Response

The recent meta-analysis by Green et al, titled ‘Early antibiotic exposure and risk of psychiatric and neurocognitive outcomes: systematic review and meta-analysis’, published in the British Journal of Psychiatry,^{Reference Green, Wrobel, Todd, Marx, Berk and Lotfaliany1} is a significant contribution. The authors’ commendable effort in synthesising available data on early antibiotic exposure and neuropsychiatric consequences is particularly relevant in the context of the widespread use of antibiotics and their potential effects on the developing gut–brain axis, which is crucial for normal development. However, there are several points to which we wish to draw attention, criticise specific methodological components of the paper and propose an analysis based on recent literature.

This meta-analysis contributes valuable information to the extant literature, using data from 30 studies with more than 7 million participants. The authors found weak evidence linking early antibiotic exposure to autism spectrum disorder (ASD), attention-deficit hyperactivity disorder (ADHD) and major depressive disorder (MDD). Significant associations were found between prenatal antibiotic exposure and ASD/ADHD, as well as between childhood exposure and ASD/ADHD/MDD. The authors performed subgroup analyses controlled by the siblings, which is crucial for any observational study due to issues with genetic and confounding familial problems. Furthermore, the emphasis on prenatal and early childhood exposure gives a better picture of the timing of antibiotic use and its benefits and harms. The study’s high quality and low risk of bias are supported by the assessment tools applied by the authors – the Newcastle–Ottawa Scale and GRADE – for the assessment of study quality and the certainty of the evidence, respectively.

The microbiota–gut–brain axis, a vital communication network that supports neurodevelopment via immune and neurochemical pathways, is of utmost importance in this context. Interruptions in this axis during early life, such as those caused by antibiotics, can result in lasting changes to brain structure and function.^{Reference Collins, Cryan, OMahony, Hyland and Stanton2} Additionally, evidence suggests that antibiotic-induced alterations to the gut microbiota could contribute to neuroimmune pathway dysregulation, potentially increasing susceptibility to neurodegenerative conditions later in life.^{Reference Loh, Mak, Tan, Ng, Chan and Yeow3} The involvement of gut dysbiosis in the progression of diseases such as Parkinson’s disease highlights the necessity of further research into the timing and type of antibiotics administered during developmental windows.^{Reference Quigley4} Experimental research in animal models also provides mechanistic evidence, such as that of Volkova et al.^{Reference Volkova, Ruggles, Schulfer, Gao, Ginsberg and Blaser5} Low-dose penicillin exposure in newborn mice altered gut microbiota composition and impacted gene expression in brain regions linked to neuropsychiatric disorders; this further supports the hypothesis of microbiota-mediated effects. Translational studies in humans are needed to validate these mechanisms.

However, like any other study, the present meta-analysis by Green et al has limitations that need to be investigated. These limitations underscore the need for further research to address these issues and enhance our understanding of the topic. To that end, the following are some limitations that have been considered to ensure that a critical approach to the news item is followed.

First, significant statistical heterogeneity (I ² > 75% in most analyses) discusses the similarities between the studies included. Although the authors acknowledge this issue, they could have provided more detailed subgroup analyses or conducted meta-regressions to identify potential sources of heterogeneity. For instance, stratifying studies by antibiotic type, dosage or specific neuropsychiatric outcomes (e.g. ASD v. ADHD) might have offered deeper understanding and insights. Second, many included studies relied on administrative data and retrospective designs, which are prone to biases, such as misclassification of antibiotic exposure or outcome ascertainment. For example, prescriptions do not always equate to actual antibiotic use, and medical records may underreport subclinical neuropsychiatric cases. Prospective cohort studies with stringent data collection protocols would yield more reliable evidence.

Thirdly, the GRADE framework classified most outcomes as having ‘very low certainty’, primarily due to methodological shortcomings and the absence of sibling-controlled analyses for outcomes such as MDD. The authors acknowledged these issues, yet they do not provide concrete suggestions for addressing these research gaps in the future. Last, the potential for confounding by indication – where the underlying infections requiring antibiotics, rather than the antibiotics themselves, may influence observed outcomes – has been inadequately explored. Although briefly discussed, more profound adjustments for variables such as infection severity and maternal health would strengthen the authenticity of the findings.

Recent studies offer complementary perspectives that help contextualise and extend the findings of Green et al. One example is a large population-based cohort study from South Korea that found no significant associations between early antibiotic exposure and risks of ASD, intellectual disorder or language disorder after accounting for familial confounding.^{Reference Choi, Lee, Jeong, Lee, Kwon and Han6} Nevertheless, it did report a slight increase in epilepsy risk among infants with prolonged antibiotic use, emphasising the importance of sibling-controlled designs and distinguishing between short- and long-term exposure. Likewise, a cohort study in New Zealand^{Reference Slykerman, Neumann, Underwood, Hobbs and Waldie7} demonstrated associations between antibiotic exposure in the first year of life and behavioural challenges and reduced language development by age 4.5 years, even after adjusting for confounders such as socioeconomic status and otitis media. These results highlight the importance of the timing of antibiotic exposure, underscoring sensitive developmental windows.

Moreover, protective strategies need exploration. Evidence suggests that probiotics can reduce neuropsychiatric risks by mitigating dysbiosis caused by antibiotics, restoring gut microbiota balance and promoting resilience in the gut–brain axis.^{Reference Bicknell, Liebert, Borody, Herkes, McLachlan and Kiat8} Similarly, breastfeeding has been identified as a mitigating factor that could buffer against microbiota disruptions caused by antibiotics.^{Reference Volker, Tessier, Rodriguez, Yager and Kozyrskyj9} Exploring these strategies might inform clinical guidelines and significantly improve developmental outcomes for affected children.

To better understand this critical area, several key questions remain. What roles do different classes of antibiotic play in shaping neurodevelopmental and neuropsychiatric outcomes? Stratifying future research by antibiotic type (e.g. broad- v. narrow-spectrum) could reveal differential risks. How does the severity of infection and maternal health interact with antibiotic use in influencing neurodevelopment? Disentangling the effects of antibiotics from those of underlying infections is essential. Can interventions such as probiotics or breastfeeding lessen the potential adverse effects of early-life antibiotics? Identification of protective factors could inform clinical guidelines and is a key focus for future research. What is the impact of cumulative antibiotic exposure across childhood on long-term neuropsychiatric outcomes? Longitudinal studies with repeated exposure measurements and results are crucial to answering this question. These questions underscore the need for ongoing research in this area, engaging researchers and clinicians in pursuing a deeper understanding of these complex issues and making them feel motivated and focused on identifying protective factors. Future research should prioritise prospective cohort designs, incorporate sibling-controlled analyses and investigate the roles of antibiotic type, infection severity and mitigating factors such as probiotics and breastfeeding. We can better understand the intricate relationship between early antibiotic exposure, gut microbiota and neurodevelopment by addressing and attempting to fill these gaps. This understanding could lead to informed clinical practices that optimise long-term outcomes for patients in the fields of psychiatry and neurodevelopment.