Assessing longitudinal patterns of depressive symptoms and the influence of symptom trajectories on HIV pre-exposure prophylaxis adherence among adolescent girls in the HPTN 082 randomized controlled trial

Introduction African adolescent girls and young women (AGYW) eligible for HIV pre‐exposure prophylaxis (PrEP) experience high levels of depressive symptoms. Depression can reduce PrEP adherence among adults, although analyses have considered depression as a time‐varying exposure rather than modelling distinct patterns of symptoms. The association between depressive symptoms and PrEP adherence has not been explored for AGYW. To address these gaps, we sought to understand depressive symptom trajectories among African AGYW initiating PrEP and the impact of time‐varying depressive symptoms and symptom trajectories on PrEP adherence. Methods HPTN 082 was an open‐label PrEP study among AGYW (ages 16 to 24) in Zimbabwe and South Africa from 2016 to 2018. Depressive symptoms were measured at enrolment and Weeks 13, 26 and 52, using the 10‐item Center for Epidemiologic Studies scale; a score ≥10 is indicative of elevated depressive symptoms. PrEP adherence was defined as any detectable tenofovir diphosphate (TFV‐DP) levels. Group‐based trajectory modelling was used to model longitudinal patterns of depressive symptoms. We assessed psychosocial and behavioural predictors of depressive symptom trajectory membership (e.g. PrEP stigma, intimate partner violence [IPV], sexual behaviour). We modelled associations between (1) group trajectory membership and PrEP adherence at Week 52 and (2) time‐varying depressive symptoms and PrEP adherence through follow‐up. Results At enrolment, 179 (41.9%) participants had elevated depressive symptoms. Group‐based trajectory models revealed persistent elevated depressive symptoms in 48.5%, declining symptoms in 9.4% and no consistent or mild depressive symptoms in 43.3%. AGYW who engaged in transactional sex, reported IPV, or had traumatic stress symptoms were more likely to be assigned to the persistent elevated symptom group compared with the consistent no/mild symptom group (Wald test p‐value all