Predictors and consequences of HIV status disclosure to adolescents living with HIV in Eastern Cape, South Africa: a prospective cohort study

Introduction: The World Health Organization recommends full disclosure of HIV-positive status to adolescents who acquired HIV perinatally (APHIV) by age 12. However, even among adolescents (aged 10–19) already on antiretroviral therapy (ART), disclosure rates are low. Caregivers often report the child being too young and fear of disclosure worsening adolescents’ mental health as reasons for non-disclosure. We aimed to identify the predictors of disclosure and the association of disclosure with adherence, viral suppression and mental health outcomes among adolescents in sub-Saharan Africa. Methods: Analyses included three rounds (2014–2018) of data collected among a closed cohort of adolescents living with HIV in Eastern Cape, South Africa. We used logistic regression with respondent random-effects to identify factors associated with disclosure, and assess differences in ART adherence, viral suppression and mental health symptoms between adolescents by disclosure status. We also explored differences in the change in mental health symptoms and adherence between study rounds and disclosure groups with logistic regression. Results: Eight hundred and thirteen APHIV were interviewed at baseline, of whom 769 (94.6%) and 729 (89.7%) were interviewed at the second and third rounds, respectively. The proportion aware of their HIV-positive status increased from 63.1% at the first round to 85.5% by the third round. Older age (adjusted odds ratio [aOR]: 1.27; 1.08–1.48) and living in an urban location (aOR: 2.85; 1.72–4.73) were associated with disclosure between interviews. There was no association between awareness of HIV-positive status and ART adherence, viral suppression or mental health symptoms among all APHIV interviewed. However, among APHIV not aware of their status at baseline, adherence decreased at the second round among those who were disclosed to (N = 131) and increased among those not disclosed to (N = 151) (interaction aOR: 0.39; 0.19–0.80). There was no significant difference in the change in mental health symptoms between study rounds and disclosure groups. Conclusions: Awareness of HIV-positive status was not associated with higher rates of mental health symptoms, or lower rates of viral suppression among adolescents. Disclosure was not associated with worse mental health. These findings support the recommendation for timely disclosure to APHIV; however, adherence support post-disclosure is important. The study traced all ALHIV (aged 10–19 at baseline) who had initiated ART from all 52 ART clinics in a large urban, peri‐urban and rural district in Eastern Cape, South Africa [33]. Adolescents were identified via paper and computerized records and traced home. At baseline (2014–2015), 1046 ALHIV were recruited, representing 90% of the 1176 patient records identified. These ALHIV were followed‐up over a 4‐year period for three rounds of data collection (Round 1: 2014–2015, Round 2: 2016–2017 and Round 3: 2017–2018). Quantitative interviews were self‐administered using standardized questionnaires on tablet devices. Available viral load data were extracted from participants’ clinical records in rounds 1 and 2 and linked to their questionnaire data [33, 34]. This analysis was restricted to APHIV (Figure 1) determined by ART initiation age ≤10 years and validated using supporting evidence, such as history of parental death, maternal HIV status and self‐reported sexual history [33, 35]. Flow diagram of study and analytic sample. The outcome for the first objective was learning one’s HIV‐positive status (disclosure). Mental health symptomology (anxiety, depression and suicidality) and HIV treatment outcomes (ART adherence and viral suppression) were the outcomes for the second objective, while mental health symptomology and ART adherence were outcomes for the third objective. At baseline, awareness of HIV status was assessed firstly through clinic records and healthcare worker interviews, and then with primary caregivers during the consent process. In cases of discrepancies, interviewers asked adolescents if they knew what their illness was, if they had ever tested for HIV and if they knew what their medication was [36]. Adolescents were considered fully aware if they reported both knowledge of living with HIV and knowledge of their medication being ART used to treat HIV. For APHIV unaware at baseline, awareness was reassessed at subsequent study rounds from primary caregivers and adolescents during the consent process. Adolescents unaware of their HIV‐positive status were asked about “illness” and “medication” as opposed to “HIV” and “ART” in study questionnaires. Disclosure was defined as being unaware of one’s HIV‐positive status at baseline or the second round and being aware at the subsequent study round. Anxiety symptoms in the past month were assessed using a 14‐item abbreviated version of the Children’s Manifest Anxiety Scale‐Revised [37]. This scale, which has previously been validated in studies among children living with HIV [38], included “no” and “yes” responses to the experience of each symptom, coded as “0” and “1” with a total score range of 0–14. Depression symptoms in the past 2 weeks were assessed using the Child Depression Inventory (short form) 10‐item version [39]. This scale, which has also been used and validated in other SSA populations [40, 41, 42, 43], had a 3‐point Likert‐type scale ranging from 0 to 2 with a total score range of 0–20. Suicidality symptoms in the past months were assessed using the Mini International Psychiatric Interview for Children and Adolescents suicidality and self‐harm subscale [44]. This 5‐item scale, which has been validated in developed world populations and adapted in SSA settings [45, 46, 47], included “no” and “yes” responses to the experience of each symptoms, coded as “0” and “1” with a total score range of 0–5. All symptoms in these measures had equal weight. Due to the small number of participants endorsing the most severe symptoms, we created binary variables for any symptoms versus none on each of three scales. An adapted version of the standardized Patient Medication Adherence Questionnaire was used to assess self‐reported ART adherence in the past week, alongside measures developed in Botswana [48, 49]. Adherence was defined by reporting currently taking ART and not having missed any doses in the past 7 days (including weekdays and weekend) [34]. We included the closest viral load results that were no more than 3 months before or 1 year after the questionnaire interview dates for the respective study rounds. Viral suppression was defined as viral load <1000 copies/ml. The main explanatory variable of interest for the mental health and HIV treatment outcomes was awareness of HIV‐positive status, as described above. Other control variables were age, age at ART initiation, sex, dwelling type, orphanhood status, relationship with primary caregiver and household poverty. Household poverty was assessed by measuring access to the top eight socially perceived necessities for children as defined by the Centre for South African Social Policy [50]. Adolescents were classified as living in poverty if they reported not having access to all eight necessities. We also included measures of abuse (physical and emotional) and stigma (anticipated and secondary) as control variables. Physical and emotional abuse were measured using items from the UNICEF Measures for National‐level Monitoring of Orphans and Vulnerable Children [51]. Anticipated stigma was measured using two items from the ALHIV‐Stigma Scale, which assessed adolescents’ views of the community's perception towards HIV and has been used previously among ALHIV in SSA [52]. Secondary stigma due to HIV in families/households was measured using the 6‐item Stigma‐By‐Association scale, which has been validated in South Africa [53]. Adolescents were categorized as having experienced physical or emotional abuse and anticipated or secondary stigma if they self‐reported at least one experience of these in the past year. Characteristics of study participants overall and by awareness of HIV status in each round were summarized using means, standard deviations, median, interquartile ranges and proportions. Differences between participant characteristics by awareness of HIV status and availability of viral load results at all study rounds were calculated using t‐tests for continuous variables and chi‐square tests for categorical variables. Second, among APHIV who were unaware of their HIV status at rounds 1 and 2, we used random‐intercepts logistic regression to identify factors associated with disclosure between rounds. The outcome was learning one's HIV status at round 2 or 3, and the explanatory variables were demographic, psychological and social factors at the survey prior to disclosure. Individual‐level random intercepts were used to account for the repeated observations of the same individuals. Variables identified a priori to be associated with disclosure, such as age, age at ART initiation, sex, dwelling location, caregiver relationship and orphanhood status [12, 14, 15, 54, 55], and the study round, were included in a multivariate regression model. Third, to assess if awareness of HIV‐positive status was associated with self‐reported ART adherence or poor mental health symptoms, data for all APHIV who were interviewed at any of the three rounds were analysed. Analysis of the association between awareness of HIV‐positive status and viral suppression was restricted to APHIV with viral load results. Logistic regression with individual‐level random intercepts was used to estimate the odds of ART adherence, viral suppression and reporting any symptom of depression, anxiety and suicidality. We adjusted for potential confounders identified from our conceptual framework and the study round in a multivariate random‐intercepts logistic regression model. A sensitivity analysis was conducted using only those interviewed at all three rounds. Lastly, again among those who were unaware of their HIV status at rounds 1 and 2, we analysed whether there was a differential change in reporting any mental health problems symptom or ART adherence between study rounds for APHIV who were disclosed to versus those who were not. We did not examine the differential change in viral suppression between disclosure groups due to the paucity of viral load results and wide interval between result dates and interviews. We specified the following logistic regression model to estimate if the odds of reporting any anxiety, depression or suicidality symptom, or adherence between study rounds (rounds 1–2 and 2–3), was different between those who learnt their HIV status and those who did not: where Y represents our mental health outcomes or ART adherence, time is a dummy variable indicating round 1 or 2 and awareness is a dummy variable indicating awareness of status at round 2. β 3 indicates the difference between the log‐odds ratio comparing round 1 versus 2 in those who learnt their status at 2 and the log‐odds ratio comparing round 1 versus 2 in those who did not. We reported exponentiated β 3 estimates (both crude and adjusted for factors hypothesized to be associated with mental health symptoms and adherence from our conceptual framework [Figure S1]). We considered p‐values ≤0.05 as statistically significant, and all p‐values are two‐sided. Analyses were conducted in R version 3.6.1 [56]. Ethical approval for the study was granted by the Institutional Review Boards at the Universities of Cape Town (CSSR 2013/4) and Oxford (SSD/CUREC2/12‐21). Provincial approval was obtained from the Eastern Cape Departments of Education and Health and participating health facilities to conduct the study and access medical records. Written informed consent for the interviews and to access clinical records was obtained from participants and their primary caregivers. Ethical approval for the secondary analyses of study data was obtained from the Imperial College Research Governance and Integrity Team (20IC6451).