Scale-up of a paediatric HIV disclosure intervention in Ghana using a stepped wedge cluster randomised trial design: Sankofa 2 protocol and implementation

Abstract

Introduction

There is a pressing need for effective interventions that can support healthcare workers and caregivers in the challenging yet crucial task of disclosing the HIV status to infected children and adolescents. Previously, we developed and tested a successful disclosure intervention called Sankofa in Ghana. In an ongoing 5-year follow-up study, Sankofa 2, we aim to build on the successful Sankofa trial by testing the intervention on a larger scale.

Methods and analysis

This study is a pragmatic, stepped-wedge cluster randomised trial.

Setting

It is being conducted in 12 HIV paediatric clinics in Ghana to examine the effectiveness, health benefits, cost and implementation of the Sankofa intervention. Caregiver–child dyads (n=700) will be enrolled. Evaluation of effectiveness, health benefits, cost and implementation of the Paediatric HIV disclosure intervention, Sankofa 2, is posed to offer valuable insights for scale-up and sustainability.

Ethics and dissemination

Ethical clearance has been obtained from the Ghana Health Service Ethics Review Committee, the University of Ghana Ethical and Protocol Review Committee, the Committee on Human Research Publication and Ethics of the Kwame Nkrumah University of Science and Technology, the Johns Hopkins Medicine Institutional Review Board and the Yale School of Medicine Human Investigation Committee. The clinical trial was registered on ClinicalTrials.gov on 5 March 2021. All caregiver participants are required to provide written informed consent and the children assent before enrolment. If either the child or caregiver says no to the study, the dyad is not eligible for the study. No study-related procedures are performed until consent is obtained. The results of the trial will be added on ClinicalTrials.gov, published in peer-reviewed journals and presented at international conferences.

Trial registration number

NCT04791865.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The Sankofa 2 study builds on a previously successful intervention trial, aiming to test its effectiveness in broader, real-world conditions across multiple clinics, which can provide valuable insights into scalability and sustainability.

• The study uses a pragmatic, stepped wedge cluster randomised trial design, which is appropriate for evaluating both the effectiveness and implementation of the intervention, and also accounts for the natural history of disclosure and factors affecting enrolment.

• The study is comprehensive, incorporating the assessment of various outcomes, including disclosure rates, health benefits and costs, and it uses a mixed-methods approach for implementation evaluation.

• Complete blinding is not possible in the stepped wedge design, which could introduce assessment bias despite measures taken to blind outcome assessors and investigators.

• The study’s reliance on delivery of the intervention by specially trained staff in each participating clinic may present challenges in terms of standardization and fidelity in diverse real-world clinical settings.

Introduction

Effective interventions are needed to support healthcare providers and caregivers in the complex, but crucial process of disclosing HIV status to infected children and adolescents. Both the American Academy of Pediatrics and the WHO strongly recommend disclosure to school-aged children and older, citing ethical obligations and clinical benefits.^{1 2} Yet, globally, and especially in sub-Saharan Africa, disclosure to children of any age continues to be a significant challenge.³ Despite being home to 85% of the world’s HIV-infected children, as many as 60–91% of children in sub-Saharan Africa have not been informed of their HIV status.³,⁶

Barriers to disclosure are multifaceted. Parents, caregivers and healthcare providers often express uncertainty about how to approach disclosure, concerns about a child’s cognitive and emotional readiness, fear of potential negative effects on family relations and apprehension about stigma and social rejection.^{3 7} Although research indicates that disclosure can lead to improved antiretroviral adherence, better psychological adjustment, enhanced clinical outcomes and reduced HIV transmission risk during adolescence, these benefits have not translated into widespread practice.^{3 8}

Implementation remains limited due to inadequate provider training, absence of culturally tailored, standardised procedures and guidelines, time constraints in busy clinical settings and persistent stigma.³⁷,¹⁰ A few interventions have been evaluated in the region, including a caregiver training video in Uganda, structured disclosure tools and peer support in Kenya and child-centred educational materials in Botswana, but most were tested in a single clinic, have lacked rigorous evaluation over time or were not scaled beyond preliminary testing.⁹¹¹,¹³

In Ghana, an estimated 18 000 children aged 0–14 years were living with HIV as of 2024, with disclosure rates remaining suboptimal.¹⁴,¹⁶ These figures underscore the critical need for interventions tailored to the paediatric population. We previously developed and tested a practical, clinic-based disclosure intervention that is delivered as an integral component of routine care in Ghana (R01 HD074253).^{710 15},¹⁹ Disclosure in this context refers to informing the child of their HIV-positive status. We distinguish between partial disclosure (where the child is told they have a health condition requiring medication, but not the HIV diagnosis) and full disclosure (where the child is explicitly informed they have HIV).

The intervention, ‘Sankofa’, is guided by an HIV paediatric disclosure model that incorporates bioecological systems theory and core elements of the Information–Motivation–Behavioural Skills (IMB) model of Health Behaviour Change.^{7 20} The individualised counselling intervention is delivered by a trained clinic staff member during routine clinic appointments. It targets modifiable IMB skills of caregivers to facilitate their engagement in the process of disclosure in a manner suitable to the context, developmental stage and needs of the child.⁷

In a trial of 446 child–caregiver dyads (n=240 treatment group; n=206 control group), we found that the Sankofa treatment group had greater disclosure at each time point (3, 6, 9 and 12 months) (p<0.001) at follow-up.¹⁶ At 1 year (primary endpoint), 51.4% of children in the treatment group had been disclosed to versus 16.2% in the control (p<0.001). While highly promising, the Sankofa intervention has only been examined at a single academic hospital clinic under optimal conditions. Thus, in Sankofa 2, we aim to build on the successful Sankofa trial by testing the intervention in Ghana in a larger-scale study delivered in a range of ‘real-world’ clinic conditions, defined as routine government clinic settings with typical staffing, infrastructure and resource constraints. Specifically, we aim to: (1) test the effectiveness of the disclosure intervention at scale in routine care settings; (2) evaluate health and psychosocial benefits among children and caregivers; (3) assess the cost-effectiveness of the intervention; and (4) identify implementation facilitators and barriers across diverse sites. If effective, this intervention has the potential to inform national policy and scale up as a sustainable model of paediatric disclosure in Ghana and other low-resource settings, thereby improving antiretroviral therapy (ART) adherence, mental health and long-term care engagement.

Methods and analysis

Study aims and design

In the 5-year Sankofa 2 study (R01HD103512-02), we are conducting a pragmatic, stepped wedge cluster randomised trial in 12 HIV paediatric clinics in Ghana to examine the effectiveness, health benefits, cost and implementation of the intervention to inform scale-up and sustainability. In this design approach, groups of participating sites are randomly assigned to a sequential introduction of the intervention.

Such designs are increasingly used to promote the implementation of evidence-based practice and are appropriate for simultaneously evaluating the effectiveness and implementation of interventions. Moreover, the varying durations of control and intervention periods at the various sites will help us appreciate the natural history of disclosure as well as factors that affect entering the study, allowing child–caregiver dyads to be exposed to either the standard of care alone (a preswitch control period) or intervention alone (postswitch intervention period) or the combination of the two, depending on the time of enrolment relative to the switch schedule.²¹,²³

Our hypotheses are: H1: The intervention will improve the probability of HIV paediatric disclosure at 1 year postintervention initiation (primary outcome). H2: The observed predictors of disclosure (eg, child’s age, caregiver knowledge, skills) will be consistent with the conceptual model. H3: The intervention-assisted disclosure will result in favourable health and cost outcomes. Research questions are: R1: To what extent is the intervention implemented and maintained? R2: What are facilitators of and barriers to implementation? Study methods are described per SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) reporting guidelines.²⁴

Study setting and epidemiological context

Ghana has 16 main regions and 261 administrative districts (https://www.citypopulation.de/en/ghana/admin/). Although the HIV epidemic in Ghana is generalised with a relatively low but stable prevalence estimated in 2022 at 1.7% within the general population, there are regional differences in the prevalence.^{9 25} The Greater Accra and Ashanti regions are among the regions with the highest prevalence of HIV. These two regions have the highest population density in Ghana with similar demographics and amenities. The study is being conducted at district-level clinics in the Greater Accra region (6 out of 16 treatment centres in Accra District) and Ashanti region (6 out of 8 treatment centres in Kumasi district). Approximately one-third of children aged 7–18 years living with HIV in the two regions reside in these two metropolitan districts (Accra and Kumasi), with a male-to-female ratio of 0.9:1.0 (https://www.ghanaids.gov.gh/).

Study entry

Participants (child–caregiver dyads, n=720) are being recruited through participating district-level clinics. All children infected with HIV, aged 7–18 who are receiving care at the study sites and who do not know their HIV diagnosis (based on caregiver account and medical records confirmation), and their caregivers are eligible to participate in the study. Children with congenital or developmental disorders, comorbidities such as sickle cell disease or diabetes that require frequent clinic visits or hospitalisations, AIDS-defining illness or end-stage AIDS and age <7 or >18 years of age are excluded. The purpose of the study is explained to participants. After agreeing to participate in the study, the parent/caregiver provides written informed consent, and the children give assent before enrolment. If either the child or caregiver says no to the study, the dyad is not eligible for the study. Participants have the opportunity to refuse or withdraw from the study at any point.

Stepped wedge design and randomisation

The 12 participating clinics, with equal representation from each region, were chosen based on Ghana Health Service (GHS) routine data identifying the highest paediatric HIV caseloads. To ensure adequate numbers of eligible participants, the highest-volume sites in the two most affected regions were included.

Groups of three clinics were randomly assigned to one of four intervention roll-out waves (‘wedge steps’) (figure 1) using computer-generated random numbers. Randomisation was conducted by an independent statistician with no role in implementation.

Figure 1. Stepped Wedge Design. 24 dyads recruited per month (2 dyads per each clinic) for 30 months = 720 dyads recruited by month 30; Follow up Months 31-42 limited to dyads recruited during Months 25-30 (2 dyads per month x 12 clinics x 6 months = 144 dyads).

Groups of three clinics per step are switched to the intervention such that each region is equitably represented across time and intervention/control periods. Each clinic will contribute at least one 6-month control/pre-intervention period and at least one 6-month intervention period (shown in dark gray). There is no washout period; the transition is managed operationally by trained counselors during the final month of the control period, with intervention delivery beginning the following month. Implementation is carefully staged to avoid overlap.

Enrolment is targeted at two child–caregiver dyads per month per clinic, with approximately equal distribution of children aged 7–10 years and >11 years of age. Using the available information on the number of patients receiving care at each participating clinic and the current disclosure rate of approximately 21% in the two regions, we should have a sufficient number of eligible participants to be enrolled at the rate of two dyads (child and legal guardian) per clinic per month over the 30-month active enrolment period.

Blinding

While complete blinding is not feasible in a stepped wedge design, the outcome assessors are unaware of which group or cluster is receiving the intervention at any given time and study investigators will be blinded to study outcome data until data collection is complete.

Retention strategies

All participants will be reminded of upcoming study visits by phone at 1 month, 2 weeks, 1 week and the day before the appointment. Those who cannot make scheduled appointments will be rescheduled at the time of the phone call. Members of the study team will visit the homes of participants who miss their visits and cannot be reached by phone. With the permission of the parent or guardian during these home visits, participants will have study assessments done. We will also fast-track the experience and flow of participants during clinic visits to reduce the burden of waiting time. We recently ended the Sankofa 1 trial (R01 HD074253) that examined the efficacy of the HIV disclosure intervention in one of two participating clinics. Participants (446 caregiver and child dyads) were randomised to either intervention or treatment as usual, and the loss to follow-up rate was only 3.3% (n=15) in 5 years. We, therefore, anticipate that retention is not likely to be a problem with the Sankofa 2 study.

Description of the study arms

Control: usual care period

During the control period, children and caregivers recruited at the clinics receive the current practice in the clinic, where the healthcare provider is expected to assess the caregiver and child’s readiness for disclosure during clinic appointments and provide health maintenance guidance. There are no formal paediatric disclosure guidelines or educational materials used in Ghana.

Intervention: usual care plus disclosure intervention period

A manualised Sankofa intervention is delivered during the intervention period. The intervention manual includes session-by-session guides, age-specific disclosure scripts and case scenarios. It was developed iteratively based on stakeholder input and revised following Sankofa 1 pilot feedback.⁷

The intervention is guided by a disclosure model (described fully elsewhere) and contains two key elements to target well-documented, modifiable barriers to promote disclosure.⁷ The first component is the adherence and disclosure counsellor. The counsellor is selected from the current HIV clinic staff (eg, nurses or social workers) by the clinic leadership based on predefined criteria, including at least 1 year of experience in paediatric HIV care, fluency in the local language and strong communication skills. The counsellor, who is familiar with the sociocultural norms of the community, is trained by an experienced disclosure counsellor to assist families in the process of disclosure (ie, predisclosure, disclosure and postdisclosure phases) (see box 1).

Box 1. Content outline of intervention process.

Predisclosure sessions

• The counsellor meets with the parent/caregiver to build rapport and start the process.

• The counsellor discusses whether the parent/caregiver has considered disclosure using Motivational Interviewing skills to explore reasoning.

• The counsellor is sensitive to and shows respect for the intense feelings the parent/caregiver may have about disclosure.

• The counsellor explores whether the parent/caregiver has an interest in learning more about disclosure (eg, possible benefits).

• The counsellor determines whether the parent/caregiver wishes to develop a disclosure plan.

• If the parent/caregiver is ready to disclose, the counsellor guides them in considering how to disclose (eg, ways of approaching disclosure, age-appropriate skills), possible response of the child and having a support system in place for and after disclosure.

• If the parent/caregiver is not ready to disclose, the counsellor schedules a follow-up meeting to discuss further and address any questions the parent/caregiver may have after reflecting on the prior discussion.

• The counsellor respects the parent/caregiver’s timing, but encourages the caregiver not to lie to the child about his/her diagnosis.

Disclosure sessions

• When the parent/caregiver is ready, the counsellor has him/her think through how they want the discussion to go.

• Encourages the parent/caregiver to choose a place where the child will be most comfortable talking openly.

• The counsellor provides the parent/caregiver with questions the child may ask so he/she is prepared to respond with open and honest answers.

• If the parent/caregiver is not comfortable disclosing alone, the counsellor can arrange to be present during disclosure or available by phone.

• At disclosure, information is provided to the child at a developmentally appropriate level and medical facts are kept to a minimum. The child needs to hear that he/she did not do or say anything to cause the illness and that the family/caregiver will always support him/her.

• The counsellor offers to give the parent/caregiver a call after disclosure and arranges a time for a follow-up visit.

Postdisclosure sessions

• Follow-up session with the parent/caregiver within 2 weeks following disclosure and regularly for 6 months to assess the impact of disclosure, to answer questions and to foster support between the parent/caregiver and the child.

• During the postdisclosure sessions, the counsellor will ask the child what he/she has learnt about the virus. This offers the counsellor an opportunity to clarify any misconceptions.

• The counsellor will assess changes in the child’s emotional well-being and offer additional support as necessary.

• The counsellor offers support to the parent/caregiver for having disclosed to the child.

• The counsellor reminds the parent/caregiver that disclosure is not a one-time event.

Our preliminary work has shown that caregivers have a clear preference for assistance with disclosure from a healthcare provider (as opposed to other possibilities such as Priest, head of family or teacher).^{4 7} Furthermore, the complexity of the disclosure process suggests that having a specific point person, rather than a team, is advisable. This allows for continuity and uniformity in approach and fosters the development of a trusting relationship between the caregiver and child.

The second component is disclosure as a process whereby the counsellor guides the intervention counselling sessions to the IMB Skills needs of the caregiver and the neurocognitive development of the child over time. The process has three main phases: predisclosure, disclosure and postdisclosure, as outlined in the HIV disclosure intervention schema⁷ (see box 1). During these individualised sessions, the caregiver learns the benefits of disclosure, how to disclose to the child in a developmentally appropriate manner, adjustment and coping after disclosure. Caregivers’ concerns about the negative consequences of disclosure are discussed, and coping skills, social problem-solving and communication skills developed with didactic and interactive techniques are discussed. For example, the counsellor will have a mock disclosure session with the caregiver, where all difficult questions anticipated from the child will be discussed. The date and venue of the disclosure are also discussed and individualised to caregiver preferences. If the caregiver chooses to disclose at home, the counsellor provides a phone number for the caregiver to call to register the exact date of disclosure and provide support to the caregiver if needed. The caregiver may elect to have a counsellor present at the time of disclosure. After disclosure, the counsellor continues to meet with the caregiver and child at frequent and scheduled intervals to assess postdisclosure problems and coping, provide referral to services that the family may need and continue to help the child to understand HIV infection in an age-appropriate manner, HIV infection and its implication on their day-to-day activities. The counsellor aims to meet with the caregiver during each clinic appointment during the predisclosure and postdisclosure processes for at least 6 months following disclosure. The sessions are individualised and can range from as little as 10 min to as much as 60 min, depending on the needs and stage of the caregiver in the process.

Intervention fidelity

To achieve consistency in the delivery of the intervention, we are employing several strategies. Standardisation is achieved using a scripted intervention protocol. The counsellor is trained to follow the protocol as designed. Standardisation of content and approach is facilitated with use of a manual. The counsellor receives competency training from an expert (counsellors who were trained and delivered the intervention throughout Sankofa 1). During training, the experienced study counsellors oversee the counsellors’ delivery until mastery of the protocol is demonstrated. In addition, counsellors maintain a log of each contact to assess intervention delivery, as described below. The quality and consistency of the intervention are addressed by ongoing on-site supervision provided by experienced counsellors from Sankofa 1. If fidelity issues are identified through supervision or logs, counsellors receive immediate feedback and participate in remedial training sessions. Sites with recurrent fidelity lapses are flagged for intensified support. In addition, the investigators conduct regular face-to-face or video conferencing sessions with the study coordinators and experienced counsellors as needed to actively address training and supervision issues.

Study outcomes

The primary outcome is disclosure after 1 year (48 weeks) of the intervention. Secondary outcomes are antiretroviral medication adherence, health (virological, immunological, psychosocial, behavioural), cost and implementation of the disclosure intervention. Table 1 summarises the outcomes and outcome measures for each aim.

Table 1. Summary of variables, measures, source and timing.

Variable	Measure and source	Timing
Rate of full HIV disclosure (primary outcome)	Chart review and project specific form.	Every 3 months; 12 months primary
Child medication adherence	ACTG Adherence Questionnaire. Pharmacy-based time to refill.	Baseline and every 3 months
Child HIV-1 RNA	Chart review.	Baseline and every 6 months
Child mental and behavioural health	Child Depression Inventory. Child Behavioural Checklist—parent and child versions.	Baseline and every 3 months
Caregiver mental health	Beck Depression Inventory.	Baseline and every 6 months
Cost-effectiveness variable	Project-specific form.	Baseline and every 6 months
Demographic characteristics (eg, age, gender)	Project-specific form.	Baseline
Caregiver knowledge, motivation, skills	Social Provisions Scale. Brief HIV Knowledge Questionnaire. Brief Illness Perception Questionnaire. HIV Stigma Scale.	Baseline and every 6 months
Implementation (acceptability, applicability, feasibility, fidelity, penetration, maintenance, context)	Log maintained by the ADDs for each encounter with the participant that details for each subject: date, length of visit, whether full intervention protocol was completed; content of visit; and, difficulties encountered and strategies used to manage them. Screening/enrolment forms and chart data. Participant questionnaires and semistructured interviews. Provider and ADDs focus groups.	Ongoing after intervention initiated Baseline and annual Every 6 months and exit Annual

Note: See Reynolds et al.(2015) for measure source

ACTG, AIDS Clinical Trials Group; ADDs, Adherence and Disclsoure counselor.

Data collection and management

Data are collected at baseline and follow-up with standardised instruments (table 1) selected based on our guiding conceptual model and successfully used in our prior study.⁷ Child and caregiver data are collected during routine clinic appointments which typically occur every 3–4 months, depending on patient stability. The primary outcome, disclosure, is assessed through chart review and a project-specific form developed in our prior study, which itself is based on caregiver and child questionnaires developed by Mellins et al.²⁶ Full disclosure is defined as the child being informed of their HIV diagnosis and demonstrating understanding of its meaning, as confirmed by caregiver and/or child report using a structured tool. Partial disclosure is tracked separately but is not the primary endpoint.

Table 1 summarises all primary and secondary outcomes, including disclosure status (chart-confirmed and self-reported), ART adherence (via pharmacy refill and caregiver report), clinical indicators (CD4, viral load), psychosocial outcomes (eg, depression, stigma) and implementation indicators. Secondary outcomes were selected to assess domains theoretically affected by disclosure per our IMB model—including health behaviour (adherence), clinical health (virology) and psychosocial adjustment. Instruments were adapted and pilot-tested for comprehension and cultural relevance during the initial Sankofa study in Ghana.⁷ Where available, previously validated Ghanaian versions were used; new instruments were cognitively tested and refined with local input.

Following accepted standards for health cost-effectiveness analysis (CEA), we are collecting measures of direct intervention costs, indirect costs and measures of health benefit, based on a public healthcare delivery system perspective and a societal perspective that includes caregiver costs.²⁷,²⁹

A mixed-methods approach will be used for the implementation evaluation guided by the recommendations of Proctor and colleagues for the measurement of implementation outcomes.^{30 31} Measures will include indicators of acceptability (the extent to which implementation stakeholders perceive intervention as agreeable, palatable or satisfactory), applicability (the perceived fit, relevance or compatibility of the intervention for the practice setting), feasibility (extent to which a new intervention can be successfully used or carried out within a given clinic setting), fidelity (degree to which the intervention (eg, content, dosage) was delivered as intended), penetration (extent to which the intervention is integrated within the clinic setting and its subsystems), maintenance (extent to which the intervention is maintained and/or institutionalised within the clinic setting’s ongoing, stable operations) and context (individual and clinic-level factors in the setting that facilitate or are barriers to implementation). Evaluation methods include (1) a log maintained by the clinic counsellor for each encounter with the participant that details each subject: (a) date; (b) length of visit; (c) whether full intervention protocol was completed; (d) content of visit; (e) difficulties encountered and strategies used to manage them; (2) structured random observations of the intervention components; (3) screening/enrolment forms and chart data; and (4) participant questionnaires, semistructured interviews and provider and clinic counsellor focus groups (eg, participants’ description of factors that were helpful or not in the delivery of the intervention, the counsellors’ description of the positive and negative aspects of the intervention).

All data collected from the study assessments are entered directly into Yale’s secure web-based system, OnCore (https://medicine.yale.edu/ycci/oncore/), which is Yale’s enterprise-wide clinical research management (CRM) system. OnCore is a fully integrated system developed by Forte, which includes three modules: (1) CRM, (2) biospecimen management and (3) patient registry. This project uses the CRM module as an electronic data capture system for all case report forms and to support subject tracking, which is US Health Insurance Portability and Accountability Act-compliant. Data are entered into the OnCore system by research staff. To ensure and maintain the scientific integrity of this human subject research project, and to protect the safety of its research participants, a Data Safety Monitoring Board with appropriate scientific and clinical expertise who are independent of the sponsor and investigators has been assembled to monitor the scientific integrity of the project (details may be found in full protocol available by request to National Institutes of Health or the project leaders Paintsil and Reynolds).

Statistical considerations

A closed-form sample size approximation for stepped-wedge design studies was used as the basis for the sample size calculation to examine the primary outcome.²³ Following that approach, we estimated an inflation factor based on our SWD specifications and applied it to the sample size needed for a parallel-group trial design. We assumed an intraclass correlation coefficient of 0.10, four intervention roll outs (switches), one baseline measurement, an average of two measurement times per dyad between switches and four dyads per clinic measured at each time point. These assumptions yielded an inflation factor of 2.5.

Therefore, since a sample size of 250 dyads is needed to achieve 90% statistical power to detect a HR of at least 2.0 in a parallel group design, the required sample size for our SWD would be 2.5×250 = 625 dyads. After adjusting for an anticipated 13% attrition rate, the final sample size required is approximately 720 dyads, to be recruited over a 30-month recruitment period (figure 1).

The primary analysis will use a Cox proportional hazards model with a frailty term to account for clustering at the clinic level. The intervention will be modelled as a time-varying exposure, and the HR for the intervention effect over the 48-week follow-up will be estimated with corresponding 95% CIs. Given the stepped-wedge rollout and terminal nature of the outcome (ie, disclosure at a specific time post enrolment), a supplementary analysis will employ a discrete-time survival model on a time-on-study scale, as recommended for stepped wedge designs.²¹

Effect sizes will be reported as HRs with 95% CIs. Although a two-sided alpha of 0.05 will guide significance testing, emphasis will be placed on the magnitude and precision of effect estimates (95% CI) rather than p values alone.

A priori-specified subgroup analyses will stratify outcomes by age group (7–11, >11 years of age) to reflect the intervention’s age-appropriate components, and by child gender, due to known gender-related differences in HIV outcomes and social determinants of health. Analyses will be implemented using SAS V.9.4 (Cary, North Carolina, USA).

Secondary outcomes are collected longitudinally and will be analysed using multilevel generalised linear mixed-effects regression models³² (with appropriate link functions for the mean response, eg, link=logit for binary outcomes, link=glogit or link=clogit for multinomial outcomes, link=Identity for continuous outcomes) with random intercepts for clinic and child, as well as a possible random slope for time (tested via the likelihood ratio test, LRT),³³ and fixed effects for time, treatment condition and relevant baseline covariates. The functional form (eg, constant or not-constant slope) of time (weeks on study) will also be tested using the LRT for nested models and Akaike Information Criterion³⁴ for not nested models. These models will allow us to assess main effects, time trends and treatment-by-time interactions. Magnitudes of associations will be summarised as estimated retrogression coefficients with 95% CIs and further transformed as needed (eg., exponentiated to obtain an OR).

Our approach to controlling the Type I error rate at 0.05 is three-pronged: (1) a priori identify the primary and secondary outcomes of interest and outline analytical considerations for each, (2) among the secondary outcomes of interest, the individual statistical hypotheses are considered as independent, thus no adjustment for multiple comparisons is needed and (3) a priori identify subgroup analyses and focus on interpreting 95% CIs for these analyses with particular attention to the coverage of clinically meaningful effect sizes.

The objective of the CEA is to evaluate the direct and indirect (opportunity) costs and health benefits of the paediatric HIV disclosure intervention compared with the status quo and standard of care. Cost-effectiveness will be assessed using the incremental cost-effectiveness ratio (ICER), calculated over 1-year, 5-year and 10-year time horizons.

The 1-year ICER will reflect the immediate, directly observed economic impact of the intervention, while 5-year and 10-year ICERs, estimated using Markov model-based projections, will provide insight into potential long-term economic impact.²⁶,²⁸ Intervention costs will include training, staff time, materials and supervision. Cost data will be collected prospectively using standard costing templates and adjusted using local price indices. Costs across clinics will be reported in both US dollar and Ghanaian cedi, standardised using purchasing power parity.

Health effects will be measured in disability-adjusted life years averted, and ICERs will be computed in accordance with WHO guidelines. Sensitivity analyses will be conducted to assess cost variability across sites and scalability of the intervention.

Progress to date

The trial is being conducted with a strong team of interdisciplinary investigators at the University of Ghana (UG), Kwame Nkrumah University of Science and Technology (KNUST), Yale University and Johns Hopkins University, in partnership with the Ghana Ministry of Health (MOH), GHS and a community advisory board. Ethical clearance was obtained from the GHS, UG, KNUST, Johns Hopkins University and Yale School of Medicine Human Investigation Committee.

The investigators at the participating institutions have worked closely on the implementation of the project. The project was launched with an intensive, hybrid meeting, conducted face-to-face at a central location in Accra and Kumasi and by video conferencing with US investigators. The meeting focused on training on all aspects of the study procedures and building collaborative team relationships and patterns of open communication. This was followed by biweekly video conferencing calls with the US and Ghanaian investigators and on-site project coordinators, and intermittent video conferencing and email communication carried out on an ‘as needed’ basis with the team and counsellors. In addition, one of the US investigators meets with the Ghanaian investigators in Ghana at least quarterly, and follow-up face-to-face meetings are held annually with the full team. In addition, one of the US investigators has held several training sessions with experienced counsellors who will be conducting on-site training with counsellors at each of the participating sites.

Patient and public involvement

Patients, healthcare providers and members of the public played an integral role in planning this study to ensure alignment with their priorities and experiences. Engagement began in the early conceptualisation phase, where input from representatives helped refine the study’s relevance to the target population, shape research questions and inform the methodological approach. The research team maintains contact with representatives to evaluate study implementation (eg, acceptability, applicability) and co-develop dissemination plans to ensure findings are shared widely in accessible and meaningful formats.

Recruitment

Recruitment began in June 2021. Intervention delivery began at three sites in December 2021 and at an additional three sites in June 2022, December 2022 and June 2023. 592 caregiver–child dyads have been enrolled (from June 2021 to September 2023). 17% (n=103) of the caregiver/child dyads who have been screened (n=695) have not been enrolled to the study. Reasons for ineligibility or refusal to participate in the study included the following: the child is already aware of his/her HIV status (n=61); the child is medically excluded (n=11); the child has a developmental disorder (n=1); the caregiver is <18 years of age (n=1); and the caregiver is not interested in participating in the study and/or able to comply with protocol requirements (n=29). Enrolment and characteristics of the enrolled dyads have been comparable across clinical clusters (table 2).

Table 2. Caregiver–child dyad characteristics at enrolment (through September 2023, N=592).

	Cluster 1 clinics*	Cluster 2 clinics†	Cluster 3 clinics‡	Cluster 4 clinics§
	(n=151)	(n=165)	(n=117)	(n=159)
Caregiver’s age, years
Mean (SD)	42.8 (10.8)	44.5 (10.0)	42.0 (10.8)	40.9 (9.9)
Caregiver’s sex, n(%)
Female	129 (85.4)	135 (83.3)	101 (86.3)	136 (87.2)
Caregiver’s HIV status, n(%)
Positive	108 (72.0)	95 (58.6)	72 (62.0)	110 (69.2)
Caregiver’s employment, n(%)
Unemployed	26 (17.2)	25 (15.1)	17 (14.5)	24 (15.1)
Self-employed	15 (9.9)	24 (14.6)	22 (18.8)	12 (7.6)
Private or government sector	110 (72.9)	116 (70.3)	78 (66.7)	123 (77.4)
Child’s age
Mean (SD)	10.6 (2.7)	11.0 (2.5)	10.1 (2.7)	10.2 (2.5)
≤11 years old, N(%)	97 (64.2)	95 (57.6)	82 (70.1)	111 (69.8)
Child’s sex, n(%)
Female	75 (50.0)	72 (44.7)	49 (43.0)	75 (47.8)
Child attending school, n(%)
Yes	147 (99.3)	158 (99.4)	115 (99.1)	150 (96.8)
HIV transmission, n(%)
Mother-to-child	141 (99.3)	143 (96.6)	103 (94.5)	139 (93.3)

^*Cluster 1 Clinics. (1) Kumasi South, (2) Accra Lekma, (3) Accra Princess Marie.

^†Cluster 2 Clinics. (4) Kumasi Tafo Government, (5) Accra Korle-Bu, (6) Kumasi Komfo Anokye.

^‡Cluster 3 Clinics. (7) Kumasi Suntreso Government, (8) Accra Ridge Region, (9) Accra Ga South Municipal.

^§Cluster 4 Clinics. (10) Kimasi Maternal and Child, (11) Accra Tema General/Ashaiman Polyclinic, (12) Kumasi Manhyia Government.

The mean age of the caregivers was 42.5, 65.8% were HIV-positive and 85.2% were female. The children were 46.4% female, with a mean age of 10.5 years. No adverse events have been reported to date, even among the participants who have elected to disclose while enrolled in the study.

Unexpected challenges

The chief challenge faced by the team to date is the conduct of the study in the midst of the COVID-19 travel restrictions. Critical to the success of the project to date have been the two dedicated on-site project coordinators (one full-time in Accra and one in Kumasi), who were also the coordinators for Sankofa 1. Assuring strong Internet connectivity has also been key, given the project’s reliance on regular virtual discussions and data entry. Development of a close working relationship among team members and regular, open communication in Sankofa 1 has facilitated prompt action when needed and monitoring of project data in ‘real-time’ in Sankofa 2.

Discussion

Although compelling evidence supports the advantages of informing children of their HIV status, disclosure recommendations have generally not been implemented. Little emphasis has been placed on equipping a child’s caregiver with information and skills to promote disclosure. At the completion of the project, we will provide information on a structured disclosure intervention that can be integrated into the usual care in Ghana and other low- or middle-income countries (LMICs) to improve the welfare of HIV-infected children and their caregivers (physical and mental health). We will also learn about individual-level (child and caregiver) and clinic-level factors, including the costs associated with the implementation and maintenance of the disclosure intervention in real-world clinic settings in Ghana. These findings will further the scientific understanding of organisational-level facilitators and barriers to effective and sustainable delivery of the intervention and inform future adaptation and roll-out of the intervention in other LMICs. We will also provide empirical insights into the benefits of early disclosure (≤11 years) of paediatric HIV status. This trial is being conducted by a strong team of multidisciplinary investigators in Ghana and the USA, who have collaborated successfully on prior research that informs the development of the proposed study.^{710 15},^{19 29 35} The project is conducted in partnership with the MOH and GHS and with the advice of a Community Advisory Board who is representative of the community.

Strengths and limitations

The Sankofa 2 study extends a previously successful intervention to evaluate its effectiveness in a wider, real-world context across multiple clinics. This approach aims to provide crucial insights into the intervention’s scalability and sustainability. The implementation of the intervention by specially trained staff poses challenges in achieving standardisation and maintaining fidelity across various clinical settings. The study team opted for a stepped wedge design over a parallel cluster design due to the high costs and the limited empirical evidence supporting the benefits of paediatric HIV disclosure. Additionally, the risk of intervention and outcome contamination is notably higher in a parallel design. However, the stepped wedge design may encounter confounding from underlying temporal trends. Moreover, complete blinding is not feasible in a stepped wedge design, potentially introducing assessment bias despite efforts to blind outcome assessors and investigators.

If the intervention is efficacious, feasible and cost-effective, it could be scaled up by the MOH as the standard of care in Ghana. The ultimate beneficiaries of this trial will be HIV-infected children and their caregivers by informing changes in routine care.

Ethics and dissemination

Ethical clearance was obtained from the GHS Ethics Review Committee, the UG Ethical and Protocol Review Committee, the Committee on Human Research Publication and Ethics of the KNUST, the Johns Hopkins Medicine Institutional Review Board and the Yale School of Medicine Human Investigation Committee (IRB Protocol ID 2000028349). After agreeing to participate in the study, the parent/caregiver provides written informed consent, and the children give assent before enrolment. If either the child or caregiver says no to the study, the dyad is not eligible for the study.

The data that support the findings of this study are available from the principal investigators (EP, NRR), on reasonable request. Data will be freely distributed, as available, to qualified academic

investigators for non-commercial research. Following publication of the main paper(s) for this project, we will make the data publicly available in the form of an electronic database. Data will be deidentified and will not contain any direct identifiers or indirect identifiers.