This secondary analysis of the HPTN-068 trial in rural Agincourt, South Africa, characterised longitudinal patterns of household hunger among adolescent girls and young women (AGYW). Using the Household Hunger Scale across baseline (2011/12) and two subsequent follow-up visits, the investigators applied Group-Based Trajectory Modelling to identify distinct hunger trajectories over time. Two trajectories emerged: no hunger (82% of participants) and marginal hunger (18%). Reported prevalence of moderate-severe household hunger was 11.0% at baseline, 10.8% at the first follow-up, and 6.0% at the second follow-up.
Mental health outcomes were measured using self-reported instruments at the last two follow-up visits. The analysis focused on incident mental health outcomes: new-onset anxiety symptoms, incident depressive symptoms, and measures of hope (reported as hopelessness prevalence). In this sample, incident anxiety symptoms were reported by 4.5% of AGYW, incident depressive symptoms by 20.0%, and hopelessness by 52.8%.
The study used complete data from 1,779 AGYW enrolled in HPTN-068 with baseline and three annual follow-up visits. Household food insecurity was operationalised via the Household Hunger Scale. Trajectories were estimated through Group-Based Trajectory Modelling to categorise temporal patterns of household hunger. Mental health outcomes were self-reported and derived from the participants' responses at the last two follow-up visits; incident outcomes reflect new reports during that interval.
Following trajectory identification, the authors used covariate-adjusted modified Poisson regression models to estimate relative risks (RRs) for the association between hunger trajectory group membership and each mental health outcome: incident anxiety, incident depression, and hope. The models adjusted for relevant covariates (details of specific covariates were not reported in the source summary). Confidence intervals (95% CI) for RRs are presented to assess statistical precision.
Two hunger trajectories (no hunger and marginal hunger) described the longitudinal household hunger experience for AGYW in this cohort. Prevalences of moderate-severe household hunger at the three time points were 11.0%, 10.8%, and 6.0%, respectively. Reported incident mental health outcomes were: 4.5% for anxiety, 20.0% for depression, and 52.8% for hopelessness.
In adjusted regression analyses, hunger trajectory membership was not associated with increased risk of incident anxiety [RR 1.09, 95% CI 0.55–2.18], incident depressive symptoms [RR 0.97, 95% CI 0.72–1.33], or hopelessness/hope [RR 1.00, 95% CI 0.81–1.23]. The confidence intervals include the null value in all reported comparisons, indicating no statistically significant associations in this sample.
In this longitudinal sample of AGYW from rural South Africa, distinct household hunger trajectories were identified but were not associated with the measured incident anxiety, incident depressive symptoms, or hope after covariate adjustment. The authors highlight the need to better understand factors that support resilience and mental health among AGYW in this setting to guide intervention design. The paper notes that existing literature has many cross-sectional studies linking food insecurity and mental health, and this longitudinal analysis provides additional temporal perspective but did not find the hypothesised associations.
The original HPTN-068 trial received ethical approvals from the Institutional Review Board at the University of North Carolina at Chapel Hill, the University of the Witwatersrand Human Research Ethics Committee, and the Mpumalanga Province Health Research and Ethics Committee. Written informed consent was obtained from participants aged 18 or older and their parents or caregivers; written assent was obtained for AGYW under 18. The secondary analysis was deemed “Not a Human Subjects Research” by Indiana University Bloomington because de-identified data were used. The trial is registered at ClinicalTrials.gov (NCT01233531). The authors declared no competing interests.
The HPTN-068 dataset is archived at Fred Hutch Cancer Center, Seattle, WA. Requests for the full dataset can be sent to HPTN-Data-Access@scharp.org. The source indicates the HPTN-068 data are publicly available via the Harvard Dataverse repository (https://dataverse.harvard.edu/dataverse/HPTN068).