Browsing by Author "Zhou, Siyanai"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Open Access
    ART adherence trajectories and correlates of treatment outcomes among adolescents in the Eastern Cape Province of South Africa
    (2025) Zhou, Siyanai; Knight, Lucia; Toska, Elona
    Background: Sustained adherence to antiretroviral treatment (ART) is essential for viral suppression among adolescents living with HIV. In turn, suppressed viral load reduces the risk of onward transmission, HIV-related morbidity, and mortality, optimising overall health and well-being. Yet, adherence to ART among adolescents living with HIV (ALHIV) remains suboptimal compared to children and adults. Adherence measurement varies widely with no gold standard, and composite assessments such as longitudinal self-reported adherence have not been thoroughly evaluated among ALHIV in low-resource settings. In sub-Saharan Africa, evidence on longitudinal sustained ART adherence and its impacts on subsequent HIV treatment outcomes for ALHIV remains scarce. Prior studies on ART adherence—largely cross-sectional—have relied on traditional aggregate methods, dichotomising adolescents as adherent versus non-adherent which is insufficient to capture variations in adherence over time. Longitudinal ART adherence trajectories may aid the assessment of sustained long-term adherence and its impact on HIV treatment outcomes among ALHIV. To address these gaps, this PhD examined the validity of longitudinal self-reported measures of ART adherence among ALHIV in South Africa. It used group-based trajectory models (GBTMs) to identify ART adherence trajectories and examined their association with HIV treatment outcomes—viral suppression and mortality. It further explored factors contributing to distinct adherence trajectories, and the impact of psychosocial and structural provisions on long-term HIV treatment outcomes among ALHIV in South Africa. Methods: This PhD is an analysis of a three-wave longitudinal cohort study of 1046 ALHIV and 473 HIV-negative peers aged 10-19 years at baseline, in the Eastern Cape province of South Africa. From 2014 to 2018, participants completed questionnaires that included their socio-demographic information and self-reported adherence at all three visits. All participants and their caregivers gave informed consent to participate. In parallel, routine viral load data were extracted from physical and electronic patient files in health facilities, and the National Health Laboratory Services (NHLS) of South Africa (2014-2019). Viral load laboratory test results for 75.2% of the participants were obtained. Mortality was ascertained through community tracing and reporting from 2016 to 2022. All analyses in this thesis used the longitudinal cohort data to address the specific objectives. Statistical analyses included mixed effects logistic regression, group-based trajectory modelling, and path analysis. Findings: The first analysis (Chapter 4) explored the validity of five longitudinal self-reported adherence measures to screen for non-adherence and identify ALHIV with elevated viral load (>1000 copies/mL). Self-report measures included any missed dose in the past 3-days, past-week, past-month, days missed in the past-month, and any past-year missed clinic appointment and were all significantly associated with elevated viral load. The individual sensitivity of each measure varied from 79.5% to 91.6%, and positive predictive values were above 75%. Using the five self-reported adherence measures and group-based trajectory modelling, four adolescent adherence trajectories were identified (Chapter 5), namely ‘consistent (49.8%), ‘low start and increasing' (20.8%); ‘gradually decreasing' (23.5%), and ‘low and decreasing (5.9%) adherence. Compared to the consistent adherence trajectory, membership in each of the three inconsistent adherence trajectories was significantly associated with lower adjusted odds of durable viral suppression—having suppressed viral load (<1000 copies/ml) at two or more consecutive study waves. Complementary mortality analysis (Chapter 6) showed that all-cause mortality rates differ across the four adherence trajectories, and sustained adherence to ART was associated with lower rates of mortality among ALHIV. The analysis found that ALHIV experience higher all-cause mortality than their HIV-negative peers, despite ART rollout in South Africa. Guided by the socio-ecological model (Chapter 7), pathways linking barriers contributing to the distinct longitudinal adherence trajectories were established using path analysis. Experiencing mental health symptoms, medication side effects, internalised HIV stigma, food insecurity, longer clinic travel time (>1 hour), and witnessing domestic violence or conflict were associated with inconsistent adherence trajectories relative to the ‘consistent adherence' trajectory. More importantly, the pathways linking co-occurring barriers, and mental health symptoms to longitudinal adherence were unique for each trajectory or ALHIV sub-group. The final analysis (Chapter 8) revealed that access to government-provided social protection and family psychosocial support was associated with greater improvements in long-term ART adherence and viral load outcomes. Conclusions: This PhD documents multiple ART adherence trajectories, low rates of viral suppression, and high all-cause mortality among ALHIV in South Africa. Self-reported adherence shows promise for low-cost and relatively easy-to-administer alternative measures to ensure timely identification of poor adherence to ART among adolescents, particularly in settings where virologic monitoring is limited. ALHIV displayed distinct adherence behaviour patterns over time, which provides an opportunity to identify and target specific sub-groups of ALHIV in need of adherence support interventions. The multiple barriers identified and their pathways to longitudinal adherence trajectories present another opportunity to tailor care to specific groups of ALHIV, possibly informing differentiated service delivery to different sub-groups of ALHIV. Government-provided social protection and strategies to address psychological well-being can be key add-on support provisions to enhance the effectiveness of ART treatment among ALHIV in low-resource settings. Overall, this PhD demonstrates the need to shift from a one-size-fits-all model of care to customised HIV care and combine biomedical with psychosocial and structural interventions to address the needs of distinct groups of ALHIV.
  • Thumbnail Image
    Item
    Open Access
    Ethnic migration patterns in South Africa: A case of Zimbabwean immigrants
    (2018) Zhou, Siyanai; Moultrie, Thomas
    Zimbabwean-born migrants form the largest proportion of the foreign-born population in South Africa; however, relatively little is known about both the international and internal migration behaviour of this population group. This study reviews several migration theories, and existing evidence on Zimbabwean-born migration to South Africa and uses the 2011 Census and 2016 Community Survey data for South Africa to explore both provincial and metropolitan migration patterns of Zimbabwean migrants. The study also seeks to identify characteristics of migrant internal movers and how these are associated with their propensity to migrate. To accomplish this, two modelling approaches are used, namely: the multiplicative component model (TODA) and the logistic regression model. The first modelling approach describes the origin-destination specific migration patterns of this group of migrants. The results from this analytic model show that Zimbabwean migrants are concentrated in three provinces, namely, Gauteng, Western Cape, and Limpopo; particularly in respective metropolitan areas. Also, post-arrival migration is largely focused on these major receiving provinces. The findings suggest that this pattern of concentration is likely to continue. The second modelling approach examines the inter-provincial migration of Zimbabwean migrants. At the aggregate level, the study results show a degree of selectivity of Zimbabwean immigration by factors such as age, sex, marital status, and education. Also, the results show that post-immigration migration is not necessarily significant to migrants’ redistribution. However, for those who move, the logistic regression analysis reveals that province of residence, age, marital status, and sex significantly impact on migrants’ propensity to move from “pioneer settlements” and their effects are consistent over time. Finally, despite concerns about the quality of the data collected in the 2011 census and 2016 Community Survey, our results show that there is general consistency of migration 'patterns’ between the two datasets.