Maternal diabetes mellitus (MDM), encompassing both gestational diabetes mellitus (GDM) and pregestational diabetes mellitus (PGDM), is increasingly prevalent globally, linked to rising obesity rates, advancing maternal age, and improved diagnostic techniques. Concerns exist regarding the potential adverse effects of MDM on offspring neurodevelopment. Animal studies suggest that hyperglycemia during pregnancy may negatively impact inflammatory responses, oxidative stress, and epigenetic mechanisms, potentially contributing to poor neurodevelopmental outcomes. Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disorder characterized by inattention, hyperactivity, and impulsivity. Previous studies have reported an association between MDM and ADHD in offspring, but these studies often lacked sufficient statistical power, relied on self-reported data, or inadequately addressed potential confounders, particularly shared genetic and environmental familial factors. This study aims to address these limitations by leveraging a large multinational cohort of linked mother–child pairs to rigorously assess the association between MDM (GDM and PGDM) and the risk of ADHD in offspring, while accounting for multiple confounding factors.

Prior research has indicated a potential link between maternal diabetes and ADHD in offspring. A previous meta-analysis suggested a substantially increased risk of ADHD in children of diabetic mothers. However, limitations in these studies included reliance on self-reported data, limited statistical power, insufficient adjustment for confounders (especially familial factors), and a predominantly White population sample. The current study sought to overcome these limitations through a larger, more diverse, and rigorously controlled investigation.

This population-based cohort study utilized linked mother–child pair data from seven regions: Hong Kong, New Zealand, Taiwan, Finland, Iceland, Norway, and Sweden (the latter four under the Nordic Pregnancy Drug Safety Studies (NorPreSS) collaboration). The study included over 3.6 million mother–child pairs with follow-up until 2020. Data sources varied across regions, including electronic health records, national registers, and insurance claims databases. A common data model and analytic approach were implemented to ensure consistency across diverse data sources. MDM was categorized into any MDM, GDM, and PGDM, with further sub-categorization of PGDM into type 1 and type 2. ADHD diagnosis was defined using region-specific criteria based on diagnostic codes and medication prescriptions. The primary analysis employed Cox proportional hazards regression, adjusted for numerous covariates (including demographics, socioeconomic status, maternal health conditions, and medication use). Propensity score (PS) fine-stratification weighting was used to address potential confounding. Sibling-matched analyses were conducted for GDM to control for shared familial factors. Sensitivity analyses were performed to assess the robustness of the findings.

The study found a small-to-moderate association between MDM overall and ADHD (pooled hazard ratio (HR) = 1.16, 95% confidence interval (CI) = 1.08–1.24). Higher risks were observed for both GDM (pooled HR = 1.10, 95% CI = 1.04–1.17) and PGDM (pooled HR = 1.39, 95% CI = 1.25–1.55). However, the sibling-matched analysis for GDM revealed no significant difference in ADHD risk between siblings with discordant GDM exposure (pooled HR = 1.05, 95% CI = 0.94–1.17), suggesting that the association between GDM and ADHD may be largely due to confounding by unmeasured shared familial factors. Children of mothers with medicated GDM had a similar risk of ADHD compared to those with unmedicated GDM. Sensitivity analyses, including those stratified by sex and using alternative regression models, yielded consistent results.

This large-scale multinational study provides more nuanced insights into the relationship between MDM and ADHD in offspring compared to previous studies. While the overall findings indicate a small-to-moderate increased risk of ADHD associated with MDM, the lack of association between GDM and ADHD in the sibling-matched analysis raises important questions about the role of confounding factors. Shared familial factors, genetic predisposition, or other unmeasured environmental influences may explain a substantial portion of the observed associations, particularly concerning GDM. The similar risk of ADHD in offspring of mothers with medicated versus unmedicated GDM suggests that the severity of GDM itself might be a more critical factor than treatment. The E-values suggest that any unmeasured confounding is unlikely to entirely explain away the observed associations, although residual confounding could reduce the risk estimates. These findings challenge previous reports of substantially higher risk estimates and underscore the complexity of the relationship between MDM, hyperglycemia, and ADHD.

This study, using a large multinational cohort, demonstrates a small-to-moderate association between maternal diabetes mellitus (MDM) and ADHD in offspring. The association with GDM appears largely attributable to confounding by shared familial factors. Further research is needed to identify and control for these confounders and to explore the roles of both genetic and environmental factors in the etiology of ADHD in the context of maternal diabetes. Investigating the influence of disease severity and paternal factors is also crucial.

The study acknowledges several limitations. The use of administrative data from diverse sources could introduce heterogeneity in diagnostic criteria and coding practices. Inherent limitations of observational studies, such as potential for unmeasured confounding and residual confounding, remain. The definition of ADHD using both diagnostic codes and medication prescriptions may have resulted in some misclassification. Finally, certain maternal lifestyle factors may not have been fully captured, potentially contributing to residual confounding. The sibling-matched analyses helped address some of these concerns, but this design might also amplify confounding from factors unique to each sibling.