Reducing mental health inequalities through hypothetical interventions in early childhood: Evidence from mediation analysis using the UK Millennium Cohort Study

The early childhood environment influences later mental health, but it remains unclear which aspects are most important for reducing inequalities. We quantified how socioeconomic inequalities in mental health at 5 and 17 years (y) would change under hypothetical interventions on five aspects of the early childhood environment. We analysed parent-reported data from the UK Millennium Cohort Study (n=15513). Socioeconomic circumstances (SEC) were assessed using maternal education at 9 months. Mental health was measured using the Strengths and Difficulties Questionnaire total difficulties score at both ages. Five mediator pathways - neighbourhood, social support, material home environment, household adversity, and parent-child relations - were identified through stakeholder consultation and literature. Interventional Disparity Measures (IDMs) estimated how mental health inequalities would change if the confounder-adjusted mediator distributions for children in low SEC were shifted to match those in high SEC, using Monte Carlo simulations with imputed missing data. Mental health difficulties were higher among children in the low SEC (Primary or Lower) compared to high SEC (Tertiary) group (5y: 3.3[95%CI: 3.0;3.6] points; 17y: 3.1[2.8;3.5] points). Shifting all five pathways reduced these inequalities by around one-third (5y: -32%[-40;-23]; 17y: -35%[-48;-26]). At 5y, most of the inequality reduction was attributable to shifts in parent-child relations (-22% [-27;-16]). At 17y, both parent-child relations (-15%[-22;-9]) and social support (-11%[-21;-6]) were equally important. To reduce socioeconomic inequalities in mental health, policymakers should prioritise early interventions that improve parent-child relations and access to social support, while recognising the broader influence of neighbourhood, material, and household psychosocial contexts on these pathways.