An Online Support Group Intervention for Adolescents Living with HIV in Nigeria: A Pre-Post Test Study

Background: Adolescents living with HIV (ALHIVs) enrolled in HIV treatment services experience greater loss to follow-up and suboptimal adherence than other age groups. HIV-related stigma, disclosure-related issues, lack of social support, and limited HIV knowledge impede adherence to antiretroviral therapy (ART) and retention in HIV services. The 90-90-90 goals for ALHIVs will only be met through strategies targeted to meet their specific needs. Objectives: We aimed to evaluate the feasibility of implementing a social media-based intervention to improve HIV knowledge, social support, ART adherence, and retention among ALHIV aged 15-19 years on ART in Nigeria. Methods: We conducted a single-group pre-post test


Introduction
Despite important gaps in our understanding of the HIV epidemic among adolescents (age, 10-19 years), available data indicate that this age group experiences high rates of new HIV infection, has less access to treatment, and is more likely to discontinue treatment and die from HIV-related causes than other age groups [1,2]. More than 80% of adolescents living with HIV (ALHIVs) are in sub-Saharan Africa, where, despite considerable progress, treatment coverage remains challenging: By 2016, less than 50% of ALHIVs received treatment [1][2][3]. Between 2010 and 2016, HIV-related mortality among adolescents decreased by 5% globally and 14% in Eastern and Southern Africa; however, in Western and Central Africa, the related mortality increased by 15% [2].
In 2016, nearly 50% of HIV-related deaths among adolescents in Western and Central Africa occurred in Nigeria [2]. One contributing factor to the high mortality may be poor treatment retention among ALHIVs. A 2016 study of youth (age, 14-24 years) initiating antiretroviral therapy (ART) in Nigeria found that they were less likely to remain in treatment at 12 months (46% vs 53%) and more likely to experience treatment disruptions (>3 months between clinic visits) in their first year on ART than older adults (relative risk=1. 15, P=.008) [4].
Fear of stigma or disclosure, lack of social support, and limited knowledge about the disease, compounded by the physical, social, and psychological changes of adolescence, create numerous challenges to successful treatment [5][6][7][8][9][10]. For example, adolescence is a time when relationships outside of the immediate family, such as those with peers, become very important [11]; however, because of HIV-related stigma and discrimination, ALHIVs rarely disclose their status to their peers for fear of losing them, which can negatively impact their psychosocial well-being and treatment [12][13].
Evidence of effective interventions to improve treatment outcomes among ALHIVs, particularly in low-and middle-income countries, is limited [14][15]. ALHIVs are also less likely than adults or younger children to be targeted with interventions to improve health outcomes [14,15]; however, some intervention strategies designed for adults may be effective for ALHIVs. Emerging evidence supports group counseling and structured support groups as effective ways to improve adherence and retention among adults [16][17][18][19][20][21]. Digital health interventions such as mobile phone reminders improved adherence among adults in some low-and middle-income countries [22][23][24][25][26][27]. Two recent studies targeting youth in South Africa and the United States demonstrated that online social network interventions could be acceptable and feasible in this population [28][29].
According to a 2014 survey, 89% of Nigerians aged ≥ 18 years owned a mobile phone [30]. A 2012 study in Nigeria found that more than half of the females aged 12-30 years owned a phone, and almost all who did not own a mobile phone had access to one [31]. The rapid increase in mobile phone use in Nigeria indicates the potential of digital health strategies to help meet the support needs of ALHIVs.
In this study, we aimed to develop and test the feasibility and acceptability of a structured support group intervention-SMART (Social Media to promote Adherence and Retention in Treatment) Connections-which is delivered through a social media platform to improve retention in HIV health services and ART adherence among ALHIVs aged 15-19 years in periurban southern Nigeria.

Design
We conducted a mixed-methods, single-group, pre-post test study from June 2017 to January 2018. Participants were recruited from 3 health facilities in Akwa Ibom State and enrolled to receive 5 intervention sessions [32]. Data were collected through face-to-face interviews using a structured questionnaire at baseline and endpoint. In-depth interviews (IDIs) with a subset of purposively selected participants were conducted at endpoint and stratified by participation level. Participation data were collected using a Facebook group analytics tool [33]. The FHI 360 Protection of Human Subjects Committee, Durham, North Carolina, and the University of Uyo Teaching Hospital Ethics Committee Uyo, Akwa Ibom State, Nigeria, approved this study. Written informed consent was obtained from adult participants aged ≥18 years. Written parental permission and adolescent assent were obtained from participants aged 15-17 years.

SMART Connections: Development and Description
SMART Connections was designed to promote ART adherence and retention in HIV services by leveraging informational, emotional, and network dimensions of social support. Content was adapted from sessions of the Positive Connections: Leading Information and Support Groups for Adolescents Living with HIV guide, to be delivered online through "secret" Facebook groups [34]. Secret Facebook groups limit membership and access to those invited and added by a group administrator. Content in secret Facebook groups can be seen only by members. These groups cannot be found through online search engines and do not appear on Facebook users' timelines; therefore, others who are not part of the group cannot determine whether a person is a member of a secret group [35].
Intervention components included informational messages and moderated group discussions, and 5 of 14 Positive Connections sessions were included in this study: Understanding HIV; Disclosure and Developing Trust in Relationships; Treatment and Adherence; Nutrition and Health; and Sex and Relationships. Sessions included activities for the facilitators to post and subsequently lead discussions. Within each session, the activities included the following: At a glance, to introduce the topic of the week ( Figure 1); word of the week, to define one key concept for the topic ( Figure 2); cartoons, to tell a story related to the topic ( Figure 3); key messages, to deliver important information; quizzes, to assess participants' knowledge and stimulate discussion; and group discussions moderated by facilitators. Four facilitators, recruited from existing community-based organizations and previously trained to lead in-person support groups for people living with HIV, received a 1-week training on the intervention. Study support groups began with an initial, in-person meeting during which participants met each other and the facilitator. The facilitator described the intervention and the group agreed upon ground rules for interactions, emphasizing the need to maintain confidentiality for group interactions. All participants received a basic mobile phone (also known as a feature phone) that could access Facebook, regardless of current phone ownership, because our primary interest was to determine whether the intervention could be implemented as designed and if participants would engage in it, given the opportunity. Participants selected the cellular network on which the phone was registered and were allowed to keep the phones at the end of the study. Each participant received 1000 Naira (USD 3.33) of data each month to facilitate participation, corresponding to 1.5-2 gigabytes of data monthly, depending on the network provider. This amount of data translates roughly to 300-400 social media posts with photos [36].

Sample Size and Sampling Design
We aimed to enroll 40-50 ALHIVs to form 5 support groups of 8-10 individuals. Eligibility criteria for ALHIVs for inclusion in the study were age, 15-19 years; on ART for >6 months; and basic literacy to participate in online chats. Literacy was assessed by data collectors during eligibility screening by asking the participant to read aloud three short sentences from the intervention content. If the data collector determined that the participant could read all or most of the sentences, the person was deemed eligible. Participants who could not read at all or struggled with all three sentences were deemed ineligible. ALHIVs who planned to move from the area before the end of the study, were enrolled in an in-person support group, were enrolled in another HIV-related research study, or were critically or severely ill at enrollment were excluded from the study.
Eligible participants were sequentially recruited during clinic visits. Additionally, participants were identified from medical records by a clinic staff and contacted by telephone to tell them about the study and, if interested, to come to the facility to learn more about the study. IDIs were conducted for 8 participants with moderate to high active participation and 8 participants with no to low active participation at the study endpoint.

Measures
We collected demographic information of participants (sex, age, relationship status, education, occupation, and religion) and their HIV infection (date of diagnosis, date of start of ART, disclosure to others, viral load, and CD4 [cluster of differentiation 4] results). We also collected data on preintervention access to mobile phones, the internet, and experience with social media.
We measured intervention implementation and participation using Grytics software [33]. To measure fidelity, we determined if intervention activities occurred within 1 week of when they were scheduled and the proportion of scheduled activities completed. In addition, we measured if and how participants participated in each scheduled activity. Group members could participate in several ways: commenting on a scheduled post, "liking" or "reacting" to a post or comment, making a new post, or commenting on or reacting to others' posts. IDIs explored participants' experiences with the intervention.

Data Analysis
Quantitative data were analyzed descriptively and independently verified by a second analyst. For participation data, posts (original message) and comments (replies to a post) were equally weighted and each assigned a value of 1. Participation was categorized into approximate quartiles of total posts over the 5 sessions: none to very low (0-5 total posts), low (6-20 posts), moderate (21-50 posts), and high (≥51 posts).
IDIs were audio recorded and transcribed into English. Some IDIs were conducted in local languages and then simultaneously transcribed and translated. Qualitative data were analyzed by an applied thematic approach using NVivo 11 [37,38]. A codebook was created on the basis of the interview guide, and emergent thematic codes were added during analysis. Two analysts coded the transcripts and checked 12% of the transcripts for intercoder reliability. Summary memos documented overall themes related to the study objectives.

Results
A total of 41 adolescents enrolled in the study and completed a baseline interview, of whom 38 started the intervention and 35 completed the endpoint questionnaire. A flow chart describing the identification, enrollment, and follow-up of participants is presented in Figure 4.
Several individuals initially deemed eligible according to clinical records were later found to be ineligible due to incorrect age information in their medical records. Of those confirmed to meet eligibility criteria, 13 parents or guardians refused to allow their child to participate, 5 of whom stated they had not yet disclosed their child's HIV status to him or her.
Participants were nearly equally divided by sex (Table 1). About half of the participants were enrolled in school, mostly in secondary school, and one-quarter were employed. All participants had a home or primary residence. Most participants had been on ART for >4 years; 90% (37 of 41) had at least one CD4 cell count test recorded in their medical record, but only 73% (30 of 41) had a viral load test recorded in their medical record (Table 1). Among those with a recorded viral load, half were virally suppressed ( Table 1). All participants who were not virally suppressed had viral loads >5000 copies/mL (data not shown).   Three-quarters of all participants lived with one or both parents, and the rest lived with other family members, including one participant who was married and lived with the spouse ( Table  2). The majority (68%) reported that at least one parent knew of their HIV status; fewer participants had other family members who knew their status, and only 3 (7%) reported that a close friend knew their status. Six participants stated they were in romantic relationships, and all but one participant reported that their partner knew their HIV status; 5 of the 6 participants knew their partner's HIV status.

Feasibility
At baseline, nearly all participants (93%) had existing access to a phone, and most (66%) ever used the internet (Table 3). Nearly all participants (96%) who used the internet had used Facebook. Among those who had phone access before this study, most reported difficulty charging phones (79%) or running out of airtime or phone credit (97%) sometimes or often.
Nearly all participants (92%) who completed the endpoint structured interview reported that connecting to Facebook was somewhat or very easy, and 97% stated that they wrote comments or asked questions during sessions (Table 4).

Participant Engagement
Most participants who completed the endpoint questionnaire (34 of 35) participated in the intervention sessions ( Table 5). The majority of intervention participants took part in all sessions and reported actively commenting or asking questions at some point.
Participation varied widely by session and group ( Figure 5). Within groups, participation varied considerably ( Figure 6). Each group had 1-3 members who were considerably more active than the others. Two participants made >100 posts (327 and 155 posts; Figure 6).     One group (Group 5) was the most active group overall, and the group's facilitator was the most active among all facilitators (Figure 7). Similarly, the group with the lowest member participation had a facilitator with the lowest activity.

Challenges to Participation
Through the IDIs, we explored the challenges faced during implementation that might affect participation. Just over half of the 16 IDI respondents mentioned occasional problems with charging their phone or running out of data. About one-third of IDI respondents mentioned issues with cellular networks, including poor connectivity or slow data speed, of whom 2 participants lacked network coverage when they traveled and 3 had persistent network problems at home. One of them commented the following:   During the session on adherence, participants were asked to upload a photo of an adherence plan; however, none of the participants did so. Several IDI respondents described difficulties with photo uploads as the reason for not uploading an adherence plan photo. One other IDI respondent noted that the pictures did not always display correctly on his phone.
One IDI respondent said he thought another group member was sharing information publicly from the Facebook group, although he did not state that his HIV status or that of others was compromised:

Fidelity to Design
Overall, each of the facilitators for the 5 groups posted most of the scheduled activities and did so on time (Table 6). Quizzes (polls) aimed to assess knowledge and stimulate discussion proved problematic. IDI respondents noted that quizzes did not appear correctly formatted on their phones. In response, facilitators posted quiz questions and response options as comments.

Intervention Acceptability
All participants who completed the endpoint questionnaire agreed that the intervention was useful: They enjoyed taking part in the Facebook group, felt comfortable with the facilitator and other group members, and wanted to continue to participate in the group (Table 7). In addition, participants unanimously felt that the Facebook group intervention was a good way for ALHIVs to interact and would recommend the group to other ALHIVs.
When asked to describe what they liked about the intervention, IDI respondents, both low and high engagers, most commonly reported that they enjoyed the intervention because it was educational or informative. They enjoyed "chatting," sharing experiences, and communicating with other ALHIVs; it helped them take care of themselves and their health (eg, most commonly, taking drugs on time) and felt supported or encouraged: A few participants said they did not like low participation from other group members: There's something I didn't like because we were 8 in number in that group and anytime or sometimes when I go online, I will only see only one chat or sometimes I won't see anybody.

Future Directions
Participants had few specific recommendations to change the intervention. Recommendations from the endpoint questionnaires (Table 7, n=34) included the following: encourage group members to be more active (23%); increase the size of the group (14%); provide better instructions for Facebook (8%); and use of other social media, specifically WhatsApp (5%). Two participants thought some group members may require more explanation on using Facebook in terms of etiquette for posting comments. When asked whether both sexes should be included in future groups, 94% of endpoint questionnaire respondents said yes (Table 7). Just over half the participants (55%) favored inclusion of older youths, but several wanted to cap the age at 19 years. Participants who wanted to include older youths said younger participants could learn from older participants; those who preferred the current age range said younger ALHIVs may be afraid or less open to interacting with older youths. When asked whether they would prefer to be part of an online support group, in-person support group, or both, 58% of participants preferred an online group only and 41% preferred both groups. None of the participants preferred an in-person support group without the online group.

Discussion
This online support-group intervention was feasible to implement and highly acceptable among the ALHIVs who participated in the study. We obtained important information on specific challenges and ways to improve the intervention in order to enhance delivery and participation.
The level of active participation varied, within and across groups, which is similar to in-person support groups [21]. Group participation appeared to correlate with facilitator activity, although other factors such as logistic issues also played a role in individual participation within groups. Notably, the facilitator with the lowest overall participation performed the worst on timely and complete posting of sessions. In the future, experienced, dynamic facilitators with adequate mastery of intervention delivery should be recruited; however, even participants with low active participation reported high levels of satisfaction and appear to have benefitted from the intervention. With social media, the ability to follow conversations may allow those who are not comfortable commenting, to learn from others in the group [39,40]. Basic literacy was an important factor to the success of the intervention. Education levels are relatively high in southern Nigeria; 75.9% of women aged 15-49 years have completed some secondary school or higher education [41]. As such, implementing this intervention for individuals with low education levels could prove more difficult.
The time taken for identification and enrolment of adolescents through health facilities was longer than anticipated. Although Nigeria bears the second-highest burden of HIV on the continent in terms of absolute numbers of people infected [42], its prevalence is relatively low, with higher pockets of concentration in certain geographic areas [43]. According to the recently conducted AIDS Indicator Survey in Akwa Ibom State (2017), HIV prevalence among adolescents aged 15-19 years is 1.5% [44]. Facility staff attempted, without success, to reach several patients through their recorded contact information, which highlights the challenge of poor retention in health care services among this age group. Although inclusion of fewer eligible participants than anticipated complicated study enrollment, it supports the potential role of an online intervention that does not require individuals to live near each other or travel to a designated location and could connect people across a broader geographic area.
We also faced a few challenges during implementation of the study. Considering the limitations of the study phones, future interventions should use smartphones, which are increasingly popular and affordable, even in low-and middle-income countries [45]. Network coverage problems also limited use of the intervention among some participants. Despite these challenges, most subjects participated in most sessions.
Surprisingly, all participants who voiced an opinion preferred online support groups, either alone or in combination with an in-person group, but none of them preferred in-person groups alone. The global trend toward use of and comfort with social media among young people seems to have reached youths in periurban and rural southern Nigeria [46]. This study had a few limitations. The nonprobability sample prevents generalization of the results beyond the study sample. Social desirability bias may play a role in participants' responses to questions about the intervention. In addition, we adapted only 5 of 14 sessions from Positive Connections; therefore, we do not know if participation may change with a longer curriculum. Moreover, we could not determine if facilitators drove participation or group member participation drove facilitator engagement.
In conclusion, this feasibility study demonstrated that an online support group intervention was both feasible and acceptable among ALHIVs in southern Nigeria. Our results provide guidance on changes required to enhance participation. Based on these findings, we have adapted the intervention and expanded it to include the remaining Positive Connections topics. A subsequent randomized controlled trial will test the cost-effectiveness of this revised intervention to improve HIV treatment outcomes.

Acknowledgments
The authors would like to acknowledge the contributions of the United States Agency for International Development (USAID)-funded Strengthening Integrated Delivery of HIV/AIDS Services (SIDHAS) project for successful implementation of the intervention under study. Drs Hadiza Khamofu and Edward Oladele of FHI 360 ensured the coordination of intervention implementation with the SIDHAS project. Dr Oluwasanmi Adedokun from FHI 360 provided key information and direct support for selecting study sites and informing sampling through the use of the SIDHAS program data. Dr Ogha Okpokam from FHI 360 played a pivotal role in coordinating the logistics for study-related trainings and liaising between the health facilities, the study team, and the local partners. We would also like to acknowledge the CPRH study team and their efforts to successfully carry out data collection and enroll study participants in the intervention. We especially would like to acknowledge Ms Hannah Silverstein from the University of North Carolina for her creative efforts in illustrating the intervention. Finally, we would like to thank the facilitators and adolescents who participated, without whom this research would not have been possible.
This study was funded by the USAID through the US President's Emergency Plan for AIDS Relief, under task order contract number AID-OAA-TO-15-00003 and YouthPower Action, under IDIQ contract number AID-OAA-I-15-00009, YouthPower: Implementation. The contents of this publication are the sole responsibility of FHI 360 and do not necessarily reflect the views of the USAID or the United States Government.