From protocol to practice: evaluating the real-world effects of decentralising HIV care on adolescents’ care outcomes and experiences in South Africa

Background: To rapidly scale up HIV care, several sub-Saharan African countries have decentralised antiretroviral therapy (ART) and HIV care delivery. Decentralisation triages patients by intensity of care need, with clinically “stable” patients being down-referred to primary care clinics from secondary or tertiary hospital settings. Compared to other age groups, adolescents demonstrate both high rates of new HIV infections and the worst care outcomes once initiated on ART. However, the effects of decentralising HIV care remain poorly characterised for adolescents, despite its extensive and continuing rollout. Therefore, this thesis aims to characterise the current reality of HIV care delivery for adolescents in South Africa’s decentralised HIV care system, in order to identify potential ways to tailor services for improving care outcomes in this vulnerable population. Objectives: (1) Summarise current evidence for effects of decentralising ART delivery for youth health outcomes; (2) Assess adolescent attainment of HIV care targets in South Africa’s decentralised healthcare system; (3) Characterise adolescent experiences of transition out of paediatric HIV care in South Africa’s decentralised healthcare system Methodology: This thesis consists of a systematic review of decentralising HIV care for youth in low- and middle-income countries (Paper 1), and an epidemiological study of HIV care outcomes and experiences for ART-initiated adolescents (n=1080) in South Africa’s decentralised HIV care system (Papers 2-3). Primary data analyses for Papers 2-3 were based on a community-traced cohort (2014-2018) of all adolescents who had ever initiated ART in 52 public healthcare facilities of a health sub-district of the Eastern Cape. This candidate was a lead Co-Investigator of this longitudinal cohort study. At each of the three study waves, adolescents completed questionnaires about their health on digital tablets, administered by research assistants. In parallel, participants’ clinical records were extracted in two waves (2014-2015 and 2016-2017) from electronic and paper-based patient files in healthcare facilities. At each facility, healthcare staff completed a semi- structured interview that provided a facility “profile” of available services. Results: Paper 1: A systematic review was conducted in accordance with PRISMA guidelines to critically appraise evidence on the effects of decentralising ART delivery on health outcomes for adolescents and young people (10-24 years old) in low- and middle-income countries. An extensive search was conducted through 12 electronic databases, contacting relevant experts, and hand-searching references. Implementation fidelity, study quality, and risk of bias were assessed using the TIDieR checklist, CASP checklists, and ROBINS-I tool, respectively. Of 5302 records identified, 11 studies were potentially eligible but required age disaggregation. Only 2 studies could provide age- disaggregated data, but meta-analysis was not possible due to limited data availability and heterogeneity in implementing decentralisation. Results from these 2 studies suggest the potential for at least equivalent attrition outcomes within decentralised care, but both studies faced significant selection and allocation bias. This review highlighted three key limitations in the current evidence base: (1) a general paucity of evidence on decentralisation for adolescents and youth in resource-limited settings, (2) a critical gap in recent evidence (post-2011) on decentralising HIV care, and (3) the focus of evidence on highly resourced models of HIV care delivery, rather than public care models. This paper is published in Global Health Action. Paper 2: Using clinical records through 2017, this study evaluated adolescent progression along an extended HIV care cascade in South Africa’s decentralised public HIV care system, including operational and care-terminating outcomes. Mortality and loss to follow- up recorded in clinical records were adjusted for unreported deaths and “silent” care transfers. Sociodemographic and treatment-related predictors were tested through sequential multivariable logistic regressions. Predicted probabilities for the effects of predictors were estimated by sex and mode of infection. Rates of mortality and loss to follow-up in the total cohort were 3.3% and 16.9%, respectively. Although almost all participants with available clinical records had at least one recorded viral load, only 51.1% of these adolescents had viral loads from the past 12 months. Having a recent viral load was associated with experiencing decentralised care and longer time on ART. The protective effect of decentralised care was greater for female and sexually infected adolescents. At most recent available viral load, 58.4% of adolescents were fully virally suppressed. Among the total cohort, only 23.2% of adolescents were fully virally suppressed in the past 12 months. Younger age and longer time on ART were associated with full viral suppression. Thus, although overall rates of viral load coverage are high, adherence to routine testing guidelines and viral suppression remain low for adolescents living with HIV in South Africa. This paper is under review at BMC Infectious Diseases. Paper 3: This study identified adolescents’ pathways in HIV care across facility care types and levels in South Africa’s decentralised HIV care system, including transitions out of paediatric care. Associations between transition pathways and care outcomes were tested in sequential multivariable regressions. Thematic analysis of clinic-level questionnaires identified transition support available at facilities. In the total cohort, 57.8% had initiated ART in paediatric care, and 20.4% had transitioned out of paediatric care, with median age at first transition of 14 years. Among those who transitioned, two main pathways were identified: down-referral transition to generalised primary care clinics (56.7%) and classical transition to specialised adult HIV care (43.3%). Across pathways, 27.3% experienced cyclical transition, with repeated movement between paediatric and non-paediatric care. Experiencing down-referral transition was protective against viral failure, and median post-transition viral load change was not clinically significant. Healthcare providers at hospitals and community health centres described informal “protocols” used to mitigate risk of negative post-transition care outcomes for adolescents. This study suggests a new, contextually relevant model for adolescent transitions out of paediatric HIV care, beyond models found in high-income countries. This paper is published in the Journal of Acquired Immune Deficiency Syndromes. Conclusions: This thesis provides new and urgently required evidence on the reality of HIV care outcomes and experiences for ART-initiated adolescents in sub-Saharan Africa. This thesis highlights adolescents’ high level of inter-facility mobility in South Africa’s decentralised HIV care system, as well as low rates of recent viral load availability and viral suppression in clinical records. Further, this thesis identifies multiple pathways of transition out of paediatric HIV care in South Africa. Down-referral transition to generalised primary care clinics, a result of decentralising HIV care, was associated with viral suppression and facilitated by informal protocols developed by healthcare providers. Findings suggest that decentralising HIV care has substantially transformed HIV care experiences for adolescents, including greater agency in their care-seeking. This unique context requires a shift in understanding sub-Saharan adolescent HIV care beyond current models from high-resource settings. With increased mobility across facilities and care levels, it is crucial to ensure that health information systems accurately reflect adolescents’ current clinical status. Thus, this thesis highlights both the need for structural interventions to improve clinical data monitoring and the potential for feasible protocols to mitigate risk of negative care outcomes for adolescents as they approach adulthood. Still, further studies with greater longitudinal coverage are required to confirm the dynamic and long-term effects of decentralising HIV care.