Exploring parental prenatal influences on child health: A multicohort study and data visualisation tool

BACKGROUND: The Developmental Origins of Health and Disease (DOHaD) hypothesis suggests that early life environmental exposures, especially during pregnancy, can impact long-term health. Research has largely relied on correlational evidence and has focussed on maternal factors, with less attention given to paternal, postnatal, and broader social determinants. This focus could complicate efforts to determine the most effective strategies for improving population health. METHODS AND FINDINGS: Using harmonised data across four population-based longitudinal cohort studies from the UK and Norway (births between 1991 and 2008), we took a systematic approach to explore associations of parental prenatal health behaviours (smoking, alcohol, and caffeine consumption) and low socioeconomic position (SEP) with 72 child health-related outcomes (e.g., related to body size and composition, cognitive function, mental health, blood pressure, allergy, etc.) from birth up to age 11. Where possible, cohort estimates were meta-analysed, yielding a maximum sample size of 232,139. We triangulated evidence of causality using different analytical approaches, including a Mendelian randomisation-informed approach using genetic risk scores, negative controls (maternal versus paternal and during versus post-pregnancy comparisons), and dose-response analyses. This comprehensive set of analyses generated more than 594,000 effect estimates. We developed a web app, 'EPoCH Explorer' to visualise and share our results in an accessible format. We did not find strong evidence for widespread or large effects of parental health behaviours on child health and wellbeing. Only 6% of analyses had a Cohen's D value >0.2 and FDR-adjusted P < 0.05. In most analyses, the effect estimate was similar for mothers and partners, with 51% showing a larger effect for mothers and 49% for partners. Despite the lack of widespread associations, we found consistent evidence of association between maternal smoking and small for gestational age, higher childhood body mass index (BMI), depressive symptoms, and behavioural issues, while partner smoking was consistently associated with childhood BMI and social communication difficulties. Overall, we found stronger evidence of child outcomes being associated with low SEP than with health behaviours: 15% of results for low SEP had a Cohen's D value >0.2 and FDR-P < 0.05, compared to 6% for parental smoking, 3% for alcohol, and 0.4% for caffeine. Sample sizes and statistical power varied across outcomes and there was limited ethnic diversity in our study samples. CONCLUSION: Our findings suggest that wider familial socioeconomic conditions may be a more important determinant of child health than specific parental health behaviours prenatally. Interventions to improve population health may be most effective if they target wider social inequalities, rather than individual behaviours and mothers specifically. We encourage researchers to use EPoCH Explorer to prioritise associations for exploration in their own datasets, thus enabling replication and cross-context comparison to validate and extend the generalisability of our findings.