Scaling a waterfall: a meta-ethnography of adolescent progression through the stages of HIV care in sub-Saharan Africa

Abstract Introduction: Observational studies have shown considerable attrition among adolescents living with HIV across the “cascade” of HIV care in sub-Saharan Africa, leading to higher mortality rates compared to HIV-infected adults or children. We synthesized evidence from qualitative studies on factors that promote or undermine engagement with HIV services among adolescents living with HIV in sub-Saharan Africa. Methods: We systematically searched five databases for studies published between 2005 and 2016 that met pre-defined inclusion criteria. We used a meta-ethnographic approach to identify first, second and third order constructs from eligible studies, and applied a socio-ecological framework to situate our results across different levels of influence, and in relation to each stage of the HIV cascade. Results and discussion: We identified 3089 citations, of which 24 articles were eligible for inclusion. Of these, 17 were from Southern Africa while 11 were from Eastern Africa. 6 explored issues related to HIV testing, 11 explored treatment adherence, and 7 covered multiple stages of the cascade. Twelve third-order constructs emerged to explain adolescents’ engagement in HIV care. Stigma was the most salient factor impeding adolescents’ interactions with HIV care over the past decade. Self-efficacy to adapt to life with HIV and support from family or social networks were critical enablers supporting uptake and retention in HIV care and treatment programmes. Provision of adolescent-friendly services and health systems issues, such as the availability of efficient, confidential and comfortable services, were also reported to drive sustained care engagement. Individual-level factors, including past illness experiences, identifying mechanisms to manage pill-taking in social situations, financial (in)stability and the presence/absence of future aspirations also shaped adolescents HIV care engagement. Conclusions: Adolescents’ initial and ongoing use of HIV care was frequently undermined by individual-level issues; although family, community and health systems factors played important roles. Interventions should prioritise addressing psychosocial issues among adolescents to promote individual-level engagement with HIV care, and ultimately reduce mortality. Further research should explore issues relating to care linkage and ART initiation in different settings, particularly as “test and treat” policies are scaled up.


Introduction
Globally, it has been estimated that approximately 2.1 million adolescents aged 10-19 years are living with HIV, with nearly 85% of this population residing in sub-Saharan Africa (SSA) [1]. Whilst initiatives to improve access to antiretroviral therapy (ART) in low-and middle-income countries (LMIC) contributed to a 30% decline in AIDS-related mortality between 2005 and 2012 [2], adolescents have been left behind. In the same time-period, AIDS-related mortality amongst adolescents increased by 50%, with over 90% of these deaths occurring in SSA [1].
The clustering of adolescent AIDS-related mortality in SSA is largely attributable to an approximate 10-year time-lapse in ART interventions compared to high-income nations [3]. Many of the children perinatally infected in SSA, while prevention of mother-to-child transmission (PMTCT) programme coverage was still low, are now reaching their adolescence [4,5]. However, less than 30% of adolescent girls and 10% of adolescent boys aged 15-19 years in SSA report ever having undergone an HIV test [6]. In addition, the regional adolescent population has been projected to double by 2050, with adolescents making up approximately 46% of the total population [6].
Recent global initiatives have called for adolescent health to be prioritized as young people represent a "triple dividend" investmentof benefit now, to the next generation of adults and to the health of generations to come [7]. At the heart of these initiatives lies the need to prevent and treat infections among this population, making care and treatment for ALHIV a policy priority for many African governments. This places new pressures on health systems and services to meet the needs of a rapidly expanding generation of both diagnosed and undiagnosed adolescents living with HIV (ALHIV) [5,8].
Transition from childhood into adulthood is characterized by rapid physical, mental and sexual development, and in addition to this, ALHIV have complex emotional, psychological and health-related requirements that need to be reconciled to effectively manage their lives with HIV [4]. In SSA, the majority of ALHIV under 14 years of age have been infected perinatally [9] and many are orphaned by one or both parents [10] forcing them into early emotional maturity and adult responsibilities. Furthermore, while some have been on ART since a very young age, others have been living with the virus untreated for over a decade. Without access to ART, those untreated are not expected to survive beyond their teenage years [9]. For those that do survive, great efforts are needed to increase care access and improve retention on ART to reduce AIDS-related mortality and onward horizontal transmission of HIV [9].
Quantitative research indicates that in many settings ALHIV experience higher rates of loss to follow-up and virological failure than both adults and children living with HIV [11][12][13][14]. Geographically discreet qualitative studies in different countries in SSA suggest that stigma, challenges in disclosure and consent policies for testing and treatment are reasons for faltering ALHIV care engagement, however, there has been no published synthesis of these studies. Meta-ethnographic methods enable an in-depth assessment of various health-related research topics [15,16], including factors that influence patient care experiences and medication adherence in LMIC [17,18]. Meta-ethnography aims to expand upon, or develop new interpretations of qualitative data. The "constant comparison" approach is used to determine if themes across varying contexts are reciprocal (complementary concepts) or refutational (contradictory concepts), supporting further hypothesis development [17,[19][20][21]. Our meta-ethnography includes qualitative studies undertaken across the region, in order to understand the most influential issues affecting adolescent initiation of, and retention in, HIV care in SSA. Our synthesis of 10 years of qualitative research conducted since the advent of ART expansion enables a longitudinal line of inquiry, allowing us to assess changes across time, in different countries in SSA and across all stages of the HIV care cascade.

Theoretical frameworks
The HIV care cascade describes progression through HIV diagnosis, care and treatment services and is commonly defined in five stages: (i) HIV diagnosis, (ii) linkage to care, (iii) engagement/retention in care, (iv) ART initiation and (v) viral suppression [22,23]. The cascade has been widely used as a framework to explore reasons for patient attrition between these five stages, with some variations in the definitions for each stage depending on the study design or nature of the data that are available. For example, in relation to stage v, qualitative studies often explore patients' experiences of adherence to antiretroviral therapy, rather than investigating viral suppression per se, which is more commonly measured in studies with a quantitative design [24,25].
For adolescents, traversing the cascade is complicated by the lack of autonomy in care-seeking that characterizes this age-group, including the requirement for guardian accompaniment to access services. Further disruptions in HIV care among adolescents have been noted to occur during the transition from paediatric into adult care [26] or as they transition in and out of PMTCT programmes [27,28]. Such challenges perpetuate the risk of attrition and associated mortality [26,29]. "Engagement" with the HIV care cascade involves the willingness and ability of ALHIV to undergo HIV testing, initiate treatment, adhere to ART and negotiate the linkages between different HIV services and models of care (e.g. paediatric and adult; pre-ART and ART). This engagement with HIV care and treatment is influenced by a multitude of factors including basic service access, economic capability, and awareness of their HIV status [30].
The socio-ecological model (SEM) has been frequently used across HIV literature to illustrate the range of social influences on health behaviours, including care fallout [31,32]. The SEM is particularly useful in exploring health decisions made by individuals in LMIC or communities with high levels of disease stigmatization [33][34][35]. The model illustrates four levels of influence on health decisions ( Figure 1): i) structural factors, such as health systems, policies and underlying poverty; ii) community factors, including social norms around treatment-seeking and beliefs surrounding disease aetiology; iii) family and peer influences, including expectations of behaviours within these social contexts; iv) and individual factors at the model's core, including knowledge, perceptions, self-efficacy and risks associated with care-seeking. Influential factors may act across more than one level of the SEM: for example, family influences on adolescents' care-seeking behaviours may be affected by structural factors such as economic circumstances of the household or consent policies that limit access to services.

Search strategy
For each search, five essential concepts were identified: HIV, adolescents, SSA, barriers /facilitators and each of the five stages of the care cascade. All possible search terms and synonyms for each concept were established and multiple spellings were incorporated. We searched five databases known for having a high volume of social and health-related research or articles specific to adolescents: PubMed, Web of Science, Scopus, Global Health and ADOLEC. Titles and abstracts of identified literature were imported into EndNote and reviewed to eliminate duplicates. References from highly relevant articles and systematic reviews were further checked for additional eligible articles.
Inclusion /exclusion criteria Studies were included if they involved qualitative or mixed methods, were published between January 2005 and March 2016, included an adolescent population in the age range 10-19 years (if the study included children, young adults or adults, it was included so long as the adolescent population was distinct in the results), were conducted in one or more countries within SSA (or included a defined population in SSA) and stated aims that included an assessment of factors influencing adolescent engagement with at least one stage of the HIV care cascade. Only articles published in English were included.

Quality appraisal
The importance of appraising the quality of health research is increasingly recognized, although challenges include the lack of a standardized method or approach [36,37]. The Oxford Critical Appraisal Skills Programme (CASP) criteria is one of the most widely used checklists for qualitative research, enabling an assessment to be made of a study's validity, rigour and relevance [38].
Two researchers (SW, LG) reviewed all articles according to the ten-point scale proposed by CASP and scored them independently. One point was given for each checklist criterion present in the research article and zero points were given for absent criterion. Individual scores were compared and differences deliberated until an agreement was reached. Any scoring not reached in agreement was submitted to a third researcher (AW) for final review.

Thematic analysis
Thematic constructs were noted according to the methods of conducting meta-ethnographic analysis suggested by Noblit and Hare [19] and findings were applied to the SEM. In each article, first-order constructs were identified through study participant quotations in the text. Secondorder constructs were then distinguished from these based on the researchers' perspectives, mainly presented in the discussion sections of research articles. Comparison of first and second order constructs was examined chronologically across studies that explored the same stage of HIV care. Articles that covered more than one stage of care were catalogued into each stage of the cascade sequentially, for a complete analysis of themes specific to that stage. Third order constructs were then defined to contextualize all identified themes across the entire cascade of care, noting reciprocal and highly salient constructs.

Characteristics of included studies
In total, 3089 publications were identified through the search. Following the review of titles and abstracts, 2974 were excluded for not meeting the inclusion criteria, most of which were quantitative studies. The remaining 115 articles were reviewed in full, following which an additional 92 were excluded (the majority for lacking results specific to adolescents). 1 additional relevant study was discovered searching article references. In total, 24 qualitative and mixed methods articles were selected for inclusion in the meta-ethnography ( Figure 2). The eligible articles were catalogued in accordance to HIV care cascade stages (i.e. HIV diagnosis, linkage to care, engagement/retention in pre-ART care, ART initiation and adherence), and within those stages arranged chronologically by year of research. Table 1 presents the characteristics of included studies.
The majority of identified studies were conducted in Southern and Eastern Africa with notable gaps in some high prevalence countries, including Mozambique, Swaziland and Lesotho ( Figure 3).
Out of the 24 studies included in our review, 6 explored issues related to cascade stage one, HIV testing, and 11 Exclusions based on title and abstract:

Quality assessment
Eighteen out of 24 (75%) of the included studies were rated moderate to high quality, scoring 7-10 points on the CASP scale; and none of the studies scored below six points. The CASP criteria least achieved (absent in 13 out of 24 articles (54%)) was a failure to elaborate on the relationship between researchers and study participants (CASP criteria 6).
Themes influencing the engagement of ALHIV in care and treatment services Twelve third-order themes were identified, many of which acted across multiple levels of the SEM (Table 2). Salient themes are presented by order of importance, determined by number of studies they were reported in and instances of overlap within the models.

Broader social support
The inclusion of social services in communities was cited as facilitating engagement at all stages of the HIV care cascade. ALHIV support groups [44,45,48,54,56], home-based care [43,62] and HIV /sexual and reproductive health education through school or outreach programmes [40,45,50,54] were all detailed as beneficial resources.
The presence or lack of support within schools in particular was influential in terms of ART adherence, experiences with discrimination from other students or teachers negatively influenced the ability of ALHIV to carry medication and manage a dosing schedule at school [49,51,53,54], and crowded environments with no privacy, particularly in boarding schools, exacerbated fears surrounding unintended disclosure [49,52,57,58]. However, when selective disclosure to a "safe" teacher or school nurse was obtained, it frequently facilitated good adherence to ART [49,51,54,59].
Adolescent-specific health services ALHIV often faced difficulties in transitioning between paediatric and adult models of care; several studies reported that health services which catered specifically to Traditional medicine used to avoid costs of care (7) Reliance on traditional medicine n = 5 Caregiver decision to treat with TM (12,13) TM practicionerspractitioners claiming to heal HIV (7,11,18) Medication mythology /mistrust of ART (12,18) "I used those herbs and I have even drunk urine because there is someone who came and lied to us that drinking urine can heal HIV." -Kunapareddy, p 391. adolescents' needs positively influenced their progression through the cascade. Services which accommodated "youth-friendly" hours that didn't interfere with school or work were often cited as facilitating access to HIV testing, pre-ART care and ART [41,45,46,51,57,58,61], as were HCWs who had undergone specific training to work with adolescents [44][45][46]57,62]. In addition, adolescents expressed a preference for HIV counsellors who shared similar characteristics to themselves (age range or gender) to avoid feeling that counselling messages were condescending [40,46]. Sexual and reproductive health education incorporated into health services [46] and HCW participation in the disclosure processes [46,50,54,56] were found to positively influence both ALHIV retention in care and adherence.

Adequacy of health systems
Adolescents' linkage to and retention in care were affected by the ability of the health system to provide efficient, confidential and comfortable services. This included the presence or absence of outreach efforts to inform ALHIV about service availability [39][40][41]43,47], pre-and post-test counselling options [39,41,43,47] and employment of sufficient numbers of HCW of various cadre [39,43,45,47]. The infrastructural capacity to avail space within health facilities to ensure privacy [45,62] and the total time required for appointments based on queues and travel to and from the clinic further compounded barriers to both HIV diagnosis and retention in care [39,40,43,45,47,48].

Past illness narratives
Experiences with physical deterioration and severe illness could often facilitate engagement in HIV care, specifically in regard to receiving an HIV diagnosis [47] and adhering to treatment [50,[52][53][54][58][59][60]. Within families and at health facilities, these past illness narratives were used as reminders by caregivers and HCWs to facilitate ALHIV's adherence to ART [46,60]. Conversely, experiencing improvements in symptoms could also hinder adolescents' ART adherence, particularly if their HIV status had not been fully disclosed to them, or they had limited knowledge about their illness and its treatment requirements: several studies identified faltering adherence among ALHIV who had regained physical strength and a "normal" appearance [51,53,57,58,60].

Financial instability
The direct and indirect costs of care and treatment acted as a barrier to HIV care engagement, demonstrated at the family and structural level of the SEM. The cost of transportation to reach appointments [40,46,48,55,56,59] or additional food required to satisfy the nutritional needs of adolescents on ART [48,50,55,60] negatively influenced family willingness or ability to support ALHIV. At the structural level, the absence of available financial assistance to cover the costs of medications for opportunistic infections or ART counselling exacerbated the situation [50,60,61]. Conversely, one study found that the inclusion of economic incentives including food, household provisions or travel reimbursement encouraged linkage to care among ALHIV [44].

Social coping mechanisms
In multiple studies, adolescents described a variety of coping mechanisms to navigate social situations that threatened unwanted disclosure. Such strategies included ALHIV claiming that their drugs, AIDS symptoms or prolonged school absences were the result of an alternate condition, such as heart problems or malaria [49,50,[52][53][54][55][56]61]. These strategies were also used by caregivers and HCWs to avoid disclosure to younger ALHIV of their true condition at the family and structural SEM levels [51,54,57,58]. Such tactics were utilized to facilitate shortterm ART adherence, despite their apparently limited effectiveness to help ALHIV achieve sustained adherence to treatment.

Future orientation
Future aspirations, and the ability to identify personal goals including marriage, pregnancy or further education were cited in several studies as factors that motivated ALHIV to obtain an HIV diagnosis [39,41,47], to link to HIV care [58] and to adherence to their antiretroviral therapy [53,59]. However, adolescents' fear of having diminished options in the future following a positive HIV diagnosis acted as a barrier to diagnosis and ART adherence [40,41,44,59] Media influence Mass media campaigns were identified as effective in promoting knowledge surrounding HIV risk factors and diagnosis. The HIV testing campaign slogan ALHIV repeatedly referenced was the "Know Your Status" (KYS) campaign, that originated in Lesotho [63]. The "KYS" slogan and associated statements regarding the moral imperative of testing were recited by adolescents in multiple countries during interviews [39,[41][42][43]47]. However, researchers noted that despite this familiarity with the KYS campaign, rates of testing often remained low [40,41]. Media campaigns were also detailed as a means of counteracting community stigma towards ALHIV that undermined HIV service enrolment [40,61].

Reliance on traditional medicine
The use of traditional medicine was identified in the SEM at the family and community levels as a barrier to linkage to HIV care and to ART adherence. Reliance on traditional medicine for health care delayed the initiation of ART [52,57,60], and was favoured at times by caregivers [53,55] with reduced costs cited as justification for their use by economically strained families [60]. Lack of community HIV and ART education also contributed to the perpetuation of local disease mythologies associated with the use of traditional medicine to treat HIV [57,60].

Discussion
Our meta-ethnographic approach to the analysis of qualitative literature published across varying contexts in SSA revealed striking thematic similarities across different countries and over time. Stigma, family support, self-efficacy and social support were common themes resonating across multiple settings in SSA and across the 10-year period of inquiry. The degree of consistency in these findings across diverse settings suggests that similar interventions to improve the uptake of HIV care among ALHIV may be relevant in many SSA countries. Out of 24 selected articles, only seven investigated multiple stages of the HIV care cascade. Most of the studies focused on discreet aspects of HIV care and treatment, in particular testing or adherence, leaving stages such as linkage to care and initiation onto treatment under-represented. In order to better understand the impacts of efforts being made to achieve the UNAIDS 90-90-90 targets [64], further investigation of the full spectrum of ALHIV experiences with HIV care is needed.
During the time-period evaluated, the most significant efforts to promote HIV engagement have primarily acted at the structural level: for example, free ART provision, expansion of counselling and testing services and formation of support groups in many clinics. However, it was the threat of social isolation as a result of attitudes towards HIV that emerged as the most influential factor hindering ALHIV progression through all stages of the care cascade. Detrimental experiences with stigma were the most salient theme across the entire time-span, suggesting that structural-level interventions have had limited impact on individual experiences of living with HIV over the last 10 years.
Given that structural interventions alone are unlikely to act at multiple levels of the SEM, the potential of youthspecific services to improve rates of care retention in SSA (a facilitator to HIV care engagement detailed in numerous articles) is ultimately reliant on the presence of mutually improved conditions at the other SEM levels [65]. Their ability to facilitate the movement of ALHIV through the stages of the care cascade is contingent upon caregiver support, community perceptions of services, and experiences of stigma associated with their use. Further, training new or existing HCWs to work with adolescents, expanding services to youth-friendly hours and improving upon the efficiency of appointments all require a sustained and substantial economic commitment [66] which is in turn dependent on international funding priorities and domestic health policies [67][68][69][70].
We noted that stigma was commonly anticipated by ALHIV, often manifest through fear of unwanted disclosure of their positive status after being seen using HIV services or taking medication. These individual level fears pervaded into other SEM levels, where they threatened to deteriorate relationships with loved ones or result in ostracism by the community. The power of such anticipated stigma to negatively affect adolescent care engagement in SSA cannot be overstated, particularly as it compounds the hardships that many ALHIV have already gone through following early parental loss and learning their own positive HIV status. This appears distinct from the dominant manifestations of stigma amongst adults living with HIV in SSA [71,72]. The stigma experienced by many adults living with HIV is inextricably linked to blame or promiscuity [51,73]. Witnessing such enacted stigma demonstrated towards adults may not only reinforce anticipated stigma in ALHIV at the individual level, but also presents a barrier to their care at other levels of the SEM where they require the support of adults. For example, guardian experiences' with enacted stigma may result in failure to provide ALHIV with the consent and accompaniment required to access HIV care in some areas, out of their own unwillingness to be seen utilizing HIV services [51,74,75]. Our findings suggest that interventions to address stigma need to account for the inherent differences between ALHIV and adults living with HIV. Care should be taken to not assume that the same forces driving stigma amongst an adult population resonate amongst adolescents.
In the absence of coordinated support to address treatment challenges among ALHIV, motivational techniques were often applied by caregivers or HCWs, such as the use of past illness narratives to facilitate ART adherence based on ALHIVs' desires to maintain "normalcy" [49,53,57,58,60]. Our meta-ethnography highlights that HCW used the motivation of "staying well" to encourage adherence, insinuating a comparison to a past period of ill health. Further, both ALHIV and their caregivers reported inventing an alternate illness to attribute medication use to, either to avoid disclosure from caregiver to ALHIV or from ALHIV to the wider community of extended family, peers and teachers. The use of past or alternative illness narratives represent makeshift approaches to adherence and disclosure that have been utilized in the absence of broader social support to help ALHIV manage care dilemmas attached to stigma. Recent changes in guidance for ART eligibility to include all who test HIV positive regardless of their immune status or WHO staging [69,76] should reduce the occurrence of "past illnesses" and could therefore change how ALHIV engage with HIV care. Social approaches to managing engagement in HIV care and adherence should encourage early and ongoing disclosure assisted by HCWs, rather than being left to the responsibility of families alone [30,[77][78][79]. This coordinated disclosure will need to place greater emphasis on sustainable facilitators to selfefficacy, such as future orientation, to avoid use of shortterm tactics that fail to recognize the capacity of ALHIV to understand and manage their illness.
The self-efficacy of ALHIV is recognized as a key facilitator to diagnosis, retention in care and adherence to ART. However, its effect is reliant on social conditions such as family environment, presence or absence of peer support and the attitudes of HCWs working with youth. Inclusion of supportive social conditions at different levels of the SEM, such as community support groups, family counselling and HIV /sexual and reproductive health education, has the potential to bolster self-efficacy and motivate ALHIV through the stages of care [80][81][82][83]. In the absence of supportive conditions, ALHIV self-efficacy may fail to develop or result in rebellion against their care management [84] as they gain further independence with age, yet maintain a sense of social isolation and frustration surrounding treatment adaptation.
We concur with others that have taken a critical stance on the utility of a linear model of the HIV care cascade with a proposed end-point of viral suppression [85]. Our systematic review of evidence suggests that such a model is not sufficient to generalize the ongoing and evolving process of managing a lifelong condition for ALHIV. In Figure 4, we illustrate how the reality is a succession of ebbs and flows as ALHIV attempt to "scale a waterfall" to achieve viral suppression. Ethnographic and qualitative descriptions portray movement which requires upward momentum to move through these stages, with ALHIV using facilitators as footholds to manoeuvre against a downward cascade of barriers that persistently threaten to undermine their intentions to engage in HIV care.
Our study is affected by various limitations inherent to qualitative studies and the meta-ethnography design. Adolescents may be prone to social desirability bias in their responses, particularly when discussing behaviours such as ART adherence or clinic attendance. In addition, most of the studies in our analysis adopted a cross-sectional rather than a longitudinal design, which may have been better suited to identify issues related to transition between paediatric and adult models of HIV care. Our search was limited to publications in English which may partially explain the geographic gaps in the research studies that we identified; however, these gaps may also be influenced by differences in research funding across countries. Although our search criteria were broad, some relevant studies may not have been captured by our choice of terms; yet this risk should be minimized by our strategy of searching through the reference lists of relevant articles. Furthermore, due to limited available research exploring gendered differences in adolescent HIV care, our ability to unpick gender-specific factors influencing cascade progression was limited.
Additional qualitative research should be undertaken to explore the risk perceptions, knowledge of HIV transmission, sources of SRH education and experiences of adolescent females with HIV services, not only because they are at higher risk of acquiring HIV [6,64,86,87], but also because their experiences with stigma may profoundly reinforce barriers to testing or ART for PMTCT [80, 88,89].
To our knowledge, this is the first meta-ethnography of research on factors influencing engagement in HIV care among adolescents in SSA that examines treatment progression holistically, by synthesizing studies focused on different aspects of the care cascade. Our use of a socio-ecological lens provides a comprehensive view of the issues influencing HIV care engagement, and enables us to analyse the interplay between individual, family, community and structural factors. The findings generated through this review are timely given the current focus on adolescent health within global health initiatives and also since the adolescent HIV epidemic in the region has peaked, and readiness to test, commence treatment and adhere successfully to ART is essential in reducing transmission rates and preventing drug resistance [77]. Significant efforts are still required to ensure that this important age group do not fall further behind in the global effort to eradicate AIDS by 2030 [90].

Conclusions
For ALHIV in SSA, individual-level issues are the most influential in preventing engagement across all stages of the HIV care cascade, although family, community and structural issues also play an important role. Despite scale-up of services and increased availability of ART in SSA over the past decade, stigma remains the most pervasive barrier to HIV care engagement, present at each level of the SEM, and hindering progression across all stages of the care cascade. Psychosocial interventions among adolescents and their families should be prioritized to improve adolescents' engagement in HIV care, enhance their wellbeing and ultimately reduce their risk of ill-health and mortality.