Child marriage or cohabitation before 18 years disproportionately affects girls worldwide, but evidence on its longer-term mental health impacts has been limited and often cross-sectional. This study used longitudinal cohort data to examine whether child marriage is associated with reduced life satisfaction and increased emotional symptoms, and to distinguish pre-existing vulnerabilities from any additional effects of marriage.
Analyses drew on the Young Lives cohort across four countries: Ethiopia, India, Peru, and Vietnam. The prevalence of child marriage/cohabitation among girls in these cohorts varied by country, from approximately 5% in Peru to 20% in India. The study followed participants from childhood into early adulthood, with outcome measurement between ages 19 and 29.
The study focused on two mental health-related outcomes captured in the Young Lives data: self-reported life satisfaction and measures of emotional symptoms. Life satisfaction was examined cross-sectionally at age 19 and longitudinally across ages 19–29. Emotional symptoms were compared across groups in the same age ranges.
To estimate associations and probe causality, the authors used multiple complementary methods. They applied linear regression to estimate pooled and country-specific differences. To address confounding by pre-existing disadvantage, they implemented difference-in-differences (DiD) estimation and propensity score matching (PSM). These approaches aimed to separate baseline differences that precede marriage from any subsequent change potentially attributable to child marriage itself.
Pooled regression results indicated that girls who experienced child marriage/cohabitation had lower life satisfaction compared with those who did not. Specific pooled estimates reported were a difference at age 19 of -0.25 (95% CI = -0.40 to -0.10) and across ages 19–29 of -0.22 (95% CI = -0.32 to -0.13). Heterogeneity estimates suggested consistently negative impacts across the four countries.
Importantly, the study found that children who later became child brides already had lower life satisfaction at age 8, prior to marriage. This pattern points to pre-existing social and economic disadvantages—such as poverty and other risk factors—that both reduce early life satisfaction and increase the probability of child marriage.
When accounting for these baseline differences using DiD and PSM, much of the observed association was explained by prior vulnerability. The pooled matched estimate from PSM indicated a small additional adverse effect of child marriage on life satisfaction at age 19 (matched estimate -0.09, 95% CI = -0.19 to 0.00). This indicates that while most of the life satisfaction gap is attributable to earlier disadvantage, there remains a modest additional negative association with child marriage itself at least at the earlier adult measurement point.
The analyses did not detect differences in emotional symptoms between girls who married or cohabited before 18 and those who did not. Additionally, within the group of child-married participants, the study did not find evidence of differential outcomes by exact age at marriage.
The authors interpret the results as indicating that girls who become child brides have persistently lower life satisfaction from childhood through early adulthood, but that this is largely explained by pre-existing disadvantage and risk factors that precede marriage. There appears to be a small additional negative effect of child marriage on life satisfaction measured at age 19 after matching on baseline characteristics. No effect was observed for emotional symptoms, and no within-child-married heterogeneity by age at marriage was detected.
These findings emphasise the importance of accounting for early-life vulnerability when studying the consequences of child marriage and suggest that interventions to reduce child marriage should be linked with broader efforts to address poverty and disadvantage.
The research was conducted under the Young Lives study governance with ethical approvals from the University of Oxford Central University Research Ethics Committee and relevant committees in each study country. The authors declared no competing interests. Data used in the analysis are available through the UK Data Service. Funding support was declared from the Wellcome Trust. The article is a preprint and has not yet been peer reviewed; it should not be used to guide clinical practice.