HIV Care Preferences among Young People Living with HIV in Lesotho: A Secondary Data Analysis of the PEBRA Cluster Randomized Trial

Introduction . Sub-Saharan Africa is home to 89% of all young people living with HIV, a key population with specifc challenges and needs. In-depth knowledge of service demands is needed to tailor and diferentiate service delivery for this group. We evaluated HIV care preferences among young people living with HIV who were part of the PEBRA (Peer Educator Based Refll of ART) cluster-randomized trial. Methods . Te PEBRA trial evaluated a novel model of care at 20 health facilities in Lesotho, Southern Africa. In the PEBRA model, a peer educator regularly assessed participant preferences regarding antiretroviral therapy (ART) refll location, SMS notifcations (for adherence, drug refll, viral load), and general care support options and delivered services accordingly over a 12-month period. We present these preferences and their changes over time. Results . At enrolment, 41 of 123 (33.3%) chose ART refll outside the health facility, compared to 8 of 123 (6.5%) after 12 months. Among those selecting clinic-based ART refll, many preferred collecting ART during the peer educator led Saturday clinic club, 45 of 123 (36.5%) at the beginning and 55 of 123 (44.7%) at the end. SMS reminders for treatment adherence and ART refll visits were chosen by 51 of 123 (41.5%) at enrolment and 54 of 123 (44.7%) at the last assessment. Support by the peer educator was popular at the beginning (110 of 123 (89.4%)) and lower but still high at the end (85 of 123 (69.1%)). Tirteen of 123 (10.6%) participants chose support by the nurse, without the involvement of any peer educator, at the frst and 21 of 123 (17.1%) at the last assessment. Conclusion . Our longitudinal preference assessment among young people living with HIV in Lesotho showed a sustained interest in SMS notifcations for adherence and refll visits as well as in additional support by a peer educator. ART refll outside the health facility was not as popular as expected; instead, medication pick-up at the facility, especially during Saturday clinic clubs, was favoured. Te PEBRA trial was registered with clinicaltrials.gov (NCT03969030. Registered on 31 May 2019)


Introduction
According to the UNAIDS 2021 report, 2 out of 7 new HIV infections in 2019 occurred in young people aged 15 to 24 years and sub-Saharan Africa is home to 89% of all young people living with HIV [1,2].Tis age group makes up a substantial part of the HIV-positive population in sub-Saharan Africa with a share of 20% [3].AIDS-related deaths are the leading cause of mortality in this population [4].
In Lesotho, the Demographic and Health Survey of 2014 showed that 10% of young people were living with HIV with women being more afected (13%) than men (6%) [5].Young people living with HIV are a vulnerable subpopulation that faces distinctive challenges and therefore needs special attention on the way to the goal of ending the AIDS epidemic by 2030 [4,6].
Diferentiated Service Delivery (DSD) is an approach that shifts from a one-size-fts-all model to a patientcentered approach, with the idea of better delivery according to individual needs [7].For adults living with HIV, evidence about the efectiveness of such DSD models exists, including data about their sustainability and costefectiveness [8][9][10][11].However, for adolescents and young adults, the evidence for DSD models is scarce and DSD models are rarely designed and led by adolescents [12,13].To tailor programs more adequately according to the needs and demands of young people living with HIV, knowledge of their preferences is required.
In this secondary analysis, we used data collected in the intervention arm of the PEBRA (Peer Educator Based Refll of ART) cluster randomized trial in rural Lesotho [14].We evaluated participants' HIV care preferences, their feasibility, and intraindividual changes of preferences throughout the 12-month study period.

Study Design and Participants.
Tis is a preplanned secondary analysis of data collected in the intervention arm of the PEBRA trial, a cluster randomized controlled trial conducted at 20 nurse-led health facilities in three districts of Lesotho between November 2019 and April 2021.Te PEBRA trial assessed the efectiveness of a peer educator-coordinated preference-based antiretroviral therapy (ART) service delivery model among young people living with HIV in Lesotho (PEBRA model).PEBRA enrolled young people living with HIV aged 15-24 years taking ART.Te 20 health facilities (clusters) were spread over three mostly rural districts in Lesotho: Leribe, Butha-Buthe, and Mokhotlong.Te clinics were randomized in a 1 : 1 allocation to an intervention (PEBRA model) and a control arm.Detailed information about the setting, design, eligibility, randomization, and data collection and management, as well as about the primary and secondary endpoint is published in the PEBRA study protocol [14], and the main results were published (https:// journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1004150).

Te PEBRA Model.
Te PEBRA model was designed in collaboration with peer educators, young people living with HIV, youth advocates, clinical staf, and mobile application developers during several workshops supported and coordinated by two local nonproft organizations (SolidarMed and Sentebale) as well as the Ministry of Health of Lesotho.It consists of three pillars: ART refll location, SMS notifcations, and general care support, and made use of preexisting structures at the local clinics.For the ART refll location, participants could choose from the following options: refll at the clinic with the option of pick-up within the Saturday clinic club, refll at the village health worker's home, home delivery by the peer educator, refll at the community adherence club, or refll by a treatment buddy (Table S1).Regarding SMS notifcations, the participants could choose to get a notifcation reminding them to take their ART (adherence reminder), to remind them of the next ART refll visits (refll reminder) and to receive a viral load result message (viral load result notifcation) (Table S2).It was possible to opt into more than one notifcation option, and for each notifcation, they could specify the message content, time, and frequency.Te third pillar of the PEBRA model is the additional support that participants could choose from.Te diferent possibilities were support through the nurse at the clinic, Saturday clinic club, community youth club, phone call by peer educator, home-visit by the peer educator, school visit and health talk by peer educator, pitso (a community gathering) visit and health talk by peer educator, condom demonstration, more information about contraceptives, more information about voluntary male medical circumcision (VMMC), linkage to young mothers group (for pregnant women), linkage to a female social asset building model, and more information about gender-based violence/ legal aid.It was possible for participants to choose multiple sources of support.Each option is explained in more detail in Table S3.
At each of the 10 intervention facilities, a trained peer educator delivered the PEBRA model using the PEBRApp, a tablet-based application designed specifcally for the PEBRA study.Te peer educator conducted a preference assessment among his/her participants every three months or every month for virally suppressed or unsuppressed (>999 copies/mL) participants, respectively.At every preference assessment visit, all the options were shown to the participants visually in the PEBRApp and explained individually.Subsequently, participants were asked which options they preferred.Ten, the peer educator assessed if the chosen options were feasible (e.g., not every community had an established community youth club or the participants lived too far from the peer educator's home), and if not feasible, the second-best option was chosen and delivered.Te PEBRApp helped the peer educator to keep in regular contact with participants and keep track of participants' preferences and ART reflls.Together with the nurses and other clinic staf, the peer educator delivered services according to preferences and feasibility.Te chosen SMS notifcations were sent out automatically from the PEBRApp including a call-back option to the peer educators' number.

Variables and Time Points of Interest.
We included preference data for all three pillars of the PEBRA model over the course of the 12-month trial period.Te main variables of interest for this analysis were the proportion of participants requesting alternative ART refll than individual pickup at the clinic, adherence and refll reminder notifcations, and additional support options provided by the peer educator.We assessed the feasibility of providing selected options during all PEBRA preference assessments.For the 2 AIDS Research and Treatment longitudinal assessments in preferences over the study period, three time points were considered: (1) enrolment, (2) 6 months after enrolment (range: 5-7 months), and (3) 12 months after enrolment (range: 11-14 months).We chose these three timepoints following the SPIRIT diagram of the PEBRA trial as these were the 3 outcome data collection windows [14].
2.4.Sankey Diagrams.We created Sankey diagrams using the three defned timepoints and grouping preferences within each pillar of the PEBRA model.ART refll locations are shown in the categories of inquiry.SMS notifcations were grouped as adherence and/or refll reminders, which are not available in standard care; only viral load notifcations or no notifcations; and no cell phone available (see also Table S2).Support options were grouped as peer educator support, nurse support, and other support (see also Table S3).Peer educator support included Saturday clinic club (monthly gathering, led by the peer educator), community youth club, a phone call by the peer educator, a home visit by the peer educator, a school visit and health talk by the peer educator, or a pitso visit and health talk by the peer educator.Tese options were developed specifcally for the PEBRA model and were not otherwise available.Nurse support corresponded to the usual standard of care."Other" support options included support that was one-time support on the day of the assessment such as condom demonstrations, information about contraceptives, information about VMMC, linkage to young mothers' groups (for pregnant participants; DREAMS or Mothers-to-Mothers), linkage to a female social asset building model (for female participants; WORTH), and information about legal aid and genderbased violence.Tese "other" support options could be provided either by the peer educator, the nurse, or other staf at the health facility.

Statistical Analyses and
Software.We used absolute and relative frequencies to describe categorical data and medians and interquartile ranges for continuous variables.Te data analysis was performed in R (Version: R 4.1.1GUI 1.77 High Sierra build).Te Sankey Diagrams were built with SankeyMATIC (https://sankeymatic.com/build/).

Characteristics of the Study Population.
Te PEBRA model group enrolled 150 participants, of whom 123 (82%) were still in care at 12 months.Detailed sociodemographic and clinical characteristics including viral loads disaggregated by sex, follow-up status, and pregnancy/breastfeeding status can be found in Tables S4 and S5.In brief, the median age was 18.7 (interquartile range [IQR] 16.8-22.1)years, 99 of 150 (66%) were female, 148 (98.7%) were heterosexual, and the median number of completed school years was 9.0 (IQR 7.3-10.0).Asked about their occupation, 57 (38.0%) answered that they were attending school, 13 (8.7%) that they were (self-) employed, and the remaining 80 (53.3%) stated that they did not have an occupation.Of the 150 participants, 107 (71.3%) were single, 39 (26.0%) were married, 3 (2.0%) were separated/divorced, and 1 (0.7%) was widowed.At the time of enrolment, 41 (27.3%) participants had one or more children and among women, 19 of 99 (19.2%) were pregnant or breastfeeding.Te median number of years since HIV diagnosis was 5.5 (IQR 2.9-11.0),and the median number of years since starting ART was 4.9 (IQR 2.7-9.4).At the baseline, 82 of 150 (54.7%) had a documented viral load <20 copies/mL.Participants that were lost to follow-up at 12 months (27/150, 18%) had similar characteristics to those still in care, although it seemed that more were married, without employment nor school attendance (Table S4) with shorter time since HIV diagnosis (Table S5).
For the longitudinal assessment of service preferences, we restricted the study population to those still in care at 12 months.Tis allowed us to assess the individual changes over time from the baseline up to 12 months in detail.

ART Refll Preferences and Changes over Time.
We assessed changes in preferences over the 12-month study period for the three pillars of PEBRA: ART refll options, messages, and support options (Figures 1-3).We report here only preferences that were eventually also carried out.Te number of service preferences that were not feasible to deliver is reported in the last chapter of the Results section.
At enrolment, 41 of 123 (33.3%) intervention participants made use of the ofer of an alternative ARTrefll option outside of the clinic (Figure 1, Table 1).ART pick up by a treatment buddy was chosen by 16 of 123 (13.0%) participants, and 15 of 123 (12.2%) wanted to get their medication delivered to their home by the peer educator.Te village health workers' home was the preferred refll site for 9 of 123 (7.3%) participants, and 1 of 123 participants wanted to pick up the ART in the community adherence club.Out of the 82 of 123 (66.7%) participants who chose to pick up their medication in the clinic, 45 of 123 (36.6%) did so within the Saturday clinic club.Participants lost to follow-up at 12 months seemed to choose more often the nurse at the clinic and less often the Saturday clinic club (Table S6).Te same was true for pregnant/breastfeeding women.Participants who reported infection through their mother more often chose the Saturday clinic club and home delivery by the peer educator and less often the nurse at the clinic (Table S6).
At the end of the study, the proportion of the outside clinic refll preferences had shrunk to 8 of 123 (6.5%).Similarly, only 3 of 123 (2.4%) still chose home delivery by the peer educator, another 4 of 123 (3.3%) the village health workers' home, and 1 of 123 (0.8%) wanted to pick the ART refll up at the community adherence club.Te remaining 114 (93.5%) preferred to pick up the medication at the health facility, with about half of these 55 of 123 (44.7%) reflling ART at a Saturday clinic club.Troughout the study period, men more often chose to pick up the medication at the clinic within the Saturday clinic club than at a regular clinic visit.Te exact opposite was the case among women who preferred to pick up their ART at the regular clinic visit than AIDS Research and Treatment during the Saturday clinic club.Tis pattern became even more pronounced over the course of the study (Table 1).women.At enrolment, 88.9% of the participants lost to follow-up at 12 months had access to a cell phone and the majority chose refll and adherence reminders as well as SMS for VL notifcations (Table S7).

Support Options and
Changes over Time. Figure 3 summarizes the changes over time of those still in care at 12 months in terms of support options, grouped into four overarching support categories, whereby each participant could only be a part of one support category at each time point (see Methods).Table 3, on the other hand, lists the chosen support options in detail and not by the participant but by support option since one participant could choose several support options (see Methods).
Support by the peer educator was chosen by 110 of 123 (89.4%) participants at the frst timepoint and decreased to 85 of 123 (69.1%) by the end of the study (Figure 3).At enrolment, there were 13 of 123 (10.6%) participants who chose only support from the nurse at the clinic.At the last timepoint, 21 of 123 (17.1%) participants chose only support by the nurse, 1 of 123 (0.8%) chose only support without nurse or peer educator involvement, and 16 of 123 (13.0%) participants wanted to have no support at all.Similar to the ART refll preferences (Table 1), support preferences followed the same sex diference pattern regarding Saturday clinic club support and regular clinic nurse support.Participants lost to follow-up at 12 months showed a similar preference pattern at the baseline, with the majority requesting support from the peer educator (Table S8).

Overall Preferences and Feasibility.
During the entire study period, the peer educators conducted a total of 800 preference assessments among the 123 participants.Te median number of assessments per participant was 6 (IQR [5][6]. ART refll at the clinic was chosen in 671 of 800 (83.9%) and refll via a treatment buddy in 35 of 800 (4.4%) assessments.In all other instances, a refll outside the clinic was

AIDS Research and Treatment
School visit and health talk by the peer educator (%) More information about contraceptives (%)

AIDS Research and Treatment
the frst choice.However, among the 53 cases where the peer educator-delivery option was the frst choice, 12 (22.6%)had to be changed to another refll option due to feasibility constraints.
Among the 1839 support options chosen in the 800 assessments, peer educator support, nurse support, other support, and no support were selected in 1014 (55.2%), 622 (33.8%), 168 (9.1%), and 35 (1.9%) times, respectively.As with the refll options, chosen support options were not always feasible.In 27 instances, a home visit by the peer educator was not feasible due to distance, and in six instances, a pitso visit was not possible for the same reason.Community youth clubs and Saturday clinic clubs were selected but not available in the participant's community in 39 and at the participants' clinic in eight instances, respectively.In 3 cases, linkage to a female WORTH group was not feasible.In total, 83 support choices could not be delivered; this was 4.5% of all care support demanded.

Discussion
HIV care preference data among young people taking ART at HIV clinics in rural Lesotho revealed that ART refll outside the clinic was not as popular as expected.At enrolment, 33.3% chose ART refll outside the health facility; however, twelve months later, more than 90% were choosing to get reflls at the clinic despite little issues of feasibility.Furthermore, the data showed that SMS notifcations such as adherence and refll reminders were widely chosen throughout the one-year study period.Additional support by the peer educator was feasible and highly popular when the study was introduced (89.4%) and decreased over the 12month period but remained popular towards the end (69.1%).
Despite the high interest and urgency statements by the WHO and other international organizations and the known knowledge gaps regarding this population, research to address these shortcomings is limited [2,15].Tere is some research showing promising fndings regarding DSD models for young people in Africa, but randomized and high-level evidence is scarce with the exception of the Zvandiri trial [16][17][18].A recent systematic review highlights the paucity of data evaluating adolescent HIV service delivery models [13].
Regarding preference data among young people living with HIV in sub-Saharan Africa, studies and discrete choice experiments on HIV testing [19], pre-exposure prophylaxis [20,21], and fnancial incentives [22] exist but none assessing general care preferences.In a formative study about the engagement of young people in sexual and reproductive health, there was great interest in accessing health services at community hubs rather than the health facilities [23].
For adults living with HIV, the study situation is somewhat diferent.Tere are studies from the last three years from Kenya, Zambia, and Ghana that have examined treatment preferences among people living with HIV [24][25][26][27].All three studies found a preference for facilitybased care even though in most cases, this also entailed higher costs for the participants but with less frequent visits and individual consultations.Te main reason cited by participants was fear of HIV status disclosure to their own community.All three studies excluded adolescents, and in the Kenya and Ghana studies, the average age was 41 years and just under 50 years, respectively, and in the Zimbabwean study, half of the participants were over 30 years of age.All three studies were based in urban areas.Interestingly, similar preference patterns to our data emerged.
SMS notifcations and peer support were the service options with the highest uptake.Peer support may generate trust and, thus, stability in the long run [28,29].However, it is important to note that 13% of participants wanted no support at all, and 17% only wished for support from the nurse at the health facility.Rather surprising was the high share of participants preferring to come to the clinic for their ARTrefll rather than the more decentralized options such as home delivery and pick up at the village health worker's home.Te reason was not that the decentralized options were unfeasible (only 1.5% of frst choices were turned down due to feasibility).While decentralizing ART pick-up sounds appealing by removing structural barriers [9,[30][31][32], we could not confrm this statement in practice among the study population.As the study was run during the COVID-19 pandemic when mobility and public life were reduced to a minimum, we would have expected more decentralized ART reflls in this setting.Te Saturday clinic clubs may have played an important role for clinic refll as they are well established at most clinics.Also, nurses at the study facilities received a training in adolescent-friendly service delivery before the inception of the study and thus might have contributed to more clinic-based service choices.Moreover, people living with HIV still face stigmatization in their community [33][34][35].Tis is another possible reason why participants may have preferred to come to the clinic to pick up their medication because it makes them feel less visible.It is also possible that the participants did not trust their peers or community members to provide ART reflls or were dissatisfed with the services.Te exact reasons remain to be explored in further research to adapt services to the needs of people living with HIV during the sensitive phase of adolescence and young adulthood.
Interesting was the sex diferences regarding the two options for refll and support at the clinic.Men more often chose the Saturday clinic club as refll and support site, while women more often preferred a regular clinic visit with the nurse as their primary support and ART refll pick up point.Over the 12 months period, this pattern became even stronger.According to the baseline characteristics, more men were employed or attending school, while more women 8 AIDS Research and Treatment reported no occupation (62% among women vs. 37% among men).Hence, the Saturday clinic club may have suited men better than women.However, this is only one explanation, and we need to investigate these diferences in more detail in a qualitative follow-up study.Participants who reported being infected through their mothers seemed to prefer the Saturday clinic club and home delivery by the peer educator for their ART refll, diferent to participants horizontally infected.Tese diferences in the preference pattern may be interesting to explore further to ofer diferentiated services for each group.
Tis study has several limitations.Te frst concerns data collection.Despite training the peer educators on data collection and providing instructions on how to present the options to the participants, we cannot exclude the possibility of peer educators' attitudes infuencing participant preferences.Conditioning cannot be ruled out since the participants and peer educators knew that not all options were feasible for all participants at all times.Second, Lesotho has a unique geography, and the presented data originate from rural areas.Participants in urban areas of Lesotho may have diferent preferences.While our data demonstrate the feasibility of and demand for alternative care options, these cannot be expected to lead to direct improvements in clinical outcomes.

Conclusions
Tis longitudinal preference assessment among young people living with HIV in rural areas of Lesotho is a frst of its kind among this key population.It shows that ART refll outside the health facility was not as popular as expected; instead, medication pick-up at the facility, especially during Saturday clinic clubs, was favoured.More research is needed to investigate the underlying reasons for each preference pattern.Overall, this key population has a clear interest in SMS notifcations to remind them about medication adherence and upcoming refll visits and support provided by a peer educator.

Figure 2 and
Table 2 summarize the SMS notifcation preferences over time.At enrolment, 72 of 123 (58.5%) had access to a cell phone where they could receive confdential information on and this number increased to 80 out of 123(65.0%)at the end.Te number of participants who wished to receive either a refll or an adherence SMS reminder was 51 of 123 (41.5%) at enrolment and 54 of 123 (44.7%) at the last assessment.Te option to receive only VL notifcations or no notifcations was chosen by 21 of 123 (17.0%) participants at the beginning of the study, respectively, by 25 of 123 (20.3%) at the end.Overall, men seemed to have less access to a cell phone and chose less notifcations than
females: linkage to a female WORTH group (social asset building model) legal aid and gender-based violence (%) 4 ( Abbreviations: community youth club (CYC), SCC (Saturday clinic club), and VMMC (voluntary male medical circumcision).

Table 1 :
ART refll site preferences over time, by sex, among those still in care at 12 months.Abbreviations: ART (antiretroviral therapy), CAC (community adherence club), SCC (Saturday clinic club), and VHW (village health worker).

Table 2 :
SMS reminder preferences over time, by sex, among those still in care at 12 months.

Table 3 :
Support preferences over time, by sex, among those still in care at 12 months.