¹Deakin University, IMPACT, Department of Medicine, Geelong, Australia, ²Deakin University, IMPACT, School of Medicine, Geelong, Australia, ³Univeristy of Tasmania, School of Psychological Sciences, Hobart, Australia, ⁴University College Cork and APC Microbiome, Department of Anatomy and Neuroscience, Cork, Ireland, ⁵Barwon Health, Department of Medicine, Geelong, Australia, ⁶Deakin University, School of Medicine, Geelong, Australia, ⁷Massachusetts General Hospital, Dauten Family Center for Bipolar Treatment Innovation, Department of Psychiatry, Boston, United States

Background: There is a compelling body of data attesting to the role of the microbiota-gut-brain axis in mental health. The early life period poses a critical window wherein disturbances in the intestinal microbiota may adversely impact neurodevelopment. Antibiotics have a clear impact on the intestinal microbiome, yet their long-term safety regarding early exposure and later life psychiatric or neurocognitive outcomes has not been broadly evaluated.

Aims & Objectives: This systematic review and meta-analysis aimed to address a critical gap in literature by evaluating the association between in-utero and early childhood (ages 0-2) antibiotic exposure and the likelihood of psychiatric and neurocognitive problems in later life.

Method: We conducted a systematic review and meta-analysis of studies evaluating in-utero or early childhood (ages 0-2) antibiotic exposure compared to unexposed controls and later psychiatric or neurocognitive outcomes. We searched MEDLINE, PsychINFO and EMBASE on 20/11/23. Quality assessments included the GRADE certainty assessment and Newcastle-Ottawa Scale.

Results: Thirty studies were included (n=7,047,853 participants). There were weak associations between in-utero antibiotic exposure and later development of ASD (OR= 1.09, 95%CI: 1.02-1.16) and ADHD (OR=1.19, 95%CI: 1.11 to 1.27). Early-childhood exposure was associated with later development of ADHD (OR=1.33, 95%CI: 1.20 to 1.48), ASD (OR=1.19, 95%CI: 1.01 to 1.40), and MDD (OR=1.29, 95%CI: 1.04 to 1.60). However, in sibling-controlled studies, there were no associations between either exposure period and later ASD or ADHD. No studies in MDD used sibling-controls. All meta-analyses were rated very low certainty using the GRADE certainty assessment, largely owing to methodological and statistical heterogeneity, except for childhood antibiotic exposure and later ASD (sibling-controlled data), which was low certainty.

Discussion & Conclusions: Weak evidence was found supporting the association between in-utero and early-life antibiotic exposure, and later neurodevelopmental outcomes. However, these associations were attenuated in studies that used sibling controls, which is a more robust study design. Thus, genetic and familial confounding may explain the associations seen, and non-sibling-controlled data should be interpreted with caution. We recommend that future studies utilise sibling-controls.