Prenatal environmental exposures associated with ADHD diagnosis in children: a multicentre exposome analysis

Attention deficit hyperactivity disorder (ADHD) is one of the most diagnosed neurodevelopmental conditions. While its aetiology is primarily attributed to genetic factors, the complex interplay with environmental exposures remains poorly understood. We investigated the relationship between prenatal environmental exposures (excluding biomarkers) and ADHD, utilizing data from seven cohorts (n = 156,563 children) across Europe. A supervised machine learning model (Random Forest) was run at each cohort site to identify the environmental exposures most informative for classifying ADHD. The results were pooled by averaging variable importance ranks across cohorts and highest-ranking variables were then used in logistic regression to obtain interpretable associations. Finally, a random-effects meta-analysis with restricted maximum likelihood estimation was used to infer the associations across cohorts. Based on Random Forest, built and natural urban characteristics, air pollution, and maternal socioeconomic factors were selected as variables with the greatest predictive contribution to ADHD classification. Associations with built and natural urban characteristics and air pollution were not robust across cohorts and largely attenuated in fully adjusted models. Meta-analysis across cohorts showed increased odds of ADHD for maternal smoking during pregnancy (OR = 1.67, 95% CI: 1.31, 2.14, I² = 65.5%) and low maternal education (OR = 1.62, 95% CI: 1.27, 2.07, I² = 51.3%), with substantial heterogeneity. Our findings indicated that maternal smoking during pregnancy and low maternal education were associated with increased odds of ADHD diagnoses in children, suggesting the potential role of socioeconomic disparities. However, gene-environment correlation cannot be ruled out.