Childhood Adversities Leave Lasting Mark on Depression Risk in Young South African Women

A seven-year study tracking more than 2,000 adolescent girls and young women in rural South Africa has revealed which childhood adversities cast the longest shadows over mental health, and the answer challenges some long-held assumptions about what hurts young people most. Researchers analyzing data from the HPTN 068 trial found that the emotional bond with a caregiver, hunger at home, violence from a partner, and violence at school each measurably raised the prevalence of depression, while the loss of both parents, often assumed to be the most devastating blow, had surprisingly little consistent effect. The findings, published in SSM – Mental Health, offer one of the most detailed causal pictures yet of how early adversity shapes depression across the transition from adolescence to young adulthood in a low- and middle-income country.

Depression among adolescents is a formidable global health problem, with roughly one-third of adolescents worldwide estimated to experience depressive symptoms. The burden falls disproportionately on young people in low- and middle-income countries, where prevalence estimates in sub-Saharan Africa range from about 27 percent to more than 70 percent in some studies, compared with roughly 23 percent in North America and Europe. Adolescence is also a sensitive developmental window: depression that begins during these years can disrupt neurodevelopment, undermine social functioning, and increase the risk of suicide, intimate partner violence, and HIV acquisition. Yet despite effective treatments existing, modeling studies suggest treatment alone can avert only a fraction of the total disease burden, making prevention a critical public health priority.

The research team, led by Valerie A. Lucas of the University of North Carolina at Chapel Hill together with colleagues including Audrey Pettifor and Kathleen Kahn, turned to the HPTN 068 trial, a randomized controlled trial conducted from 2011 to 2015 in rural South Africa that had tested whether conditional cash transfers tied to school attendance could reduce HIV risk in adolescent girls and young women. The team conducted a secondary analysis of 2,116 participants who were under 18 at enrollment, with a mean age of 15.1 years. Participants reported on six specific adverse childhood experiences at baseline: double orphanhood, food insecurity, low perceived care from their guardian, physical intimate partner violence, sexual violence, and violence at school. Depressive symptoms were then measured annually during the trial and at two additional follow-up visits in 2016/17 and 2018/19, using the 20-item Center for Epidemiologic Studies Depression scale, a validated screening tool in South African adolescents.

The scale of adversity in the sample was striking. Food insecurity was the most common experience, reported by 33.1 percent of participants in the previous year, followed by school violence at 18.4 percent, physical intimate partner violence at 15.0 percent, sexual violence at 9.6 percent, low guardian care at 5.9 percent, and double orphanhood at 4.7 percent. Nearly 40 percent of participants reported no adversities, but a substantial minority reported two or more. Intriguingly, the adversities were largely independent of one another, with the strongest correlation, between physical partner violence and sexual violence, registering at just 0.207. This statistical independence suggests each type of adversity operates through distinct mechanisms, and it allowed the researchers to estimate each one’s effect separately rather than lumping them into a single additive score, an approach they argue can obscure important differences between experiences.

To move beyond simple association toward causal inference, the team employed g-computation, a form of marginal structural model developed by epidemiologist James Robins. The technique works in two steps: first, a logistic regression model estimates each participant’s probability of depression at each follow-up visit given her actual adversity history and covariates; then, the researchers create counterfactual copies of the dataset in which every participant is set either to have experienced or not experienced a given adversity. Averaging predicted depression probabilities across these counterfactual scenarios yields an estimate of what depression prevalence would have been if the adversity had been completely eliminated, or universally experienced. The difference between the two is the prevalence difference, the estimated causal effect. Standard errors were computed with 500 bootstrap replications that sampled each participant’s full set of visits as a cluster, and the team quantified vulnerability to unmeasured confounding using E-values.

The results were revealing. At the first follow-up year, when adjusted depression prevalence peaked at 31.9 percent, the two largest single-adversity effects came from low perceived care from a guardian, which raised depression prevalence by 11.1 percentage points, and physical intimate partner violence, which raised it by 10.1 percentage points. Food insecurity added 6.3 percentage points, and school violence 5.2. Most striking of all, young women who had experienced both physical intimate partner violence and sexual violence showed a 15.7 percentage point elevation in depression prevalence, the largest effect of any exposure in the study, an effect that remained substantial at 14.4 percentage points five years after baseline.

Double orphanhood, by contrast, defied expectations. Despite a substantial body of literature linking parental loss to depression among children affected by HIV in sub-Saharan Africa, losing both parents before age 18 produced only a small and statistically insignificant effect of 3.1 percentage points at Year 1, and by Year 7 the estimate had actually turned negative at minus 5.7 percentage points. The researchers suggest a possible explanation rooted in the social fabric of rural South Africa: extended family networks frequently step in to fill the caregiver role for orphaned children. What matters for mental health, the data imply, is not whether the biological parents are alive, but whether a caring adult is present in the young woman’s life. Low perceived guardian care, measured by a single question asking whether the most involved guardian cared about her not at all, proved far more damaging than orphanhood itself.

The trajectory of effects over time also carried important lessons. Depression prevalence in the cohort fell dramatically from 31.9 percent at Year 1 to just 8.8 percent by Year 7, and most adversity effects shrank alongside it. But some persisted with remarkable tenacity. Food insecurity raised depression prevalence by 6.3 to 7.7 percentage points across the first five years of follow-up. School violence, remarkably, continued to elevate depression even after most participants had left secondary school, with a significant 6.4 percentage point effect at Year 5, suggesting that the psychological damage inflicted in unsafe school environments outlasts the exposure itself. Physical partner violence remained significant at Year 5 with an 8.3 percentage point effect. The E-value analysis indicated that an unmeasured confounder would need to be associated with both the exposure and depression by a factor of roughly two to explain away the significant findings, a threshold the authors consider implausible given the weak correlations among adversities in their sample.

The authors are candid about limitations. All exposures and outcomes were self-reported, and current depressive symptoms could color how young women recalled or reported past adversities. The CES-D measures symptoms rather than providing clinical diagnoses, and a sensitivity analysis using a stricter cutoff of 20 or higher produced similar patterns with smaller magnitudes. The study captured adversities only at baseline, ignoring new exposures, severity, and frequency after age 18, and outcome data were missing at some visits, particularly Year 3 when only 38.6 percent of the sample completed the depression assessment, though the imputation approach assumed missingness was random conditional on measured variables. Adversities with low prevalence, such as double orphanhood and sexual violence, also had limited statistical power, producing wider confidence intervals.

Nevertheless, the study’s strengths are considerable: an explicitly causal analytic framework, temporally separated measurement of exposures and outcomes, seven years of follow-up, and extensive sensitivity analyses that consistently supported the main findings. The practical implications are concrete. Supporting caregivers and their engagement in young people’s lives, regardless of orphan status, could buffer depression risk. Consumption-smoothing interventions and income diversification could reduce the food insecurity that affects a third of households. Economic empowerment programs, cash transfers, and trauma-informed care could reduce young women’s exposure to and recovery from partner violence. And school-based interventions targeting violent teachers, unsafe environments, and harassment could yield population-level mental health benefits that persist well beyond graduation. In a country where up to 88 percent of young adults report at least one childhood adversity and 30 percent experience depression, identifying which adversities matter most, and for how long, may be the first step toward breaking the cycle.

Subject of Research: Effects of adverse childhood experiences on depression prevalence in adolescent girls and young women in rural South Africa

Article Title: The effect of adverse childhood experiences on depression prevalence in adolescent girls and young women over time: a secondary analysis of HPTN 068 trial data

Article References: Lucas, V. A., Reyes, L. M., Maselko, J., Stoner, M., Rosenberg, M., DeLong, S. M., Bhushan, N. L., Moffett, B., Beidelman, E. T., Klein, M., Kahn, K., & Pettifor, A. (2026). The effect of adverse childhood experiences on depression prevalence in adolescent girls and young women over time: a secondary analysis of HPTN 068 trial data. SSM – Mental Health, 10, Article 100706. https://doi.org/10.1016/j.ssmmh.2026.100706

Image Credits: AI Generated

DOI: 10.1016/j.ssmmh.2026.100706

Keywords: adverse childhood experiences, depression, adolescent girls, South Africa, food insecurity, intimate partner violence, school violence, orphanhood, g-computation, HPTN 068, global mental health, CES-D