“We did not know what was wrong”—Barriers along the care cascade among hospitalized adolescents with HIV in Gaborone, Botswana

High mortality among adolescents with HIV reflects delays and failures in the care cascade. We sought to elucidate critical missed opportunities and barriers to care among adolescents hospitalized with HIV at Botswana’s tertiary referral hospital. We enrolled all HIV-infected adolescents (aged 10–19 years) hospitalized with any diagnosis other than pregnancy from July 2015 to January 2016. Medical records were reviewed for clinical variables and past engagement in care. Semi-structured interviews of the adolescents (when feasible) and their caregivers explored delays and barriers to care. Twenty-one eligible adolescents were identified and 15 were enrolled. All but one were WHO Clinical Stage 3 or 4. Barriers to diagnosis included lack of awareness about perinatal HIV infection, illness or death of the mother, and fear of discrimination. Barriers to adherence to antiretroviral therapy included nondisclosure, isolation, and mental health concerns. The number of hospitalized HIV-infected adolescents was lower than expected. However, among those hospitalized, the lack of timely diagnosis and subsequent gaps in the care cascade elucidated opportunities to improve outcomes and quality of life for this vulnerable group.


Introduction
Worldwide, adolescent AIDS-related deaths are increasing, despite improvements in access to antiretroviral therapy (ART) [1]. From  in all other age groups, AIDS-related deaths in adolescents increased by 50% [2]. HIV/AIDS is a leading cause of death among adolescents globally, particularly in sub-Saharan Africa [3]. Mortality increases are particularly striking among [10][11][12][13][14] year-old adolescents with HIV, reflecting failures to diagnose perinatally infected children and to successfully engage them in care early in the course of their illnesses. Gaps have been described throughout the adolescent HIV care cascade-at HIV diagnosis, linkage to and retention in care, and ART adherence and suppression of viral load [4]. There is an urgent need to identify strategies to help adolescents make the same morbidity and mortality gains seen in other age groups.
Botswana is among the countries with the highest HIV prevalence in the world, with 21.9% of the adult population living with HIV [5]. Botswana has been a pioneer in responding to the HIV/AIDS epidemic; Botswana introduced PMTCT services in 1999, and in 2002 was the first sub-Saharan African country to provide universal free ART to citizens [6]. Botswana is now approaching 90-90-90 targets; however, successes are not shared equally across age groups. While it is estimated that 84% of Botswana's adults living with HIV are receiving ART, only 60% of Botswana's children (under 15 years) living with HIV are on ART [5]. Age-disaggregated outcome data are not available for the approximately 17,000 adolescents living with HIV in Botswana [5].
By investigating delays in the HIV care cascade among hospitalized adolescents with HIV in this high-prevalence setting, we aimed to examine the full spectrum of missed opportunities for HIV care in those most vulnerable to poor outcomes.

Setting
This study was conducted between July 2015 and January 2016 at Princess Marina Hospital (PMH) in Gaborone, Botswana, the country's main government referral hospital. PMH is located adjacent to the Botswana-Baylor Children's Clinical Centre of Excellence, which has provided comprehensive HIV care to thousands of children and adolescents since 2003, and provides adolescent-centered services, such as a Teen Club support group. HIV testing is indicated for all hospitalized children and adolescents with unknown HIV status, per Botswana National HIV & AIDS Guidelines [7].

Data collection
The study nurse and other research team members consulted with ward medical teams to identify all adolescents (ages 10 to 19) with HIV hospitalized at PMH for any admission diagnosis other than pregnancy over a 6-month period. Adolescents admitted for any admission diagnosis to the medical, pediatric, surgical, oncology, or intensive care wards were included. Pregnant patients were excluded because the intent was to understand reasons for delayed access to care, and pregnant patients with HIV represent mainly those with more recent behaviorally-acquired infections in this context [5,8]. Prenatal care (PNC) uptake is high and 92.8% of mothers in Botswana's PNC program were tested for HIV in their last pregnancy [8]. Furthermore, hospitalizations of pregnant adolescents in this setting are primarily for labor and delivery, or obstetric complications, rather than complications of HIV, which were our primary interest. Adolescent HIV deaths are roughly equal among males and females, and highest among 10-14 year-olds, arguing against pregnancy being a major driver of adolescent HIV deaths [2,9].
We developed semi-structured interview guides which begin with open-ended questions querying the adolescents' prior engagement in care (S1 File). Sequential questions determined if critical failures or delays occurred in the diagnosis and treatment cascade. Probes they stated that they would not publish full interviews, and that they would not share any information that could potentially lead to identification of the individual participants. It was also conveyed in the consent forms that no information would be publicly shared that could potentially identify participants. Given the very sensitive nature of these interviews and possibility of identification of the individual participants if the full interviews were to be publicly available, in addition to their previous assurances that they would not publish the full interviews, the authors cannot make the qualitative interviews available. Data requests may be sent to The Institutional Review Board of the University of Pennsylvania, who may be reached by e-mail at irb@pobox. upenn.edu or by phone at +1 (215) 573-2540.
investigated barriers to care at each stage of the cascade where critical challenges or failures occurred. The acceptability of specific proposed interventions to improve care at each stage was explored. Separate interview guides were used for disclosed and non-disclosed adolescents, and for caregivers. Guides for non-disclosed adolescents did not mention HIV, but asked generally about barriers to medical care for adolescents, as well as any challenges with adherence to medicines and clinic visits. Eligible adolescents and their caregivers were approached with information about the study and to request informed consent/assent. Strict precautions were taken to prevent inadvertent disclosure of HIV status throughout the informed consent process and study procedures. This included avoidance of inadvertent disclosure to young adolescents who were not aware of their own status by avoiding use of the term HIV in any assent documents or in the interview guide for non-disclosed adolescents. Adolescents were presumed to be non-disclosed unless their primary caregiver confirmed that the adolescent was aware that he or she had HIV infection. Caregivers and adolescents (when medically stable) were interviewed separately in a confidential setting near the hospital ward. Interviews were conducted in Setswana by a bilingual research nurse with extensive experience caring for children and adolescents with HIV and training in interview techniques. Clinical and demographic variables were abstracted from medical records onto paper forms, and uploaded into a secure REDCap database [10]. A master list of study numbers and identifying information was kept separate from the de-identified study data used for analysis. Study materials were kept in a locked cabinet in a secured office where only the study team had access.

Data analysis
Clinical and demographic data were analyzed using standard descriptive statistics, using Stata/ SE 14.1 (StataCorp, College Station, Texas, USA). Interviews were audio-recorded, transcribed, and translated into English. The first and senior authors reviewed all transcripts and relayed queries to the study nurse, who contacted participants with follow-up questions when clarification was needed. Transcripts were coded and analyzed through thematic analysis. Initial codes were organized according to stage of the care cascade addressed by the participants, as well as hypothesized factors impacting the care cascade. Through reading and re-reading of the data, the coding structure was augmented with themes that emerged from the data [11,12]. NVivo 11 analysis software (QSR International Pty Ltd, Melbourne, Australia) was used to organize and analyze the data. enrollment and caregiver consent was withdrawn after another died. Two minors did not have a caregiver present who could consent to their participation and two declined consent/assent. Of all eligible adolescents, 4/21 (19.0%) died; among the adolescents enrolled in our study, 2/ 15 (13.3%) died during the hospitalization. Interviews were conducted with 13 adolescents and 15 adult caregivers, of whom 9 were mothers, 4 were sisters, one was an aunt, and one was a grandmother to the hospitalized adolescent.
Only the two patients admitted for elective surgical procedures were diagnosed in infancy. All other adolescents had experienced delays in HIV diagnosis (Table 3). While delays were common, only one patient was diagnosed during the current hospitalization, at the age of 12. Significant challenges to retention in care were identified, with 3/14 previously diagnosed adolescents having been out of care as much as several years. Of 13 patients on ART, 6 reported significant adherence difficulties. Specific barriers and missed opportunities for care were explored for each stage of the cascade where adolescents experienced significant delays or challenges (Table 4). Many barriers presented significant challenges at multiple stages of the cascade. Barriers related to diagnosis and adherence were most common and are presented in-depth.

Missed opportunities for diagnosis
Presenting frequently to clinics or hospitals without receiving an HIV test was reported by many. Chronic illnesses included cough, pneumonia, TB, diarrhea, ear infections, rashes, oral ulcers, and wasting. In a few cases, the child had to withdraw from school due to severe illness, but had not yet been tested for HIV. In some, failure to diagnose the mother may have contributed to delayed diagnosis of the child.

Barriers to HIV diagnosis, linkage, and retention in care
Lack of awareness. Caregivers frequently discussed their past lack of knowledge of the possibility of undiagnosed perinatal HIV infection in their children, or remarked that they "did not believe" that the child could be infected. Some noted that the child "grew up wellnourished" or "was doing fine" prior to becoming ill. When the child did become ill, multiple caregivers remarked that "it never crossed [their] mind" or that they "never thought [that the child could have HIV]." In some cases, the mother herself had not yet been diagnosed-one mother reported, "even myself I had not tested." Illness or death of the mother. Severe illness or death of the child's mother was a common contributor to delayed diagnosis or failure to retain a child in care when "there was no one to bring [the child] to clinic". Challenges in adherence and retention for mothers-"her mother stopped treatment and is now very sick"-contributed to delays in children's care. In where she was also presenting with diarrhea and they asked me if the child has been tested. Because I didn't know, I said 'no' and they told me to take her for HIV testing and bring back the results. But before I left they picked it up from the medical cards that the child had been tested long ago and was started on treatment which was stopped at some point, and they referred me to [HIV treatment clinic]."-Aunt of a 14 year-old female "His mother is the one who was living with him then, and I didn't ask if the child was tested for HIV; she also didn't mention anything. When she told us that he was diagnosed with TB we never thought of anything else. We were happy that at least there was a diagnosis."-Grandmother of a 12 year-old male Barriers to care among hospitalized adolescents with HIV in Gaborone  Peer support Retention, Adherence "If we suffer from the same illness we become free with each other. For example, one will maybe say, 'I don't like Combivir it does not make me feel good,' and the peer will respond, 'but Combivir it's good because it suppresses the virus.' So they are able to encourage each other as peers."-Sister of a 19 year-old female "[Joining a peer support group] will help me to understand that whether you are on medications or not you are still the same."-12 year-old male "We will be in the same air together, and no one will be like saying, 'this one has HIV and this one does not have,' it will be just fun."-14 year-old female "It will help to be free when I am with others."-13 year-old male Family therapy Retention, Adherence "I have an aunt who is also on ARVs. I once thought maybe my aunt could join us when we come for clinic checkups so that she discloses to the boy as part of encouragement; maybe he will see that he is not alone. these situations, other family members needed to take on the medical care of the child, including the decision to test for HIV. Lack of disclosure to other family members. Lack of disclosure of the child's status to other family members was an important barrier to diagnosis and retention in care, particularly if the mother was ill, or if the child was living with family members who were not aware of the child's status. Two adolescents had several-year gaps in care after diagnosis, due to severe illness of the mother and lack of disclosure of the child's status to other family members who potentially could have brought the child for care. I was very sick and bedridden. I had once come to clinic after he was diagnosed, for initiation [of ART], but I stopped as there was no one to bring him. I had not told anyone about my child's status at home, for fear of being discriminated against, even though they knew my status.
-Mother of a 13 year-old male Stigma and fear of discrimination. Stigma and fear of discrimination were barriers to diagnosis and care at all stages of the cascade. Caregivers expressed fear of discrimination, even from close family members, as in the preceding example. They reported that stigma was also acutely experienced by adolescents, particularly in their interactions with peers.
It hurts him; he does not like it. Especially when you speak with a raised voice, telling him he has the sickness, his perception is maybe other people are listening.

-Sister of a 17 year-old male
We need to meet with health professionals to continue talking about this illness, so that the child understands that this is an illness like any other; she does not differ from others.

Adherence challenges
Adherence challenges were commonly described among adolescents on ART. Key barriers to adherence included: lack of disclosure of HIV status to the adolescent or family members, stigma, isolation, and mental health concerns including para-suicidal behavior. (A sister of a 19 year-old female reported, "sometimes she will even be refusing to go to the clinic, saying she wants to be left to die.") For those adolescents who had been told their HIV status, adolescents and caregivers spoke of both the challenges and breakthroughs for these youth in "accepting themselves," or coming to terms with the diagnosis.

Potential interventions
Participants were asked about specific proposed interventions at each stage along the cascade where challenges were faced (Table 4). Peer support was currently available to some, but not all, of the disclosed adolescents, and was seen as a particularly vital intervention to support adolescent mental health, confidence, education about HIV, and adherence. Adolescents described the peer group setting as a means to escape isolation and stigma-to "be in the same air together"-to "be free with each other." As one 12 year-old male stated, "[the groups help us] to understand that whether you are on medications or not you are still the same." Unfortunately, many adolescents were not able to benefit from a peer group due to being unaware of their HIV status or receiving care at sites lacking peer groups.
Regular reminders from personal contacts were seen by some as helpful, though others felt that visit reminders were unnecessary for themselves. Similarly, cell phone reminders (of medication schedule or of clinic visits) were generally not perceived as being as helpful as phone calls from clinic staff in the case of missed appointments or concerning lab results. Such outreach was seen as evidence of a strong relationship of the clinic staff to the adolescent. Multiple caregivers noted, "it shows that they care." Family therapy interventions were generally viewed favorably. Several caregivers noted that the involvement of other family members could provide additional encouragement and support to the adolescent. A few caregivers and adolescents expressed concerns relating to disclosure of the adolescent's status to close family members-"I just don't want them to know." Family interventions may be challenged by limited disclosure within the family.
Caregivers and adolescents offered additional unprompted insights into key facilitators to adolescent care, including adolescent disclosure support, family disclosure and involvement, and adolescent mental health services and counseling ( Table 4). The need for additional adolescent counseling was noted by multiple caregivers, as they described significant isolation and mental health concerns among their adolescents. The importance of disclosure to the adolescent as well as to trusted family members were recognized as being critical to the discussed interventions.

Discussion
Significant delay in HIV diagnosis emerged as the most consistent challenge in the care cascade in this study of hospitalized adolescents with HIV infection. The median age at HIV diagnosis (7.5 years) is older than it was once thought that children could survive with HIV untreated [14], and a third were diagnosed in adolescence. While long-term survival of perinatally infected adolescents is increasingly recognized among clinicians and researchers, most laypeople may not know that these children could potentially survive into adolescence untreated [15]. Few of these adolescents were born in the pre-ART era, when more than half died in the first 2 years of life [16]. Many improvements in pediatric HIV diagnosis have occurred since the time these adolescents were infants, with widespread availability of early infant diagnostic services and improved implementation of "opt out" testing. However, there has been insufficient emphasis on HIV diagnosis of young children in healthcare settings outside of the PMTCT context [17]. Most adolescents in this study presented to clinics or were hospitalized on several occasions, before being tested for HIV. For some, HIV testing was reportedly not offered despite their manifestations of HIV-related illnesses. For others, there was hesitancy on the part of the caregivers to test. Clinical and public health leaders in high HIV-burden settings should eliminate barriers to universal pediatric testing. This should include efforts to increase awareness among health care workers and lay people about the possibility of survival into older childhood and adolescence with perinatally acquired HIV. Other efforts to increase uptake of pediatric HIV testing in clinical settings, and to support caregivers through counselling and stigma-reduction interventions, should be pursued.
Late presentation of perinatal HIV infection is associated with profound immunosuppression and advanced disease [14]. Perinatally infected adolescents who do not initiate treatment until after long-standing, advanced immunosuppression are medically vulnerable [18]. Prolonged untreated HIV infection contributes to chronic inflammation, impaired immune recovery, neurologic and developmental pathology, and the development of opportunistic infections and malignancies. The fact that HIV is the leading cause of death among [10][11][12][13][14] year-olds worldwide suggests that we have failed to find and adequately treat far too many children before they develop profound illness [19]. The high mortality (19.0%) among adolescents with HIV in our cohort is consistent with that observed among children and adults hospitalized with HIV globally and in the African region [20]. There are very limited data around causes of hospitalization and mortality for adolescents with HIV infection; in our series, AIDS-related diagnoses were frequent, as has been observed among children and adults with HIV globally in the ART era [20].
Once diagnosed and linked to care, disengagement of children and adolescents is a complex challenge. A few adolescents in this series had become disengaged soon after diagnosis, and were out of care for years prior to hospitalization with advanced illness. The challenge of early disengagement from HIV care has been described in children in Botswana, and in studies of youth and adults elsewhere in sub-Saharan Africa [21][22][23][24]. Prior research has demonstrated the negative impacts of stigma, discrimination, and fear of disclosure on adult retention in HIV care [25]. In this series, maternal illness, stigma, and lack of disclosure to close family members influenced poor retention. Further research is needed to define barriers and facilitators to long-term retention for children and adolescents in high-prevalence settings. In addition, interventions to minimize disengagement of children and adolescents and to track and re-engage those who are lost to follow-up should be included in HIV care programs.
Stigma played a central role in breakdowns in the care cascade, and impacted adolescent care through pervasive effects on disclosure to the adolescent and to family. Family relationships, engagement with care, experience of isolation, and adherence to ART were also directly impacted by stigma. While stigma significantly complicates care for all persons living with HIV, perinatally infected adolescents have a particularly unique and challenging experience of stigma [26,27]. These adolescents transition from dependence on caregivers (who may themselves be facing severe illness, mental health issues, poverty, stigma, and isolation); to learning their status; coping with illness or death of family members; navigating peer relationships; and developing autonomy in care; all while undergoing the developmental transition of adolescence and into young adulthood. Interventions to reduce stigma and to mitigate the experience of stigma for adolescents-by strengthening family, peer, and community support-are urgently needed, and will be critical to improving adolescent HIV outcomes in the care cascade [26][27][28][29].
Problems adhering to ART were common in this study of hospitalized adolescents, and daunting adherence challenges in this age group are well documented [18,30,31]. It is important to recognize that in Gaborone, where pioneering adolescent-centered services are available to support adherence, adherence nevertheless remains a critical and complex challenge. Participants in this series were enthusiastic about participation in a peer group to reduce social isolation and improve HIV education and ART adherence. Many of the adolescents in this series were not receiving adolescent-centered services, such as engagement in a peer support group, either because their status was not yet disclosed to them, or because they were enrolled at clinics that did not provide these and other youth-friendly care services. Indeed adolescentcentered HIV care services remain an exception to the rule, and are not widely available in most settings [18]. For those engaged at clinics providing such services, some continued to have complex difficulties adhering to therapy, including mental health needs requiring ongoing counseling and support. The current evidence base for interventions to improve adolescent adherence is severely limited [32]. Studies of developmentally-appropriate approaches to improve adolescent adherence to ART are critically needed.
Despite the gaps in care and advanced illness experienced by the adolescents in this series, the number of adolescents identified at the tertiary referral hospital in Botswana over a 6-month period was unexpectedly small. This finding contrasts with alarming recent estimates that 100 adolescents are dying daily from HIV in sub-Saharan Africa [33]. While the HIV prevalence in Botswana is high, with its small population, the approximately 17,000 HIVinfected adolescents in Botswana make up less than 1% of the global population of adolescents with HIV, and an even smaller proportion of global adolescent HIV deaths, with fewer than 500 deaths annually [5]. PMH serves as both the major referral hospital for the country and the primary hospital for the capital city. However, our study may under-represent the adolescent epidemic in Botswana if adolescents with HIV were predominantly admitted to primary hospitals other than PMH or died in the communities. It's worth noting that chronically ill patients in Botswana may be brought back to their ancestral villages for end-of-life care [34]. It is also possible that the low number of admissions reflects some relative successes of the pediatric HIV treatment program in Botswana, which may not be representative of the broader adolescent HIV epidemic in sub-Saharan Africa.
A further limitation of our study was our reliance on interviews for self-report of adherence challenges, and for recollection of delays and challenges earlier in the adolescent's history. Recall bias may have minimized some barriers while over-emphasizing others. The breadth of our semi-structured interview was designed to ascertain the full scale of missed opportunities and challenges along the HIV care cascade for adolescents. However, this did not allow for exhaustive exploration of all specific barriers or potential interventions. For example, targeted interventions to support family disclosure and to mitigate mental health concerns were discussed with all participants, but not comprehensively evaluated by this study. Finally, by excluding pregnant adolescents whom we presumed would be admitted primarily for obstetric care and more likely to have behaviorally acquired HIV, we were not able to evaluate gaps in the care cascade prior to hospitalization with any HIV-related complications in this group. These are important areas for future investigations.

Conclusion
These data highlight the failure of timely diagnosis of pediatric HIV infection among adolescents at the highest risk of poor HIV treatment outcomes. More work is still needed to address pervasive stigma, to study interventions for adherence and retention, to support disclosure to both adolescents and their close family members when needed, and to raise awareness of late presentation of perinatal HIV infection. Greater access to disclosure support, peer groups, adherence support, and mental health care is needed.
Supporting information S1 File. Interview guides. Interview guides used for caregivers, disclosed adolescents, and non-disclosed adolescents.