A youth-centred approach to improving engagement in HIV services: human-centred design methods and outcomes in a research trial in Kisumu County, Kenya

The design team

The design team comprised AYA workshop facilitators, the research team, and a design expert. The lead youth facilitator (28 years old) and two co-facilitators (24–26 years) comfortable discussing their HIV status with other youth were selected among adolescent HIV peer leads and collaborators from the study health facilities. Consistent with HCD mindsets,27 the AYA facilitators (two male, one female) demonstrated good group facilitation skills, flexibility, optimism and creativity. The research team included individuals from Kenya and the USA with expertise in paediatrics, adolescent sexual and reproductive health, HIV medicine, nursing, public health, social science and qualitative research; the design expert was from the USA with extensive experience conducting HIV research and design work in SSA. All team members were compensated for their role and are co-authors. Team composition was driven by the goal to blend content expertise with in-depth understanding of local communities. Co-creation was guided by AYA workshop participants.

Design team training

Led by the design expert, the design team engaged in collaborative HCD training, preparation and implementation over 7 months prior to the workshop. Biweekly didactic trainings were held over Zoom from July 2020 to October 2020. Topics included A4A study aims; HCD concepts, analysis approaches and practical application of HCD as a research tool; codevelopment of workshop objectives and activities; facilitation skills; adolescent health and HIV; confidence, and team building. Three practical, in-person trainings among Kenya-based team members covered practice workshop activity facilitation followed by full team debriefs. Importantly, this fostered formal reflexivity as the design team discussed their perspectives on the HCD questions prior to working with the AYA participants in the workshop.

Discover phase (March 2020–April 2020)

Drawing on: (1) knowledge and experience from delivering patient engagement and other HIV interventions in the region, (2) existing data from prior studies conducted by this research team and (3) extant literature, the research team mapped out key insights related to the A4A-proposed interventions: E-NAV, IP-NAV, CCT.28 29 The team systematically brainstormed and documented answers to the questions: ‘what do we already know?’ and ‘what do we need to learn more about?’ to formally reflect on perceived areas of strengths and potential gaps in how the interventions might improve AYA HIV treatment engagement. The areas identified for further learning represented design opportunities for exploration in the Define phase.

Define phase (April 2020–September 2020)

Building on the discover phase, the research team identified specific design challenges (HCD research questions).30Through discussion, we generated 18 ‘how might we…?’ questions targeting anticipated intervention effectiveness gaps.30 Using grouping, rating and ranking according to significance and relevance of knowledge gaps, the research team selected six initial core questions to address during the workshop (table 1). In consultation with the design team, including the AYA workshop cofacilitators, the HCD expert developed participatory design activities for each question to engage AYAs in discovery, ideation and co-creation during the workshop and created facilitation guides for each activity. She drew on underlying qualitative, participatory and design methods to create activity options tailored to the design opportunities and context (table 1). The Kenya-based design team reviewed activity options, selected the most preferred, and ensured contextual and cultural appropriateness. Example activities included journey maps, role plays, storytelling and ‘tomorrow’s headlines’ (table 1). The facilitators internally pilot tested and refined activities before the workshop. Questions and activities were influenced by FGD and DCE work through overlap of research team staff conducting data collection and analysis and iterative discussion of results as they were available.

Develop phase (October 2020–December 2020)

Our primary co-creation activities to improve the eNAV, IP-NAV and CCT interventions were conducted during a 4-day design workshop held in October 2020. Specifically, the design workshop aimed to (1) build empathy and understanding around engagement and viral suppression between the study team and AYAs; (2) co-create solutions to anticipated HIV retention challenges through adapting the planned A4A study interventions and (3) identify AYA-led priorities to which the study should respond, iterating on the discover and define phases. The HCD workshop was AYA-led with adult co-facilitator support.

Participant selection

HCD workshop participants included (1) purposefully sampled AYAs willing to participate and discuss their HIV status and (2) peer navigator implementers of the e-Nav and IP-Nav interventions, who were also AYAs openly on ART. The first group were drawn from the three study clinics and AYA study FGD participants. We purposefully sampled for variation in gender, age and childbearing history.31 Trained research staff members contacted eligible participants by phone or at clinic and conducted verbal informed consent procedures for interested AYAs and assent for caregivers of AYAs <18 years old.

Data collection, real-time analysis and documentation

Workshop facilitators led the design activities, which were adapted in real-time based on participant feedback (online supplemental appendix A : workshop agenda) and yielded visualised outputs (e.g., pictures, collages, etc) (online supplemental appendix B: photo images of workshop activities). On completion of each activity, facilitators fostered large group discussions using activity-specific guiding questions to support reflection, articulation of design or prototype-refinement ideas based on the activity results and synthesis, written on flip charts. Participants wrote down their thoughts on the most important take-away from each activity (participant insight statements). A qualitatively trained researcher took workshop ethnographic field notes.32

To assess participant workshop experience and acceptability, we conducted a 10-question, anonymous, individual participant survey self-administered on a tablet computer at workshop conclusion.

Deliver phase (December 2020–April 2021)

Informed by conclusions drawn from the first three HCD phases including FGD and DCE results, the full design and trial investigator teams reviewed and brainstormed trial adaptations, prioritised actions and noted concerns not addressed by the trial but critical to share with other stakeholders. High-level trial adaptation decisions were made through dialogue and trial investigator team consensus, led by principal investigators. Modifications are being tested, in the A4A trial that started in 2021 and will conclude in 2024 (figure 1). Peer navigator HCD participants were among the trial intervention implementers, allowing for ongoing reflection and iteration on intervention standard operating procedures and delivery during formal Navigator debriefs, refresher trainings and administrative study reviews.

Request permission

During workshop analysis

After each workshop day, the full design team held structured debriefs over Zoom lasting approximately 1–1.5 hours. These sessions included (1) General reflections and ways to improve the workshop in subsequent days, (2) For each activity: rapid review of activity-related data; documentation of high-level insights guided by questions such as ‘What seemed important?’, ‘What surprised you?’, ‘What confirmed what you already knew?’, ‘What was consistent?’, ‘What was confusing or conflictual?, (3) Rapid identification of design opportunities using brainstorming, affinity clustering and dialogue,33 (4) Reflection on emerging AYA-led priorities requiring further exploration and (5) Agreement on subsequent day agenda revisions to ensure responsiveness to emerging priorities and design insights. While priority ‘How might we…?’ questions guided initial activity plans, insights and priorities for intervention adaptations were not restricted to specific activities or questions. Key insights, new questions and trial adaptation ideas were informed by the entirety of the workshop activities and dialogue, including ongoing empathy development throughout. Debrief results were summarised in a report. Initial findings were shared with AYAs during the workshop for feedback and refinement.

Postworkshop analysis

Postworkshop, the full design team held two analysis sessions over Zoom lasting approximately 2 hours each. Rapid thematic analysis was conducted through iterative workshop data review, dialogue and affinity mapping.34 Clustered themes were categorised to elucidate higher level insights. Results were compared with ethnographic field notes for confirmation and identification of omissions. Results and interpretation were influenced by FGD and DCE results through overlap in study staff across data collection activities and formal discussion. Thematic insights were documented in translation tables applying abductive thinking35 to systematically identify trial intervention adaptations and linking in-workshop results.36 The research and facilitation teams iteratively contributed adaptation ideas for each of the key insights through in-person meetings and asynchronous contributions to a document over approximately 2 months.

Electronic navigation

The proposed E-Nav intervention involved electronic communication (SMS messages, phone calls) to provide peer support for AYA engagement in HIV care. The codesign workshop revealed that AYAs preferred less frequent communication than initially planned. AYAs concerns included message fatigue and competing school demands. We modified the study protocol, reducing communication frequency between E-NAVs and AYAs from weekly to every 2 weeks for the first 2 months of intervention assignment, then monthly, with an option for additional sessions as desired by the individual participant.

AYAs emphasised that confidentiality and privacy were critical to navigator communication acceptability. AYAs desired text messages that: (1) were not traceable, eliminating the possibility of accidental or intentional discovery of the message origin and (2) used coded language (eg, avoided words such as ‘HIV’, ‘viral load’, ‘clinic appointment’). Further, through conversations about the importance of AYAs being seen holistically, beyond the lens of HIV, AYAs expressed a desire to receive messages related to life priorities and events outside of HIV. For example, encouragement if they performed poorly on a test at school or sending birthday wishes. We revised the HCD workshop agenda to include participants creating actual messages e-Navigators might send to participants covering the range of topics and types of language they might appreciate. Additionally, we modified our text messaging architecture to ensure messages sent were not traceable and harmonised message content with preferences.

Our original trial eligibility criteria excluded AYAs without access to mobile phones, (eg, in boarding schools where phone access is curtailed) because of seemingly critical logistical barriers to the phone-delivered E-Nav arm. However, workshop findings demonstrated the exceptional characteristics and circumstances of AYAs who have limited access to mobile phones, underlined the importance of including them and problem solved options for engaging them such as youth-led connections to trusted boarding school educators with phones or communicating during permitted phone access windows. However, it was noted that delivery of E-Nav would remain limited in this subgroup as biweekly or even monthly communication would be impossible. Our study protocol was modified to include AYAs without mobile phone access, starting them in the control arm which allowed them to be eligible for re-randomisation to IP-Nav or CCT.

In-person navigation

The proposed IP-Nav involved in-person sessions with peer navigators to influence retention and adherence behaviours among AYAs. Acknowledging that not all peer navigators will necessarily be influential to all AYAs, the workshop identified opportunities to improve positive peer navigator influence. Codesign workshop participants highlighted the importance of feeling appreciated and having non-judgemental navigators who are vested in the preferences, experiences and goals of AYAs. They also desired empathetic navigators who can ‘walk with them’; and are open about sharing their lived experiences with HIV treatment. Additionally, they preferred navigators who can comfortably discuss health topics other than HIV, such as sex and sexuality, family planning, life skills and relationships. They also expressed a strong preference for navigators who demonstrate a high level of personal hygiene and neatness. Many participants expressed a desire to have IP-Navigators share informal recreational time in addition to specific HIV counselling (eg, playing a sport together or meeting caregivers/dependents). However, some older participants strongly preferred limited interaction time to accommodate busy schedules.

In response, the pretrial peer navigator training modules were modified to further emphasise active listening skills, facilitation skills, motivational interviewing, support discussion of non-HIV or health-related interests and sexual reproductive health and relationship education modules with appropriate referral options. The navigator training was adapted to encourage openness, empathy and equip them with the story-telling skills to share treatment experiences and challenges as learning points, and to emphasise the importance of professional appearance when meeting with AYAs. Based on prior peer educator experiences, the research team felt that encouraging external unstructured interactions outside of study procedures could be problematic in terms of maintaining professional boundaries and future scalability. Therefore, this recommendation was not adopted.

Additionally, participants preferred IP-Navigators of the same gender, particularly in boarding schools with gender-segregated environments that extend to visitors. AYAs preferred minimising IP-Navigator changes once assigned. Gender-matching was previously planned; We amended study procedures to avoid navigator switches whenever possible.

Across e-NAV and IP-NAV, the importance of navigators building trust with the AYAs and demonstrating themselves to be trustworthy arose. This included navigators supporting AYAs and in developing the skills needed to self-manage their HIV care, which was then included in Navigator training and re-emphasised during debriefs and annual trainings.

Conditional cash transfer

Realtime feedback on CCT prototypes largely endorsed the original plan including amount, delivery method and timing. However, two themes related to CCT arose. Some AYAs expressed a strong desire for confidentiality in who received the incentives, while others felt it was okay for their parents and guardians to know if they received an incentive. AYAs advocating confidentiality asserted, ‘funds should be given under water’ (ie, given in secret). Flexibility was consistent with the original study protocol, allowing incentives to be given either directly to the AYAs or to someone the AYAs designated. Caregiver notification was not required (though participants under 18 years had caregiver study consent which explained cash incentives may occur). In response to this topic, pretrial training materials around incentives further emphasised discretion. Additionally, there was a strong sentiment among HCD participants that viral suppression should not be incentivised, undercutting the study intent. Participants were concerned that: (1) missing an opportunity to receive an incentive would add shame and a feeling of punishment to the stress of having an unsuppressed viral load, (2) if friends knew they were in the study and were planning to use the extra funds in a social setting, not having the funds would reveal their unsuppressed status and cause social discomfort and (3) some viral failure may be outside of the control of the participant if related to drug resistance and would be unfair to withhold funds. Participants felt incentivising clinic visits would be more appropriate. The study team discussed the HCD results with care and attention. Ultimately, it was decided that the potential benefit of improving viral suppression with incentives, supported by extant literature but untested in this population, was worth investigating. Measures to attend to the very real concerns of the AYAs included training and documentation procedures to ensure incentive distribution was private and not discussed among study staff members or outside of the study and incentives were based on a combination of viral suppression and clinic attendance; further, questions around confidentiality and shame were added to the post-trial FGDs around the participant CCT experience to continue learning and improve future CCT use.

Personalisation

Throughout the workshop, a strong theme around the desire to have interventions tailored to individual preferences as much as possible was present. For example: choosing cash or mobile money for CCT receipt, timing of SMS message receipt in e-Nav, and choosing optional additional interactions or content areas for discussion in IP-NAV. In response, we modified our study standard operating procedures to include flexibility when possible and the navigator training curriculum to build navigator capacity to explore AYA preferences.

AYA empowerment

Participants expressed a desire to improve interactions with clinic-based, adult healthcare workers. In response, Navigator training was revised to include assertiveness trainings and role plays on how navigators can build AYA capacity in patient–provider communication, expressing their challenges and asking targeted questions to better understand and self-manage their health.

AYA awareness of HIV status

During the workshop, AYAs emphasised the importance of being aware of their own HIV status. Specifically, they expressed concerns about not being informed of their HIV status and believed that healthcare workers and caregivers should be responsible for disclosing this information to them rather than leaving it for them to discover on their own. Knowledge of status is critical to a young person taking responsibility and garnering support. While facilitating disclosure directly was outside of the scope of the study as disclosure was an eligibility criteria to enrol, the study team included these findings in feedback presentations to the clinic on study results and revised the navigator training to include postdisclosure support skills.

Drug holidays

AYAs described ‘drug holidays’ as periods when they decide to stop taking their ART for certain durations and for reasons that include school schedules, stigma (fear of others seeing them take their medication), peer pressure, visitors at home, food insecurity, a feeling of good health, ‘no reason’ and pill burden. The research team was aware of drug holidays but was surprised by how this term was widely applied to normalised non-adherence. Drug holiday assessments were added to navigator training and study intervention forms. Navigators were trained to provide non-judgemental support for AYAs who discussed drug holidays or intermittent medication adherence, as well as counselling around benefits for all AYAs (including the navigators themselves) to maintaining continual adherence.

Importance of transition in AYA retention

Through 18 years of age, routine services at the health facilities are provided through a youth-friendly adolescent centre. When AYAs>18 years transitioned to the adult clinic, they described an impersonal, busier, setting with few peer interactions. The challenge of this transition was raised by participants and emphasised as a critical concern. During an added session on this topic participants expressed a strong desire for more support to help AYAs prepare emotionally and better understand how to navigate the adult clinic. Solutions included AYAs being informed about the need to transition prior to age 16 years and transitioning with a group of peers to ease the shift. While not an outcome of the A4A study, the research team opted to integrate training on transition and incorporate transition issues in Navigator intervention sessions. Further, the team shared these findings during feedback presentations to the study clinics.

Youth-led HCD workshop acceptability

Engaging AYAs as codesigners in a workshop-based approach was feasible and well-received by participants. Twenty-one of the participants completed the postworkshop survey, with 100% indicating the workshop met their expectations. The workshop content, mix of activities and quality of the facilitation team was rated ‘outstanding’ (means: 4.6, 4.4 and 4.4, respectively, on a 1–5 Likert scale). All but one participant reported they would recommend the workshop to others. Different workshop sessions appealed to different AYAs, with at least 20% of participants rating each of the eight main activities as one of their top three favourites. Elicited specific improvement suggestions included having more time for the workshop, an improved venue, reducing the amount of writing participants did and, most frequently, ‘no changes’ (n=10). The workshop produced results relevant to each of the study ‘how might we…?’ questions and allowed for participant-driven inquiry.