Menu
Menu
Menu
Menu
Introduction: The Sustainable Development Goals (SDGs) present a groundbreaking global development agenda to protect the most vulnerable. Adolescents living with HIV in Sub-Saharan Africa continue to experience extreme health vulnerabilities, but we know little about the impacts of SDG-aligned provisions on their health. This study tests associations of provisions aligned with five SDGs with potential mortality risks. Methods: Clinical and interview data were gathered from N = 1060 adolescents living with HIV in rural and urban South Africa in 2014 to 2015. All ART-initiated adolescents from 53 government health facilities were identified, and traced in their communities to include those defaulting and lost-to-follow-up. Potential mortality risk was assessed as either: viral suppression failure (1000+ copies/ml) using patient file records, or adolescent self-report of diagnosed but untreated tuberculosis or symptomatic pulmonary tuberculosis. SDG-aligned provisions were measured through adolescent interviews. Provisions aligned with SDGs 1&2 (no poverty and zero hunger) were operationalized as access to basic necessities, social protection and food security; An SDG 3-aligned provision (ensure healthy lives) was having a healthy primary caregiver; An SDG 8-aligned provision (employment for all) was employment of a household member; An SDG 16-aligned provision (protection from violence) was protection from physical, sexual or emotional abuse. Research partners included the South African national government, UNICEF and Pediatric and Adolescent Treatment for Africa. Results: 20.8% of adolescents living with HIV had potential mortality risk – i.e. viral suppression failure, symptomatic untreated TB, or both. All SDG-aligned provisions were significantly associated with reduced potential mortality risk: SDG 1&2 (OR 0.599 CI 0.361 to 0.994); SDG 3 (OR 0.577 CI 0.411 to 0.808); SDG 8 (OR 0.602 CI 0.440 to 0.823) and SDG 16 (OR 0.686 CI 0.505 to 0.933). Access to multiple SDG-aligned provisions showed a strongly graded reduction in potential mortality risk: Among adolescents living with HIV, potential mortality risk was 38.5% with access to no SDG-aligned provisions, and 9.3% with access to all four. Conclusions: SDG-aligned provisions across a range of SDGs were associated with reduced potential mortality risk among adolescents living with HIV. Access to multiple provisions has the potential to substantially improve survival, suggesting the value of connecting and combining SDGs in our response to paediatric and adolescent HIV.
The study was designed in partnership with the South African National Departments of Health, Social Development and Basic Education and National AIDS Council, UNICEF, PEPFAR‐USAID, and Pediatric Adolescent Treatment for Africa (PATA). We conducted a total population sampling survey of adolescents living with HIV in a health district of South Africa’s Eastern Cape, a province characterized by poor infrastructure and limited service access. The study included ART‐initiated adolescents, irrespective of whether they were engaged in care at the time of the survey. In order to do this we targeted one urban/peri‐urban/rural health district and visited all health facilities providing HIV services (hospitals, community health centres, and primary care clinics). Of these, all 53 health facilities that provided ART for adolescents were sampled. In each health facility, we went through paper and computerized records to identify all adolescents aged 10 to 19 that had ever initiated ART. From March 2014 to September 2015, adolescents were traced to their communities and invited to participate in a study of young people, health and social services in South Africa. Of 1202 eligible adolescents, 90.1% (n = 1060) were interviewed. 3.7% of adolescents were untraceable, primarily due to inaccurate names and addresses in clinic files. 0.9% were unable to be interviewed due to very severe developmental disability, and 4.1% refused to participate (either adolescent or primary caregiver), 1.2% were excluded for other reasons. In order to prevent stigma, HIV was not mentioned in recruitment, and neighbouring adolescents were also interviewed (n = 467, not included in analyses). Community‐based tracing resulted in a sample that included high proportions of ART‐initiated adolescents who do not regularly attend facilities, default from treatment, or missed recent appointments (44% of the sample). Ethical approval was given by the Universities of Cape Town (CSSR 2013/4) and Oxford (SSD/CUREC2/12‐21), the Provincial Departments of Health and Education and ethical review committees of participating facilities. Full voluntary informed consent was obtained from both adolescents and their primary caregivers, and included interviews and access to clinical records. Given low levels of literacy, consent procedures were additionally read aloud. The study did not use financial incentives for participation, but all adolescents received a small gift pack, refreshments and a certificate of participation. Participants were informed that all responses were confidential except in the case of risk of harm to the adolescent or someone else. Where participants or caregivers reported abuse, recent attempted suicide, active untreated TB or other serious risks, immediate referrals and follow‐up were made to health, police and social services, including taking the participant to a health facility when health or abuse‐related cases were reported. 69 high‐risk referrals were made. These included 38 for severe food insecurity, 22 for psychosocial or family issues, 6 for sexual abuse, 5 for suicide attempts, 5 for physical abuse, 4 for extreme illness and 2 for drug use. Clinical records identified most recent viral load measures, diagnoses and treatment of tuberculosis. With the support of interviewers, adolescents participated in a tablet‐based questionnaire lasting approximately 90 minutes. Standardized measures were translated into Xhosa, back‐translated and provided in both Xhosa and English according to participant choice. The questionnaire was designed in collaboration with our Teen Advisory Group of 20 adolescents to be engaging and adolescent‐friendly, and was pre‐piloted with a further 25 adolescents living with HIV in the Eastern Cape. Interviewers were trained in working with HIV‐affected adolescents and their families. Potential mortality risk for adolescents living with HIV was assessed using two key predictors: treatment failure and symptomatic pulmonary tuberculosis 16, 17. Antiretroviral treatment failure was operationalized as virologic suppression failure in the past two years (defined as viral load 1000 +/ml 18). Viral load data were extracted from clinic records, however limited health service capacity in the Eastern Cape meant that viral load tests were not consistently performed or recorded: only 673 adolescents (64.5% of the sample) had any viral load recorded in their patient files. Furthermore, only 412 adolescents (38.6%) had a viral load test in the past two years. Viral load measurements that had been taken more than two years prior to the study were excluded. In the very few cases where adolescents had more than one viral load taken in the past two years, the most recent viral load was selected. Symptomatic pulmonary TB – the leading cause of death among HIV‐infected populations in the region – was unable to be assessed through clinical records as almost none reported any TB testing or results. Consequently, TB was measured as 1) adolescent‐reported TB diagnosis without subsequent treatment or 2) self‐reported current symptoms of TB using WHO diagnostic criteria, validated among 8979 participants in Zimbabwe against two sputum specimens 19, 20. This study identified that positive predictive value (PPV) was highest for the symptom combination of chronic cough and weight loss (sensitivity 72.9%, specificity 85%, PPV 11.4). Negative predictive value (NPV) was highest for the symptom combination of any cough, drenching night sweats, and weight loss (sensitivity 75%, specificity 82.4%, NPV 99.2). Area under ROC curves was estimated to provide a summary measure of diagnostic accuracy, with AUC of 0.81 for HIV‐positive TB. We required fulfilment of criteria for both positive and negative predictive values to maximize precision for each case identification of TB. We identified provisions aligned with SDG targets that may have potential to improve adolescent outcomes, using systematic reviews and studies of risk and protective factors for child, adolescent and adult HIV‐outcomes 21, 22, 23. First, adult studies suggest that poverty may present a barrier to HIV healthcare access through lack of transport and food 24, 25, 26 and that social protection provision may have potential to improve retention in care 27, 28. Given the overlap between ending poverty, food insecurity and social protection, we combined provisions aligned to SDGs 1 and 2 (“End poverty” and “end hunger”). These were operationalized as access to all of the following: eight basic necessities for children as endorsed by over 80% of the population in the nationally representative SA Social Attitudes Survey, including “3 meals per day” and “free school;” and access to a child‐focused grant (child support or foster child grant) in the household. These provisions align to SDG targets 1.1 (“end extreme poverty”), 1.2 (“reduce […] poverty in all its dimensions”), 1.3 (“Implement nationally appropriate social protection”) and 2.1 (“end hunger”). Second, studies of HIV‐affected families find improved outcomes for children with healthy and surviving caregivers 29, 30. Consequently, we identified a surviving and healthy caregiver as a provision aligned with SDG 3 (“Ensure healthy lives”). This was measured through adolescent self‐report of having a surviving parent or caregiver taking care of them at home, who was not suffering from chronic illness. This provision is aligned with SDG targets 3.1 (“reduce maternal mortality”) and 3.2 (“end the epidemics of AIDS, tuberculosis, malaria and neglected tropical diseases”). Adolescent health was not included as a potential provision due to high risk of confounding with the study outcomes of viral failure and TB. Third, studies suggest that caregiver capacity to provide sustained household income can lead to improved child outcomes in the context of HIV 31. Consequently, we included the potential provision of household employment, as aligned with SDG 8 (“Employment… for all”). This was operationalized as adolescent report of at least one employed person in their household. Fourth, there is now substantial evidence demonstrating that violence has negative health impacts 32, and that protection from abuse may improve adolescents’ psychological and physical capacity to engage with healthcare. Thus, we included a provision aligned with SDG 16 (“Peace and justice,” specifically “end abuse… and all forms of violence against… children”). Never having been physically, emotionally or sexually abused was aligned with targets 16.1 (“significantly reduce all forms of violence”) and 16.2 (“end abuse, exploitation, trafficking and all forms of violence against…children”), and measured through adolescent self‐report using the UNICEF Measures for National‐level Monitoring of Orphans and Other Vulnerable Children and the Juvenile Victimization Questionnaire (JVQ), both used previously in South Africa 33. Age (10 to 14 versus 15 to 19), gender, and urban/rural location were measured using items based on South Africa’s Census 34. Mode of HIV‐infection was measured using clinical algorithms, with adolescents coded as vertically infected if they had initiated ART prior to age 12 or if they had been on ART for more than 5 years, based on the year of widely available ART access in the study area 35, 36. Time on ART was measured via self‐report. Based on evidence of higher defaulting risk in the first year of treatment 37, a dummy variable was computed to differentiate between 1: “more than one year on treatment” and 0: “one year or less.” Analyses were conducted in six stages in SPSS 22 and STATA 14. First, eligible participants included in the study (90.1%) were compared to those excluded (not found or refused participation) on known socio‐demographic characteristics (age, gender and urban/rural household location) using χ2 tests. Second, frequency distributions for high mortality risk among adolescents living with HIV, each of the hypothesized SDG‐aligned provisions, socio‐demographic and HIV‐related covariates were reported. Third, to test initial associations of each SDG‐aligned provision against potential mortality risk among adolescents living with HIV, bivariate logistic regressions were run. Fourth, a multivariate logistic regression was run, including all SDG‐aligned provisions simultaneously and controlling for all potential covariates. All potential two‐way and three‐way interactions were tested in logistic regressions. Fifth, to compute combined effects of multiple SDG‐aligned provisions the following steps were taken, using provisions found to be significantly associated with reduced mortality risk in Stage 3. First, categorical principal components analysis established that all provisions loaded onto a first component (eigenvalue 1.2, 24.4% of total variance). Loadings for each provision were >0.35. To assess combined effects of multiple SDG‐aligned provisions, a summative index was computed, weighted by the respective component loadings. This weighted summative index was highly correlated with a simple summation of the unweighted dichotomies (Spearman’s rho, 0.961; p < 0.001), and therefore, for ease of interpretation, the unweighted scale was used. Sixth, a marginal effects model was run to assess predicted probabilities of high mortality risk by combined effects of multiple SDG‐aligned provisions, holding all socio‐demographic and HIV‐related co‐factors at mean values. This was plotted with 95% confidence intervals.
N/A
