neighbour matching algorithm without replacement.28 We estimated the propensity score, or the probability of antibiotic exposure in our study, by using a logistic regression model based on all pre-defined covariates except for smoking status and body mass index given their high proportion of missing data. We evaluated the distribution of covariates between exposed and unexposed groups by using the standardised mean difference, considering a value less than 0.1 on the absolute scale to indicate well balanced groups. In the propensity score matched cohort, we calculated the incidence of outcomes per 1000 person years and absolute rate differences with 95% confidence intervals on the basis of Poisson regression. We also estimated hazard ratios with 95% confidence intervals by using Cox proportional hazard regression models, while applying robust standard errors to account for data clustering. In each outcome model, we additionally adjusted for covariates that were unbalanced after propensity score matching. We considered a P value of <0.05 to be statistically significant. We used SAS Enterprise Guide, version 7.1, for data management and analysis.

Sibling analyses

Given the concerns about unmeasured confounding from familial factors, we additionally did sibling controlled analyses by restricting the population to children who had at least one sibling during the study period. Sibling analyses, by design, can control for time invariant shared factors at the family level. Using a stratified Cox proportional hazard regression model, we calculated hazard ratios with 95% confidence intervals for each outcome that were adjusted for all pre-defined covariates and birth order of sibling. Only siblings with discordant exposure and outcome status would contribute to the estimates; thus, informative pairs are reported. Additional details on the sibling analyses, including the assumption test for carryover effects, are available in supplementary appendix 1.29

Subgroup and sensitivity analyses

We did several subgroup analyses. Firstly, we analysed the risk of study outcomes by antibiotic groups (broad spectrum or narrow spectrum antibiotics; supplementary table B), given that broad spectrum antibiotics are reported to have a greater effect than narrow spectrum antibiotics on the gut microbiota. Secondly, we assessed whether the risk differed by specific timing of exposure (first, second, third trimester for pregnancy cohort; 0 to <2, 2 to <4, 4 to <6 months for infant cohort). Thirdly, as boys are more likely to receive a diagnosis of neurodevelopmental disorders than girls, we evaluated the association by sex of the infants.30 Fourthly, we analysed the duration-response relation and the risk of the top three most frequently prescribed antibiotic classes in each cohort. Lastly, we investigated the risk of study outcomes stratified by birth year (minimum follow-up period). As a post hoc analysis, we also estimated the risk of joint antibiotic exposure during pregnancy and early infancy.

We did multiple sensitivity analyses to test the robustness of our main findings. We accounted for potential exposure misclassification, outcome misclassification, and residual confounding, and we also evaluated whether our main findings from the propensity score matched cohorts were robust across different propensity scoring methods. Detailed information on subgroup and sensitivity analyses is given in supplementary appendices 2 and 3.

Patient and public involvement

The NHIS privacy policy and our institutional policy were not equipped to support patients or members of the public as partners. As study investigators, we honoured this policy. As a result, no patients were involved in this research although we are grateful for their data, which made the research possible.

Results

Cohort characteristics

We identified 3 665 246 children eligible for the pregnancy cohort, of whom 1 649 300 (45.0%) were exposed to antibiotics in utero. For the infant cohort, we identified 3 944 731 children, of whom 1 976 472 (50.1%) were exposed to antibiotics during early infancy (fig 1). After one-to-one propensity score matching, we identified 980 872 and 804 887 pairs for the pregnancy and infant cohorts, respectively. We observed substantial baseline differences between the exposed and unexposed groups (for example, indication) before propensity score matching, but all characteristics were well balanced (absolute standardised mean difference <0.1) after matching in both cohorts (table 1) All characteristics before and after propensity score matching are shown in supplementary tables C and D.

Table 1

Selected baseline characteristics of pregnancy and infant cohorts stratified by antibiotic exposure after propensity score matching. Values are numbers (percentages) unless stated otherwise

View this table:

Antibiotic exposure during pregnancy and early infancy: propensity score matched analyses

The median follow-up time for each outcome ranged around seven years in both pregnancy and infancy cohorts (supplementary table E). For the pregnancy cohort, the rate difference was 0.15 per 1000 person years for autism spectrum disorder (exposed versus unexposed: 1.25 v 1.10), 0.09 for intellectual disorder (0.62 v 0.53), 0.21 for language disorder (2.59 v 2.38), and 0.09 for epilepsy (1.12 v 1.03) (table 2). Exposure to antibiotics during pregnancy was associated with increased risks of all study outcomes, with adjusted hazard ratios ranging from 1.08 for epilepsy to 1.17 for intellectual disorder (fig 2, top).

Table 2

Incidence rate of autism spectrum disorder, intellectual disorder, language disorder, and epilepsy stratified by antibiotic exposure after propensity score matching

View this table:
Fig 2
Fig 2

Risk of autism spectrum disorder, intellectual disorder, language disorder, and epilepsy following exposure to antibiotics during pregnancy and early infancy. CI=confidence interval

We observed similar incidence rates and rate differences in the infant cohort (autism spectrum disorder: 0.05 (1.25 v 1.20); intellectual disorder: 0.13 (0.68 v 0.55); language disorder: 0.13 (2.71 v 2.58); epilepsy: 0.22 (1.01 v 0.80)) (table 2). Likewise, antibiotic exposure during early infancy was associated with increased risks of all study outcomes, with adjusted hazard ratios ranging from 1.04 for autism spectrum disorder to 1.27 for epilepsy (fig 2, top).

Antibiotic exposure during pregnancy and early infancy: sibling analyses

The sibling analyses comprised 843 412 and 1 082 417 exposure discordant children for the pregnancy and infant cohorts, respectively. Their baseline characteristics and informative exposure-outcome discordant pairs are shown in supplementary tables F and G. In contrast to the propensity score matched analyses, we found that all estimates were attenuated in the sibling analysis and showed no substantial associations (hazard ratio for autism spectrum disorder 1.06, 95% confidence interval 1.01 to 1.12; intellectual disorder 1.00, 0.93 to 1.07; language disorder 1.05, 1.02 to 1.09; epilepsy 1.03, 0.98 to 1.08) in the pregnancy cohort (fig 2, bottom). Similarly, in the infant cohort, we observed no associations (hazard ratio for autism spectrum disorder 1.00, 0.96 to 1.03; intellectual disorder 1.07, 0.98 to 1.15; language disorder 1.04, 1.00 to 1.08), except with epilepsy, which remained slightly increased (1.13, 1.09 to 1.18) (fig 2, bottom). Additional results from the sibling analyses are described in supplementary appendix 1 and tables H-K.

Subgroup and sensitivity analyses

We observed no notable differences in the association between antibiotic exposure during pregnancy or early infancy and all outcomes in the subgroup analyses, except for a slightly higher risk among infants who used antibiotics in the first two months of life and those who used antibiotics for longer than 15 days (fig 3; fig 4; supplementary tables L-S). In the post hoc analyses, we observed increased risks of autism spectrum disorder and epilepsy among children who were exposed to antibiotics during both pregnancy and early infancy (supplementary table T).

Fig 3
Fig 3

Subgroup analyses on risks of autism spectrum disorder, intellectual disorder, language disorder, and epilepsy following antibiotic exposure during pregnancy. CI=confidence interval. Estimates of forest plot are those from sibling cohort

Fig 4
Fig 4

Subgroup analyses on risks of autism spectrum disorder, intellectual disorder, language disorder, and epilepsy following antibiotic exposure during early infancy. CI=confidence interval. Estimates of forest plot are those from sibling cohort

Sensitivity analyses yielded estimates generally consistent with those from the main findings (supplementary tables U-X). Additional analyses to account for potential outcome misclassification also showed similar results (supplementary appendix 3 and table Y). The E value for the point estimate of epilepsy associated with antibiotic exposure during early infancy was 1.51.

Discussion

In this large nationwide cohort study, exposure to antibiotics during pregnancy or early infancy was not associated with increased risks of autism spectrum disorder, intellectual disorder, or language disorder in children. Although we observed small increased risks in the overall population, these associations were all attenuated and pointed towards the null in the sibling analysis, suggesting that the observed associations may have been confounded by shared familial factors. However, the risk of epilepsy associated with antibiotic use during infancy remained slightly elevated even after we accounted for the shared familial factors. Results from various subgroup and sensitivity analyses were largely consistent with our main findings, except for slightly increased risks observed among children who used antibiotics during very early life and those who used antibiotics for more than 15 days.

Comparison with other studies

A systematic review reported that available data on the association between prenatal or infant antibiotic exposure and autism spectrum disorder are conflicting and inconclusive.12 A later cohort study based on the Swedish population found an increased risk of autism spectrum disorder after use of antibiotics both during pregnancy and during infancy, with an odds ratio of 1.16 (95% confidence interval 1.09 to 1.23) and 1.46 (1.38 to 1.55), respectively.13 However, the authors noted that confounding by indication or genetics could not be ruled out. Likewise, a recent meta-analysis showed that an increased risk of autism (pooled odds ratio 1.13, 1.07 to 1.21) associated with antibiotic use in early life was no longer apparent when only the sibling matched studies were pooled (1.04, 0.97 to 1.11).31 In line with previous evidence, our study further supports a lack of association between antibiotic use during pregnancy or infancy and autism spectrum disorder after control for many confounders including indications and genetic/familial factors.

Our study did not observe an association between prenatal or infant antibiotic use and intellectual disorder or language disorder, which to our knowledge have not been investigated to date. Given that intellectual disorder and language disorder are increasingly recognised as another common type of neurodevelopmental disorder,32 further studies on the evaluation of these disorders are needed. Meanwhile, on the basis of the estimates observed in the comparison of the siblings, our study suggests no substantial association between prenatal or infant antibiotic use and intellectual disorder or language disorder.

By contrast, antibiotic use during infancy was modestly associated with epilepsy, even after control for familial factors in the sibling analysis, which indicates that familial confounding could not fully explain the observed increased risk. This increased risk was consistently observed across all subgroup analyses. Although several studies have investigated the risk of seizure or epilepsy in children associated with exposure to antibiotics during pregnancy, no previous studies have assessed this risk after antibiotic use in infants.141533 One hypothesised mechanism explaining the link between antibiotics and epilepsy is interference with gut microbiota, which can influence the interaction between the gut and the nervous system.34 In support, emerging evidence from case reports are indicating intestinal dysbiosis in patients with epilepsy.35 Because gut microbiota compositions are known to develop extensively during early life, antibiotic use during infancy may play a role in subsequent development of epilepsy.11 Another plausible mechanism is central nervous system toxicity of antibiotics. Several antibiotics have been suspected to provoke acute seizure, and animal studies have also shown that β-lactam associated seizures may arise from its γ-aminobutyric acid A receptor binding property.3637 Although the acute symptomatic seizures provoked by antibiotics do not necessarily result in chronic epilepsy, a potential negative effect on the developing brain may exist. Nevertheless, our study is the first epidemiological study to have investigated the association between antibiotics and epilepsy in infant populations and report increased risk; thus, additional studies are needed to confirm our findings. While awaiting future studies, the small but potential risk of epilepsy should be taken into account when weighing the benefits and risks of using antibiotics in infants.

Strengths and limitations of study

Antibiotics are increasingly used during pregnancy and infancy, with parallel concerns indicating that alterations in the microbiome may be associated with neurodevelopmental disorders; thus, our study tackles an important question. The major concern in investigating the potential role of antibiotics in neurodevelopmental outcomes is confounding by indication and familial factors. To minimise the possibility of these confounding, we included a wide range of covariates in the propensity score model, such as indications and related healthcare utilisation (for example, proxies for the severity of indication), and also implemented sibling designs to disentangle the effects of underlying infection and familial factors. Another strength of our study is the large sample size, which enabled us to have sufficient statistical power and also to do various clinically relevant subgroup analyses. In addition, by using the nationwide longitudinal database, adequate follow-up was possible to identify the risk of neurodevelopmental disorders along with no risk of selection or recall bias.

This study also has several limitations. Firstly, exposure misclassification is possible as we defined exposure on the basis of prescription of an antibiotic not the actual use. Secondly, outcome misclassification is possible. Although the algorithm for defining epilepsy in our study has been validated with high positive predictive values in young children,25 other outcomes have not been validated. Thus, we did a sensitivity analysis in which we redefined the outcomes by incorporating data from Korean Developmental Screening Test for Infants and Children,38 a validated tool to detect neurodevelopmental disorders in Korea, and the results were consistent with those of our main analyses (supplementary appendix 3). Thirdly, although we considered a broad range of covariates, we cannot rule out the possibility of residual confounding. For instance, paternal characteristics (for example, paternal age and comorbidities), which may be associated with the development of neurodevelopmental disorders, were unavailable in our database. The data on maternal smoking status and body mass index were also incomplete and thus not included in the propensity score model or adjustment. Fourthly, the indicators that we included in our study as proxies for the severity of infection are not direct measures of severity. The inclusion of the number of inpatient visits or emergency department visits with infection diagnoses and the number of distinct infection diagnoses was based on the assumption that the inpatient and emergency department visits typically necessitate a higher level of medical care (for example, close monitoring and intensive treatment) and that multiple infections suggest a greater level of severity compared with a single infection. However, they do not provide a direct assessment of the severity itself; thus, future studies incorporating direct measures of severity would be valuable. Fifthly, although sibling analyses are useful in controlling for familial factors that are not easily captured in the administrative database, time variant confounders are still of concern. To tackle this concern, we additionally adjusted for time variant characteristics; however, potential residual confounding from non-shared factors may still exist. Sibling comparison designs are also susceptible to carryover effects29; however, our analyses ruled out potential types of carryover effects. Lastly, as our study period included the early covid-19 era (2020-21), potential information bias on exposure, outcome, and covariate assessment due to delayed healthcare encounters may exist during this period.

Implications

In this study, the prevalence of antibiotic use during pregnancy and early infancy was 45.0% and 50.1%, respectively. Although antibiotics are essential for treating bacterial infections, concern is growing about their inappropriate use, which can contribute to antibiotic resistance.3940 Many countries, including South Korea and the UK, are actively working to reduce unnecessary prescription of antibiotics, especially for respiratory tract infections.414243 Despite the fact that many respiratory tract infections are viral and do not require antibiotics, studies have shown a substantial number of antibiotic prescriptions for such cases.4445 The challenge lies in the difficulty of distinguishing between bacterial and viral respiratory infections in primary care settings, a limitation also present in this study.46 Nevertheless, a considerable number of antibiotics were prescribed with respiratory infections as the primary diagnosis during pregnancy and infancy in our study. Furthermore, our additional analysis indicates that 117 572 pregnancies and 742 823 infants were treated with antibiotics for 15 days or longer for respiratory tract infections, of which 21 118 pregnancies and 167 404 infants had no other diagnoses of infection. Of note, the recommended treatment durations for most respiratory tract infections are below 15 days.44 Overall, clinicians should adhere to the recommended guidelines when prescribing antibiotics to pregnant women and infants to improve antibiotic stewardship.

Our study suggested no substantial association between prenatal or infant antibiotic exposure and autism spectrum disorder, intellectual disorder, and language disorder, but elevated estimates observed in several subgroups warrant further attention. Specifically, long term use of antibiotics (≥15 days) and antibiotic use during the first two months of life showed slightly increased risk even after control for familial factors. Moreover, in the post hoc analysis, antibiotic exposure both during pregnancy and during infancy was associated with autism spectrum disorder and epilepsy. Although residual confounding due to unfavourable health conditions cannot be completely ruled out, further investigation may be warranted. In the meantime, clinicians should carefully weigh the benefits and potential harms of antibiotics when prescribing antibiotics to pregnant women and infants.

Conclusions

In this large cohort study, exposure to antibiotics during pregnancy or early infancy was not associated with an increased risk of autism spectrum disorder, intellectual disorder, and language disorder in children. However, elevated risks were observed in several subgroups such as antibiotics use during very early life and long term antibiotic use, which warrants attention and further investigation. Moreover, antibiotic use during early infancy was modestly associated with epilepsy, even after control for indications and familial factors. When prescribing antibiotics to pregnant women and infants, clinicians should carefully balance the benefits of their use against potential risks.

What is already known on this topic

  • Antibiotics, which are known to disturb the microbiome composition, are commonly used during pregnancy and infancy to treat infections

  • Growing evidence indicates that early life microbiome disruption is associated with childhood neurodevelopmental disorders

What this study adds

  • The findings of this large cohort study suggest no association between maternal or infant antibiotic use and neurodevelopmental disorders

  • An exception was a modest association between exposure to antibiotics during infancy and epilepsy

  • Elevated risk observed in several subgroups such as antibiotic use during very early life and long term antibiotic use warrants attention and further investigation

Ethics statements

Ethical approval

The study protocol was approved by the institutional review board of Sungkyunkwan University; informed consent was waived as anonymised claims data were used (2022-04-011).

Data availability statement

No additional data available.

Footnotes

  • Contributors: AC and HL contributed equally to the paper as joint first authors. AC conceived and designed the study. AC and HL did statistical analyses and drafted the article. All authors interpreted the data, revised the article critically for important intellectual content, and gave final approval of the manuscript. JYS is the guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

  • Funding: This work was supported by a grant (21153MFDS607) from the Ministry of Food and Drug Safety of South Korea in 2021-25 and by the National Research Foundation of Korea grant funded by the Korea government (No RS-2023-00208978). AC was supported by the Health Fellowship Foundation (2022). The funders had no role in the study design; the collection, analysis, and interpretation of data; the writing of the report; or the decision to submit the article for publication.

  • Competing interests: All authors have completed the ICMJE uniform disclosure form at https://www.icmje.org/disclosure-of-interest/ and declare: this work was supported by the Ministry of Food and Drug Safety of South Korea and by the National Research Foundation of Korea grant funded by the Korea government; JYS received grants from the Ministry of Food and Drug Safety, the National Research Foundation of Korea, the Ministry of Health and Welfare, and pharmaceutical companies including Pfizer, UCB, and LG chem, outside the submitted work; no other relationships or activities that could appear to have influenced the submitted work.

  • Transparency: AC, HL, and JYS affirm that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

  • Dissemination to participants and related patient and public communities: The study results will be disseminated to the public through press releases, social media, and presentations at conferences.

  • Provenance and peer review: Not commissioned; externally peer reviewed.

http://creativecommons.org/licenses/by-nc/4.0/

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

References

  1. ↵
    America’s Children and the Environment.3rd ed. U.S. Environmental Protection Agency, 2015.
  2. ↵
    1. Arora NK,
    2. Nair MKC,
    3. Gulati S,
    4. et al
    . Neurodevelopmental disorders in children aged 2-9 years: Population-based burden estimates across five regions in India. PLoS Med2018;15:e1002615. doi:10.1371/journal.pmed.1002615 pmid:30040859
    OpenUrlCrossRefPubMed
  3. ↵
    1. Rah SS,
    2. Hong SB,
    3. Yoon JY
    . Prevalence and incidence of developmental disorders in Korea: a nationwide population-based study. J Autism Dev Disord2020;50:4504-11. doi:10.1007/s10803-020-04444-0 pmid:32347466
    OpenUrlCrossRefPubMed
  4. ↵
    1. Zeidan J,
    2. Fombonne E,
    3. Scorah J,
    4. et al
    . Global prevalence of autism: A systematic review update. Autism Res2022;15:778-90. doi:10.1002/aur.2696 pmid:35238171
    OpenUrlCrossRefPubMed
  5. ↵
    1. Lord C,
    2. Elsabbagh M,
    3. Baird G,
    4. Veenstra-Vanderweele J
    . Autism spectrum disorder. Lancet2018;392:508-20. doi:10.1016/S0140-6736(18)31129-2 pmid:30078460
    OpenUrlCrossRefPubMed
  6. ↵
    1. Carlsson T,
    2. Molander F,
    3. Taylor MJ,
    4. Jonsson U,
    5. Bölte S
    . Early environmental risk factors for neurodevelopmental disorders - a systematic review of twin and sibling studies. Dev Psychopathol2021;33:1448-95. doi:10.1017/S0954579420000620 pmid:32703331
    OpenUrlCrossRefPubMed
  7. ↵
    1. Cryan JF,
    2. Dinan TG
    . Mind-altering microorganisms: the impact of the gut microbiota on brain and behaviour. Nat Rev Neurosci2012;13:701-12. doi:10.1038/nrn3346 pmid:22968153
    OpenUrlCrossRefPubMed
  8. ↵
    1. Bookstaver PB,
    2. Bland CM,
    3. Griffin B,
    4. Stover KR,
    5. Eiland LS,
    6. McLaughlin M
    . A review of antibiotic use in pregnancy. Pharmacotherapy2015;35:1052-62. doi:10.1002/phar.1649 pmid:26598097
    OpenUrlCrossRefPubMed
  9. ↵
    1. Rogawski ET,
    2. Platts-Mills JA,
    3. Seidman JC,
    4. et al
    . Use of antibiotics in children younger than two years in eight countries: a prospective cohort study. Bull World Health Organ2017;95:49-61. doi:10.2471/BLT.16.176123 pmid:28053364
    OpenUrlCrossRefPubMed
  10. ↵
    1. McDonnell L,
    2. Gilkes A,
    3. Ashworth M,
    4. et al
    . Association between antibiotics and gut microbiome dysbiosis in children: systematic review and meta-analysis. Gut Microbes2021;13:1-18. doi:10.1080/19490976.2020.1870402 pmid:33651651
    OpenUrlCrossRefPubMed
  11. ↵
    1. Lozupone CA,
    2. Stombaugh JI,
    3. Gordon JI,
    4. Jansson JK,
    5. Knight R
    . Diversity, stability and resilience of the human gut microbiota. Nature2012;489:220-30. doi:10.1038/nature11550 pmid:22972295
    OpenUrlCrossRefPubMedWeb of Science
  12. ↵
    1. Łukasik J,
    2. Patro-Gołąb B,
    3. Horvath A,
    4. Baron R,
    5. Szajewska H,
    6. SAWANTI Working Group
    . Early life exposure to antibiotics and autism spectrum disorders: a systematic review. J Autism Dev Disord2019;49:3866-76. doi:10.1007/s10803-019-04093-y pmid:31175505
    OpenUrlCrossRefPubMed
  13. ↵
    1. Njotto LL,
    2. Simin J,
    3. Fornes R,
    4. et al
    . Maternal and early-life exposure to antibiotics and the risk of autism and Attention-Deficit Hyperactivity Disorder in childhood: A Swedish population-based cohort study. Drug Saf2023;46:467-78. doi:10.1007/s40264-023-01297-1 pmid:37087706
    OpenUrlCrossRefPubMed
  14. ↵
    1. Sassonker-Joseph N,
    2. Gorodischer R,
    3. Atar-Vardi M,
    4. Noyman I,
    5. Novack L
    . Prenatal exposure to antibiotics and development of epilepsy in children. J Clin Pharmacol2021;61:18-24. doi:10.1002/jcph.1674 pmid:32578224
    OpenUrlCrossRefPubMed
  15. ↵
    1. Meeraus WH,
    2. Petersen I,
    3. Gilbert R
    . Association between antibiotic prescribing in pregnancy and cerebral palsy or epilepsy in children born at term: a cohort study using the health improvement network. PLoS One2015;10:e0122034. doi:10.1371/journal.pone.0122034 pmid:25807115
    OpenUrlCrossRefPubMed
  16. ↵
    1. Köhler-Forsberg O,
    2. Petersen L,
    3. Gasse C,
    4. et al
    . A nationwide study in Denmark of the association between treated infections and the subsequent risk of treated mental disorders in children and adolescents. JAMA Psychiatry2019;76:271-9. doi:10.1001/jamapsychiatry.2018.3428 pmid:30516814
    OpenUrlCrossRefPubMed
  17. ↵
    1. Al-Haddad BJS,
    2. Jacobsson B,
    3. Chabra S,
    4. et al
    . Long-term risk of neuropsychiatric disease after exposure to infection in utero. JAMA Psychiatry2019;76:594-602. doi:10.1001/jamapsychiatry.2019.0029 pmid:30840048
    OpenUrlCrossRefPubMed
  18. ↵
    1. Brynge M,
    2. Sjöqvist H,
    3. Gardner RM,
    4. Lee BK,
    5. Dalman C,
    6. Karlsson H
    . Maternal infection during pregnancy and likelihood of autism and intellectual disability in children in Sweden: a negative control and sibling comparison cohort study. Lancet Psychiatry2022;9:782-91. doi:10.1016/S2215-0366(22)00264-4 pmid:36087610
    OpenUrlCrossRefPubMed
  19. ↵
    1. Cheol Seong S,
    2. Kim YY,
    3. Khang YH,
    4. et al
    . Data resource profile: the national health information database of the National Health Insurance Service in South Korea. Int J Epidemiol2017;46:799-800.pmid:27794523
    OpenUrlCrossRefPubMed
  20. ↵
    1. Choi A,
    2. Noh Y,
    3. Jeong HE,
    4. et al
    . Association between proton pump inhibitor use during early pregnancy and risk of congenital malformations. JAMA Netw Open2023;6:e2250366. doi:10.1001/jamanetworkopen.2022.50366 pmid:36626173
    OpenUrlCrossRefPubMed
  21. ↵
    1. Noh Y,
    2. Jeong HE,
    3. Choi A,
    4. et al
    . Prenatal and infant exposure to acid-suppressive medications and risk of allergic diseases in children. JAMA Pediatr2023;177:267-77. doi:10.1001/jamapediatrics.2022.5193 pmid:36622684
    OpenUrlCrossRefPubMed
  22. ↵
    1. Margulis AV,
    2. Setoguchi S,
    3. Mittleman MA,
    4. Glynn RJ,
    5. Dormuth CR,
    6. Hernández-Díaz S
    . Algorithms to estimate the beginning of pregnancy in administrative databases. Pharmacoepidemiol Drug Saf2013;22:16-24. doi:10.1002/pds.3284 pmid:22550030
    OpenUrlCrossRefPubMed
  23. ↵
    1. von Elm E,
    2. Altman DG,
    3. Egger M,
    4. Pocock SJ,
    5. Gøtzsche PC,
    6. Vandenbroucke JP,
    7. STROBE Initiative
    . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet2007;370:1453-7. doi:10.1016/S0140-6736(07)61602-X pmid:18064739
    OpenUrlCrossRefPubMedWeb of Science
  24. ↵
    1. Park BJ,
    2. Sung JH,
    3. Park KD,
    4. Seo SW,
    5. Kim SW
    . Report of the evaluation for validity of discharged diagnoses in Korean Health Insurance database.Seoul National University, 2003: 19-52.
  25. ↵
    1. Lee SY,
    2. Chung SE,
    3. Kim DW,
    4. et al.,
    5. Committee on Epidemiology of Korean Epilepsy Society
    . Estimating the prevalence of treated epilepsy using administrative health data and its validity: ESSENCE study. J Clin Neurol2016;12:434-40. doi:10.3988/jcn.2016.12.4.434 pmid:27273925
    OpenUrlCrossRefPubMed
  26. ↵
    1. Bateman BT,
    2. Mhyre JM,
    3. Hernandez-Diaz S,
    4. et al
    . Development of a comorbidity index for use in obstetric patients. Obstet Gynecol2013;122:957-65. doi:10.1097/AOG.0b013e3182a603bb pmid:24104771
    OpenUrlCrossRefPubMed
  27. ↵
    1. Metcalfe A,
    2. Lix LM,
    3. Johnson JA,
    4. et al
    . Validation of an obstetric comorbidity index in an external population. BJOG2015;122:1748-55. doi:10.1111/1471-0528.13254 pmid:25559311
    OpenUrlCrossRefPubMed
  28. ↵
    1. Rassen JA,
    2. Shelat AA,
    3. Myers J,
    4. Glynn RJ,
    5. Rothman KJ,
    6. Schneeweiss S
    . One-to-many propensity score matching in cohort studies. Pharmacoepidemiol Drug Saf2012;21(Suppl 2):69-80. doi:10.1002/pds.3263 pmid:22552982
    OpenUrlCrossRefPubMed
  29. ↵
    1. Sjölander A,
    2. Frisell T,
    3. Kuja-Halkola R,
    4. Öberg S,
    5. Zetterqvist J
    . Carryover effects in sibling comparison designs. Epidemiology2016;27:852-8. doi:10.1097/EDE.0000000000000541 pmid:27488059
    OpenUrlCrossRefPubMed
  30. ↵
    1. American Psychiatric Association
    . Diagnostic and statistical manual of mental disorders.5th ed. American Psychiatric Association, 2013.
  31. ↵
    1. Yu HY,
    2. Zhou YY,
    3. Pan LY,
    4. Zhang X,
    5. Jiang HY
    . Early life antibiotic exposure and the subsequent risk of autism spectrum disorder and attention deficit hyperactivity disorder: a systematic review and meta-analysis. J Autism Dev Disord2022;52:2236-46. doi:10.1007/s10803-021-05121-6 pmid:34081300
    OpenUrlCrossRefPubMed
  32. ↵
    1. Marrus N,
    2. Hall L
    . Intellectual disability and language disorder. Child Adolesc Psychiatr Clin N Am2017;26:539-54. doi:10.1016/j.chc.2017.03.001 pmid:28577608
    OpenUrlCrossRefPubMed
  33. ↵
    1. Miller JE,
    2. Pedersen LH,
    3. Vestergaard M,
    4. Olsen J
    . Maternal use of antibiotics and the risk of childhood febrile seizures: a Danish population-based cohort. PLoS One2013;8:e61148. doi:10.1371/journal.pone.0061148 pmid:23613800
    OpenUrlCrossRefPubMed
  34. ↵
    1. Lum GR,
    2. Olson CA,
    3. Hsiao EY
    . Emerging roles for the intestinal microbiome in epilepsy. Neurobiol Dis2020;135:104576. doi:10.1016/j.nbd.2019.104576 pmid:31445165
    OpenUrlCrossRefPubMed
  35. ↵
    1. Dahlin M,
    2. Prast-Nielsen S
    . The gut microbiome and epilepsy. EBioMedicine2019;44:741-6. doi:10.1016/j.ebiom.2019.05.024 pmid:31160269
    OpenUrlCrossRefPubMed
  36. ↵
    1. Wanleenuwat P,
    2. Suntharampillai N,
    3. Iwanowski P
    . Antibiotic-induced epileptic seizures: mechanisms of action and clinical considerations. Seizure2020;81:167-74. doi:10.1016/j.seizure.2020.08.012 pmid:32827980
    OpenUrlCrossRefPubMed
  37. ↵
    1. Sutter R,
    2. Rüegg S,
    3. Tschudin-Sutter S
    . Seizures as adverse events of antibiotic drugs: A systematic review. Neurology2015;85:1332-41. doi:10.1212/WNL.0000000000002023 pmid:26400582
    OpenUrlCrossRefPubMed
  38. ↵
    1. Chung HJ,
    2. Yang D,
    3. Kim GH,
    4. et al
    . Development of the Korean developmental screening test for infants and children (K-DST). Clin Exp Pediatr2020;63:438-46. doi:10.3345/cep.2020.00640 pmid:32683817
    OpenUrlCrossRefPubMed
  39. ↵
    1. Dolk FCK,
    2. Pouwels KB,
    3. Smith DRM,
    4. Robotham JV,
    5. Smieszek T
    . Antibiotics in primary care in England: which antibiotics are prescribed and for which conditions?J Antimicrob Chemother2018;73(suppl_2):ii2-10. doi:10.1093/jac/dkx504 pmid:29490062
    OpenUrlCrossRefPubMed
  40. ↵
    1. Frieri M,
    2. Kumar K,
    3. Boutin A
    . Antibiotic resistance. J Infect Public Health2017;10:369-78. doi:10.1016/j.jiph.2016.08.007 pmid:27616769
    OpenUrlCrossRefPubMed
  41. ↵
    1. Smieszek T,
    2. Pouwels KB,
    3. Dolk FCK,
    4. et al
    . Potential for reducing inappropriate antibiotic prescribing in English primary care. J Antimicrob Chemother2018;73(suppl_2):ii36-43. doi:10.1093/jac/dkx500 pmid:29490058
    OpenUrlCrossRefPubMed
  42. ↵
    1. Kim BN,
    2. Kim HB,
    3. Oh MD
    . Antibiotic control policies in South Korea, 2000-2013. Infect Chemother2016;48:151-9. doi:10.3947/ic.2016.48.3.151 pmid:27704729
    OpenUrlCrossRefPubMed
  43. ↵
    1. Fleming-Dutra KE,
    2. Hersh AL,
    3. Shapiro DJ,
    4. et al
    . Prevalence of inappropriate antibiotic prescriptions among US ambulatory care visits, 2010-2011. JAMA2016;315:1864-73. doi:10.1001/jama.2016.4151 pmid:27139059
    OpenUrlCrossRefPubMed
  44. ↵
    1. Zoorob R,
    2. Sidani MA,
    3. Fremont RD,
    4. Kihlberg C
    . Antibiotic use in acute upper respiratory tract infections. Am Fam Physician2012;86:817-22.pmid:23113461
    OpenUrlPubMed
  45. ↵
    1. Havers FP,
    2. Hicks LA,
    3. Chung JR,
    4. et al
    . Outpatient antibiotic prescribing for acute respiratory infections during influenza seasons. JAMA Netw Open2018;1:e180243. doi:10.1001/jamanetworkopen.2018.0243 pmid:30646067
    OpenUrlCrossRefPubMed
  46. ↵
    1. Del Mar C
    . Antibiotics for acute respiratory tract infections in primary care. BMJ2016;354:i3482. doi:10.1136/bmj.i3482 pmid:27381415
    OpenUrlFREE Full Text
View Abstract
Back to top