Girls carried a larger distress burden in every country studied
Psychological distress was higher in girls than boys at age 18 to 19 across Ethiopia, India, Peru and Vietnam, but the size of that gap varied sharply. The smallest difference appeared in Ethiopia and Vietnam, where girls scored 0.62 points higher on the Strengths and Difficulties Questionnaire Emotional Symptoms scale. India showed a wider separation at 1.76 points, and Peru produced the largest gap at 2.45 points. Those differences are not trivial in a developmental period when emotional symptoms can shape school participation, peer relationships, and the transition into adult roles. They also warn against treating the gender gap in adolescent distress as a single universal pattern with one explanation that fits every setting.
The investigators built a multi-domain framework to sort possible drivers into seven categories: biological, cognitive, psychological, behavioural, social, environmental, and broader societal factors. That structure matters because gender differences in mental health rarely emerge from one source alone. Some risks are more common in girls, some have a stronger effect on girls once present, and some do both depending on the local context. The analysis used longitudinal Young Lives cohort data from childhood through adolescence, which allowed the team to test not just whether explanatory factors differed by sex, but whether their association with later distress also differed.
Role overload appeared repeatedly, but its meaning changed by context
One of the clearest recurring patterns involved role overload. Girls experienced greater role overload in every country, making it one of the few explanatory factors that was consistently gendered in prevalence across all four settings. In practical terms, the burden of balancing school, domestic responsibilities, social expectations, and other competing demands may accumulate earlier or more intensely for girls, and those pressures can feed emotional symptoms during late adolescence.
Study amount also showed gendered prevalence in some countries, but not all. Girls reported more study time in Peru and Vietnam, while no comparable difference appeared in Ethiopia or India. That uneven pattern argues against assuming that academic pressure alone explains the distress gap everywhere. The same is true of the broader social environment: gendered exposure exists, but its shape depends on local schooling norms, household expectations, and the distribution of unpaid work.
Self-esteem produced one of the sharpest examples of differential susceptibility. Higher self-esteem was more protective for boys than girls in India, Peru and Vietnam, yet more protective for girls in Ethiopia. A single psychosocial factor can therefore move in opposite directions depending on social context. That may reflect differences in how self-regard is formed, reinforced, or challenged within families, schools, and peer groups. It may also indicate that the same psychological resource has different downstream meaning when girls and boys are navigating distinct social constraints.
The pathways to distress were not interchangeable across the four cohorts
The study’s country-specific regressions with gender-by-factor interaction terms showed that gendered impact was more variable than gendered prevalence. In other words, the question was not simply whether girls and boys experienced different levels of an exposure; it was also whether the exposure translated into distress differently by sex. That distinction is important because a factor that is common among girls may still explain little of the gap if its effect on distress is similar in both sexes. Conversely, a factor with only modest prevalence differences can still contribute substantially if its impact is stronger in girls.
The machine-learning models reinforced that point. Boosted-tree analyses, interpreted with SHAP values and permutation feature importance, identified important predictors across all seven domains, but no single domain dominated in every country. Biological, cognitive and psychological variables all entered the predictive mix, alongside social and environmental measures. The resulting pattern looks less like a hierarchy of causes and more like a country-specific signature assembled from several layers of risk and resilience. For clinicians, educators and youth workers, the implication is straightforward: screening and prevention strategies should not assume that one set of adolescent stressors explains girls’ distress everywhere.
The age window studied also deserves attention. Distress was measured at 18 to 19 years, after years of accumulated exposure to family roles, school demands and social comparison. By late adolescence, repeated small pressures can become clinically meaningful even when no single exposure seems extreme on its own. A family that normalizes greater caregiving expectations for daughters, a school environment that rewards compliance while penalizing visible emotion, or a community that constrains girls’ autonomy can each push emotional symptoms higher over time. The biological pathway may involve stress reactivity and persistent arousal, but the social architecture determines when and how that load is delivered.
Cross-country differences limit any one-size-fits-all explanation
Peru stood out with the largest distress gap, while Ethiopia and Vietnam showed the smallest. Those differences alone weaken any claim that a universal gender mechanism drives adolescent emotional symptoms in the same way across settings. Country context appears to mediate both exposure and response, and the same factor may sit at a different point in the causal chain depending on local gender norms and life-course pressures. A framework built for one country can mislead when exported unchanged to another.
The authors’ use of longitudinal data is a major strength because it links earlier-life explanatory factors to later outcomes rather than relying only on cross-sectional snapshots. Even so, the analysis remains observational and should be read as evidence of association, not proof of causality. The study also relied on the emotional symptoms subscale of the Strengths and Difficulties Questionnaire, which captures distress but not every dimension of adolescent mental health. The data are especially useful for mapping risk architecture, less so for settling which mechanism deserves the label of primary cause.
Mental health policy often searches for a single intervention point. This study suggests that adolescent girls’ distress is assembled from multiple pressures that differ by country, by domain, and by the social meaning attached to each exposure. Any prevention strategy that ignores that heterogeneity will miss the children most likely to carry the burden into adulthood.