Optimising HIV and related SRH service delivery for adolescents and young adults: a protocol for a discrete choice experiment

Abstract

Introduction

Adolescents and young adults (AYAs) represent a significant proportion of the global population, yet they face persistent barriers to accessing HIV services. In Sub-Saharan Africa, nearly 90% of AYA HIV cases occur within the region, with young women disproportionately affected. In Chad, while the overall HIV prevalence has declined, the risk remains high among AYA, accounting for 26.3% of all new infections. Stigma, lack of confidentiality, negative provider attitudes and structural barriers continue to hinder service utilisation, underscoring the urgent need for evidence-based, AYA-centred interventions. However, little is known about which attributes of HIV services AYA prioritise when accessing HIV care. This study applies a discrete choice experiment in Chad to systematically quantify AYA preferences for HIV services to inform the design of youth-responsive interventions.

Methods and data analysis

The study employs a D-efficient fractional factorial design, developed through extensive qualitative research and pilot testing. The final design comprises 80 choice scenarios, divided into 10 blocks to reduce participant burden. The study will recruit 1000 AYA living with HIV aged 15–24 years across eight provinces, ensuring geographical and epidemiological representativeness. Participants will be randomly assigned to one of the blocks and complete eight choice tasks. Choice data will be analysed using conditional logit, mixed logit and latent class models to estimate trade-offs and preference heterogeneity.

Ethics and dissemination

Ethical approval was obtained from the National Committee on Bioethics of Chad (#010/MESRS/SE/SG/2024). In addition to disseminating findings through scientific publications, policy briefs and stakeholder engagements, the study will incorporate a codesign process with AYA, healthcare providers and policymakers to translate research findings into actionable interventions that align with AYA preferences and improve HIV service delivery.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Attributes and levels for the discrete choice experiment were derived directly from prior qualitative research to ensure contextual relevance.

• A D-efficient fractional factorial design was used to maximise statistical efficiency while minimising cognitive burden for participants.

• Pilot testing with adolescents and young adults confirmed the clarity and feasibility of the choice tasks.

• The purposive maximum variation sampling approach supports heterogeneity but does not provide statistical representativeness.

• As with all discrete choice experiments, stated preferences may not fully align with the actual healthcare behaviours in real-world settings.

Introduction

The United Nations defines adolescents and young adults (AYAs) as youths aged 15–24 years.¹ AYAs represent one-sixth of the global population, with nearly 90% residing in low and middle-income countries (LMICs).²,⁵ This demographic transition presents both opportunities and challenges, as AYA navigate a period of rapid physical, cognitive and social development, which shapes their long-term health trajectories.^{6 7} Despite growing recognition of the importance of adolescent health, health systems worldwide remain inadequately equipped to meet their needs, particularly in the domain of HIV and related sexual and reproductive health (SRH) services.⁸,¹⁰ This gap in healthcare provision is particularly concerning given that HIV/AIDS remains a leading cause of death among adolescents globally, with an estimated 1.75 million AYA living with HIV.^{11 12} Sub-Saharan Africa (SSA) bears the greatest burden of adolescent HIV infections, accounting for nearly 90% of global adolescent HIV cases.¹¹,¹³ Within this region, young women (15–24 years of age) are disproportionately affected, representing 63% of all new HIV infections among youth.^{13 14} Structural and social factors, including gender inequality, economic dependence, limited access to education and restrictive sociocultural norms, further exacerbate the vulnerability of young women and girls.¹⁴ While significant progress has been made in expanding access to antiretroviral therapy (ART) and prevention programmes, challenges such as stigma, lack of confidentiality, negative provider attitudes and financial constraints continue to limit service utilisation and retention in care.¹⁵,¹⁸ To address these challenges, global health organisations, including WHO, UNICEF, UNAIDS and the Global Fund, have emphasised the need for adolescent-friendly health services (AFHS) that are accessible, acceptable, equitable, appropriate and effective.¹⁹,²³ WHO’s Global Standards for Adolescent-Friendly Health Services advocate for confidential, stigma-free and youth-centred care that prioritises the needs of adolescents.²⁴ However, despite these global commitments, many LMICs, including Chad, struggle to operationalise these recommendations, leaving AYA with limited options for HIV and related SRH care.²⁵

Chad, a low-income and conflict-affected country, faces significant health system constraints that hinder AYA access to essential healthcare services, particularly for HIV and related SRH care.²⁵ Although the HIV prevalence in Chad declined from 1.6% in 2010 to 1.0% in 2022, AYA remain at high risk, accounting for 26.3% of all new infections.^{26 27} Adolescent girls and young women are particularly vulnerable, experiencing HIV infection rates four times higher than young men, largely due to early marriage, gender-based violence and economic vulnerability.²⁵,²⁸ Recent qualitative research in Chad, involving focus group discussions with 52 AYA and 48 healthcare providers and community actors, highlights critical gaps in service delivery.^{15 16} Findings reveal that stigma, lack of confidentiality, negative provider attitudes and rigid service structures deter AYA from seeking care.^{15 16} AYA frequently express concerns about being seen at health facilities that provide HIV services, fearing social repercussions.^{15 16} Additionally, financial barriers, long waiting times and lack of psychosocial support make it difficult for AYA to engage in sustained HIV care.^{15 16} While healthcare providers acknowledge these challenges, they report limited resources, inadequate training and systemic constraints in providing adolescent-friendly care.¹⁵

Recognising the need for tailored, youth-friendly HIV services, global health organisations have emphasised the importance of adolescent-centred care models characterised by confidentiality, affordability, accessibility and stigma reduction.²⁴ Evidence from Chad and other SSA countries underscores the urgent need to redesign AYA health services to be more responsive, accessible and youth friendly.^{15 16} To reflect this emphasis, desirable AYA-centred care should be (1) designed with youth input, ensuring that AYA preferences shape service delivery; (2) accessible and stigma-free, integrating confidentiality measures, flexible hours and non-judgemental providers; and (3) integrated within existing health systems, incorporating peer-led models, mobile outreach and digital health innovations to increase uptake and retention.^{21 29 30} While these features are desirable, resource constraints necessitate prioritisation because health systems must often make difficult trade-offs among service attributes to maximise impact with limited budgets. Understanding which attributes AYA value most is crucial to ensuring efficient allocation of scarce resources and to designing feasible interventions that can be scaled effectively. If the discrete choice experiment (DCE) reveals that AYA deprioritise globally recommended features, this does not necessarily invalidate those components, but it does suggest the need to either adapt services more closely to AYA preferences or address gaps in understanding and engagement. Our qualitative study in Chad revealed that AYA face unique challenges, including stigma, financial constraints and limited access to AYA-friendly services, while also benefitting from supportive, peer-driven care.^{15 16} To translate these insights into effective service redesign, we now need to complement our qualitative insights with quantitative data that identifies which service attributes AYA prioritise and clarifies the trade-offs they make when selecting care.

A DCE is a stated preference method that allows researchers to quantify trade-offs individuals make when choosing between different service attributes.³¹,³⁵ Unlike traditional surveys that measure general preferences, DCEs present realistic scenarios in which participants must select between different healthcare options, enabling researchers to estimate the relevance and importance of each service characteristic.³¹,³⁵ DCEs have been widely used in health economics and policy research, particularly in LMIC settings, to inform evidence-based decision-making.³⁶,³⁸ They have been successfully applied to assess youth preferences for HIV services, contraceptive methods and healthcare delivery models.³⁵³⁹,⁴² However, no such study has been conducted in Chad. This study addresses that gap by implementing a DCE to systematically capture and quantify AYA preferences for HIV and related SRH services. The findings will guide policymakers, implementers and healthcare providers in designing AYA-responsive models of care that are not only grounded in evidence but also reflective of the lived realities and priorities of AYA in Chad.

Aims

The aim of this study is to identify and quantify the factors that influence AYA decisions when selecting HIV and related SRH services in Chad using DCE methods. By understanding these preferences, we can inform the development of AYA-friendly service models that improve accessibility, acceptability and utilisation. This study is restricted to AYA living with HIV, and attributes reflect treatment-related HIV and SRH services relevant to this group. Specifically, we will assess the following:

1. Preferences for key HIV and related SRH service attributes, including confidentiality, provider attitude, service location and psychosocial support.

2. The impact of service accessibility factors, such as waiting time, service hours and privacy, on adolescent healthcare choices.

3. Willingness to pay (WTP) for preferred service options and the extent to which cost serves as a barrier to healthcare access.

4. The relative importance of features commonly associated with youth-centred care models, such as non-judgemental communication, peer involvement and opportunities for AYA input into service design. Rather than comparing AYA-centred models to alternative models, the DCE will isolate and evaluate the specific attributes that define AYA-centred care to determine the extent to which AYA prioritise these features when faced with realistic trade-offs.

5. The trade-offs AYA are willing to make when selecting HIV and related SRH services and how specific barriers such as stigma and provider attitudes influence their healthcare-seeking behaviour.

Methods and analysis

Overview of the DCE

A DCE is a stated preference method used to elicit individual preferences by asking respondents to choose between hypothetical alternatives characterised by different attributes.³¹,³⁸ In the context of AYA’s access to HIV and related SRH services, the DCE provides a structured approach to understanding the trade-offs AYA make when selecting healthcare services.³¹,⁴² This method is particularly useful in low-resource settings where revealed preference data is scarce, enabling policymakers to design interventions that align with the needs and preferences of the target population.^{43 44}

The DCE follows a systematic process, beginning with qualitative research to identify key service attributes relevant to AYA.^{45 46} These attributes are then refined through stakeholder engagement and literature review to ensure their relevance and comprehensiveness.⁴⁷ Next, the attributes are assigned varying levels to create hypothetical choice sets.⁴⁵,⁴⁷ Respondents are then presented with these sets and asked to select their preferred option.⁴⁵,⁴⁷ The choices made by participants are then analysed using logistic regression models, such as conditional logit, mixed logit (MCL) or latent class models (LC), to quantify the weight individuals assign to different attributes and assess heterogeneity in preferences.^{48 49}

Economic theory underlying the DCE

The DCE is grounded in random utility theory (RUT), which posits that individuals make choices to maximise their utility, or perceived benefit.^{37 50} Utility is modelled as a function of the attributes of a given alternative, along with an unobservable component that captures factors not explicitly measured.^{37 50 51} Formally, the utility $\left(U\right)$ that an individual derives from choosing alternative $j$ in a choice set can be expressed as:

$$U_{ij}=V_{ij}+{\epsilon }_{ij},j=\mathrm{1,2},\dots ,J$$

where $U_{ij}$ is the total utility an individual $i$ derives from alternative $j$; $V_{ij}$ represents the systematic component of utility, derived from observable attributes and their associated coefficients; and ${\epsilon }_{ij}$ is the stochastic component, accounting for unobserved influences and individual variability.

When presented with a set of alternatives, individuals are assumed to select the option that provides the highest utility.⁵¹ The probability of choosing a specific alternative is modelled using discrete choice models, typically beginning with the multinomial logit model (MNL) and extending to more flexible models such as the mixed logit to account for preference heterogeneity.⁵¹ DCEs also align with Lancaster’s consumer theory, which suggests that individuals derive utility not from goods or services per se but from their attributes.^{51 52} This framework is particularly relevant in healthcare settings, where the design of services influences patient choices and engagement.^{51 52} From a policy perspective, the DCE framework enables the estimation of willingness to pay (WTP) for specific attributes, providing insights into the economic value that adolescents place on various service characteristics.⁵¹ These findings can inform the optimisation of healthcare delivery models, ensuring that interventions are not only clinically effective but also aligned with user preferences, thereby improving uptake and adherence. By leveraging economic theory and empirical data, the DCE approach provides a robust methodological foundation for designing AYA-friendly HIV and related SRH services, ultimately contributing to more effective health system responses tailored to young people’s needs.

DCE development process

The development of the DCE followed a structured, multi-step approach to ensure that the attributes and levels accurately reflected the healthcare preferences and decision-making processes of AYA in Chad.⁴⁵,⁴⁷ This process was informed by extensive qualitative research and grounded theory analysis, leading to the refinement of choice sets through piloting before final participant recruitment.^{15 16} Figure 1 illustrates the key stages involved in developing the DCE, from initial qualitative research to the recruitment phase.

Figure 1. The DCE development process. AYA, adolescents and young adults; DCE, discrete choice experiment; SRH, sexual and reproductive health; STI, sexually transmitted infection.

Qualitative research to inform the DCE

This DCE is rooted in extensive qualitative research that explored the experiences, barriers and facilitators shaping AYA’s engagement with HIV and related SRH services in Chad.^{15 16} Understanding these factors is crucial for designing youth-friendly services that align with AYA priorities. The qualitative phase informed the DCE attributes and levels, ensuring that the choice experiment reflects a real-world decision-making rather than externally imposed assumptions.^{53 54} The process began with four focus group discussions (FGDs) with AYA (aged 15–24 years) conducted in October 2022 to assess contextual factors influencing access to HIV and related SRH services in Chad. This phase also explored AYA’s experiences with confidentiality, provider attitudes and stigma in healthcare settings. The findings from this stage highlighted key barriers and facilitators that shape AYA health-seeking behaviour.¹⁶

Next, HIV-positive and HIV-negative Chadian AYA (n=52) participated in an adaptive sense-making process, using participatory methods, to develop a deeper understanding of how AYA access, interpret and make decisions regarding HIV and related SRH care and how these processes influence their health-seeking behaviour.¹⁵ The analysis described the processes through which AYA navigate HIV and related HIV care amid stigma, healthcare system gaps and financial constraints.¹⁵ Using the findings, a conceptual model was generated describing how AYA interpret and negotiate their healthcare experiences to make informed choices about seeking, initiating and continuing care.⁵⁵ By anchoring the DCE in a robust, theory-driven framework, the study will leverage a larger sample to assess the generalisability of the sense-making findings, quantify the relative importance of various service attributes and provide actionable insights to support the codesign of AYA-friendly HIV and related SRH services in Chad.

Determining the attributes and levels for the DCE

The qualitative insights and the developed theory were instrumental in shaping the attributes and levels selected for the DCE.^{15 16 55} Rather than relying solely on predefined attributes used in prior research, the DCE attributes were derived directly from the concerns and priorities expressed by AYA during FGDs and interviews.⁵³,⁵⁵ This ensured that the choice tasks presented in the DCE were realistic, contextually relevant and representative of the actual trade-offs AYA face when making healthcare decisions.^{53 54} The qualitative research phase highlighted key factors that influence AYA engagement with HIV and related SRH services. These factors include confidentiality, provider attitude, communication style, cost, accessibility and psychosocial support, which were systematically translated into measurable DCE attributes and levels.^{15 16 55} Importantly, the services explored in the DCE encompass the full continuum of HIV care that AYA interact with, including HIV testing and counselling, linkage to and retention in ART services, STI and SRH care, health education and psychosocial support. While some selected attributes such as cost, privacy and service hours apply broadly across healthcare settings, they were explicitly framed to reflect AYA’s preferences within the HIV context. The key attributes and levels used in the DCE reflect the most critical factors influencing AYA health-seeking behaviour, as identified in the qualitative phase (table 1).

Table 1. DCE attributes, levels and their theoretical connection.

Attribute	Levels	Alignment with theory generated from qualitative study and sense-making participatory work
Confidentiality	Full (no information shared without consent)	AYA fear stigma and social judgement, making confidentiality crucial for healthcare engagement.
	Partial (parents and guardians may be informed)
	None
Cost	Free	Financial constraints limit access to care, requiring an understanding of willingness to pay (WTP).
	Subsidised
	Full cost
Provider interpersonal behaviour	Excellent youth-friendliness (warm, respectful, listens actively, explains clearly)	Negative provider attitudes deter AYA from seeking care, reinforcing the importance of non-judgemental environments.
	Moderate youth-friendliness (professional but impersonal, limited communication)
	Poor youth-friendliness (judgemental tone, dismissive, avoids explaining or listening)
Convenience of service hours	Open every day, including evenings and weekends	Rigid service hours conflict with AYA schedules (school, work), necessitating flexible hours.
	Open on weekdays only 9am–5pm
	Open 2 days per week with flexible hours*
Quality of care	High	Perceived quality of care influences trust and retention, shaping service-seeking behaviours.
	Moderate
	Low
Privacy	Private consultation room	Privacy ensures a safe space for AYA to discuss sensitive health issues without fear of exposure.
	Partially private (curtains or partitions)
	No privacy (shared space)
Communication	Non-judgemental and supportive providers	Supportive communication fosters trust, while judgemental behaviour discourages service use.
	Neutral provider attitude
	Disrespectful or dismissive providers
Educational and preventive services	Comprehensive education and prevention (ART adherence, SRH and prevention of reinfection)	Health education empowers AYA with knowledge, impacting prevention and treatment uptake
	Basic education and prevention (limited ART adherence and SRH information)
	No educational or preventive services
Psychosocial support	Comprehensive (adherence counselling and peer support groups)	Psychosocial support (mental health services and peer networks) reduces distress and improves adherence.
	Basic (some counselling available)
	None
Waiting time	Less than 30 min	Long waiting times discourage AYA from accessing and continuing care, requiring efficiency improvements.
	About 1 hour
	More than 2 hours
Service location	Community-based	Service proximity affects accessibility, with community-based services reducing logistical barriers.
	Clinic-based
	Hospital

Notes: ‘Flexible hours’ are defined for participants as: services available outside school/work hours, including evenings and weekends. ‘Provider interpersonal behaviour’ refers to how providers interact with AYA, whether they are welcoming, respectful and non-judgemental in their treatment and communication.

ART, antiretroviral therapy; AYA, adolescents and young adults; SRH, sexual and reproductive health.

Each of these attributes is directly linked to the developed theory of adaptive sensemaking, which describes how AYA continuously interpret, negotiate and respond to complex, uncertain or emotionally charged situations, particularly when navigating fragmented or stigmatising systems.⁵⁵ Originally developed within organisational and cognitive psychology, sensemaking theory has been increasingly applied in health systems research to understand how individuals make decisions in the face of uncertainty, resource constraints or competing priorities.^{56 57} Its core tenets emphasise that individuals do not passively receive information or services; rather, they actively interpret their experiences, draw on social and cultural cues and adapt their behaviour based on feedback from their environment.^{56 57} In the context of AYA HIV care, adaptive sensemaking offers a powerful lens to understand how AYA process stigma, financial hardship, provider interactions and peer influence when deciding whether, when and how to engage in care. Our qualitative findings showed that AYA use iterative, adaptive reasoning to assess the safety, trustworthiness and feasibility of health services, often weighing competing concerns like confidentiality versus accessibility, or provider attitude versus service quality.⁵⁵ These insights informed the translation of theoretical concepts into practical, measurable attributes for the DCE, ensuring contextual validity and theoretical alignment.

Peer-led models, mobile outreach services and digital health innovations are increasingly recognised as key strategies for improving AYA engagement in HIV care. While elements of these delivery approaches are partially reflected in attributes such as ‘service location’ and ‘provider attitude’, they were not included as separate attributes in this first iteration of the DCE to minimise cognitive burden. Nonetheless, these models are highly salient for AYA and will be further explored during the codesign phase following the DCE, where preferences identified through this study will inform the development of AYA-centred service delivery prototypes.

Pilot study

The DCE design

Having selected the final attributes, levels and number of alternatives (two per choice set), an experimental design for the survey was generated.⁵⁸ Given the number of attributes and levels, a full factorial design including all possible combinations was not feasible. Instead, a D-efficient experimental design was developed using the Ngene software (version 1.4) to maximise statistical efficiency by minimising parameter standard errors.³²⁵⁸,⁶³ To improve statistical efficiency, prior information about the attribute-level coefficients would typically be used—either from previous studies or expert opinions.^{63 64} However, for the pilot study, the prior coefficients were set to zero, ensuring an unbiased estimate of attribute effects.

Pilot-testing attributes and levels

A pilot study was conducted with 20 AYA living with HIV to assess the clarity, relevance and feasibility of the DCE design. The primary objectives of the pilot were to refine the list of attributes and levels, evaluate participants’ ability to differentiate between levels and ensure that the survey format (online supplemental file 1) was user-friendly and cognitively manageable. Given the complexity of DCEs, the pilot study was essential in determining whether participants could effectively engage with the choice tasks and make meaningful trade-offs between service attributes. The pilot test led to several refinements in the DCE attributes and levels, based on participant responses and patterns observed during testing. Table 2 illustrates the final list of attributes and levels. Additional information on the refinement process is in online supplemental file 2.

Table 2. Final attributes and levels.

Attributes	Levels
Confidentiality	Full (no information shared without consent) Partial (parents/guardians may be informed)
Cost	Free Subsidised
Provider interpersonal behaviour	Excellent youth-friendliness Moderate youth-friendliness
Service hours	Every day (including evenings/weekends) 2 days per week with flexible hours
Quality of care	High Moderate
Privacy	Private consultation room Partially private (curtains/partitions)
Communication	Non-judgemental and supportive Neutral
Educational services	Comprehensive education and prevention (ART adherence, SRH and prevention of reinfection) Basic education and prevention (limited ART adherence and SRH information)
Psychosocial support	Comprehensive (adherence counselling and peer support groups) Basic (some counselling available)
Waiting time	Less than 30 min 1 hour
Service location	Community-based Clinic-based

ART, antiretroviral therapy; SRH, sexual and reproductive health.

Beyond attribute refinements, the pilot study also assessed the ease of completing the choice experiment. Results indicated that 70% of participants found the choice tasks easy or very easy to complete, 25% found them somewhat difficult, while 5% reported significant difficulty. Most participants understood trade-offs and were able to make confident selections, indicating that the final design was cognitively accessible for the target population.

Proposed study

Final experimental design and blocking strategy

The study employs a D-efficient fractional factorial design to maximise statistical efficiency while minimising participant burden.⁵⁸,⁶³ Given the number of attributes and levels, a full factorial design that includes all possible attribute-level combinations was not feasible due to the excessive number of potential choice sets. Instead, a fractional factorial design was developed using Ngene software, ensuring that each attribute-level combination is adequately represented while reducing the total number of choice tasks per participant.⁵⁸,⁶³ This approach allows for robust estimation of AYA preferences for key healthcare attributes without overwhelming participants with an excessive number of decisions.⁵⁸,⁶³

The final experimental design was informed by coefficient estimates derived from a pilot study conducted among 20 AYA living with HIV. Insights from this pilot study helped refine attribute-level combinations, ensuring that the D-efficient design accurately captures trade-offs relevant to AYA healthcare preferences. Based on these findings, the final design consists of 80 choice scenarios, which were divided into 10 blocks to further reduce cognitive load. Each participant will be randomly assigned to one block, completing eight choice tasks. This block approach ensures that while individual participants engage in a manageable number of scenarios, all 80 choice scenarios are distributed across the full sample, preserving statistical power and enabling precise estimation of service preferences. To further reduce decision fatigue, the 16 unique choice sets were split into two blocks, with each participant receiving one of the two blocks. This random assignment balances the representation of all attribute levels across the sample, minimising bias. The final design reflects the trade-offs identified in the pilot study, ensuring realistic and distinguishable choices. This structure enhances decision realism, facilitating AYA engagement with the DCE while reducing response bias. table 3 provides an example of a choice scenario presented to participants, illustrating the structured decision tasks within the study. The complete list of choice modules and all 80 DCE scenarios, including the 16 unique choice sets assigned across 10 blocks, is provided in online supplemental file 3.

Table 3. Example of scenario.

Attribute	Option A	Option B
Confidentiality	Full (no information shared without your consent)	Partial (some information may be shared with parents or guardians)
Cost	Free	Subsidised
Provider attitude	Neutral	Youth-friendly
Service hours	2 days per week with flexible hours	Every day (including evenings and weekends)
Quality of care	High	Moderate
Privacy	Private consultation room	Partially private
Communication	Partially private	Neutral
Educational services	Comprehensive education and prevention (ART adherence, SRH and prevention of reinfection)	Basic education and prevention (limited SRH adherence and SRH information)
Psychosocial support	Basic (some counselling available)	Comprehensive (adherence counselling and peer support groups)
Waiting time	Less than 30 min	1 hour
Service location	Community-based	Clinic-based
Which do you prefer?	()A	()B

ART, antiretroviral therapy; SRH, sexual and reproductive health.

Instructions

Below are two different service options for HIV and related SRH care. Each option includes a combination of features, such as how much it costs, how friendly the provider is, how private the consultation is and how long you might wait. Please review all the features in both Option A and Option B carefully. There are no right or wrong answers. We want to know which service you would prefer to use if both were available to you. Please compare all the features listed in the table. Then, choose the option that you would be most likely to use, based on your personal preferences. If anything is unclear, feel free to ask the facilitator for help before choosing.

Sampling size calculation

The sample size required for the main effects in a DCE depends on the number of choice sets per participant (J), the number of alternatives per choice set (T) and the number of levels in the most complex attribute (A).^{58 65} Applying these values, the minimum number of participants is the following:

$${\displaystyle {\displaystyle N\ge \frac{500\ast J}{T\ast A}=\frac{500\ast 8}{2\ast 2}=1000}}$$

Participants and survey procedures

The DCE will be conducted across multiple provinces in Chad using a purposive maximum variation sampling strategy to capture a diverse cross-section of AYA living with HIV.^{46 66} The study will exclusively enrol 1000 AYA living with HIV aged 15–24 years, recruited from both urban and rural areas with varying HIV prevalence levels and healthcare accessibility. Accordingly, the attributes and levels selected for the DCE reflect treatment-related HIV and SRH service components relevant to this group. Preventive-only services for HIV-negative or high-risk populations were not included in this protocol. Province selection was guided by a two-step approach, balancing both the number of AYA living with HIV and HIV prevalence. First, provinces with a sufficient number of HIV-positive AYA were identified to ensure recruitment feasibility. Then, among these, areas with high HIV prevalence were prioritised to capture regions where the epidemic poses a substantial public health concern. Additional considerations included urban-rural balance, healthcare infrastructure and logistical feasibility.

This approach does not aim for statistical representativeness but is instead grounded in qualitative and implementation research paradigms that emphasise capturing variation in experience over random sampling.^{46 66} The selected provinces, including Lac, Logone Occidental, Logone Oriental, Mandoul, Mayo Kebbi Est, Moyen Chari, N’Djamena and Tandjilé, will reflect a range of contexts, from densely populated urban centres to underserved rural areas and from high-burden regions to those with emerging service gaps.²⁵,²⁷ This heterogeneity is essential for exploring how contextual differences shape AYA preferences and trade-offs they are willing to make when accessing HIV and related SRH services. Table 4 presents HIV prevalence by province, supporting the rationale for selection. Online supplemental file 4 provides additional details on sampling locations and recruitment procedures, while online supplemental file 5 details inclusion and exclusion criteria for participant eligibility.

Table 4. HIV prevalence by province.

Province	Number of AYA living with HIV	HIV prevalence (%)	Urban/rural
N’Djamena	1884	4.0	Urban
Lac	1004	2.1	Rural
Logone Occidental	674	2.7	Urban
Logone Oriental	587	0.1	Rural
Moyen Chari	471	2.9	Urban/rural
Mandoul	470	0.6	Rural
Mayo Kebbi Est	441	0.7	Rural
Tandjilé	286	2.6	Rural

In addition to completing the DCE tasks, participants will provide demographic and contextual information to enable analysis of preference heterogeneity (onlinesupplemental files 1 3). All survey materials will be administered in French and Arabic to ensure accessibility for Chad’s linguistically diverse youth population. Participants will receive standardised guidance on how to complete the DCE, and comprehension checks will be implemented to confirm understanding of the choice format. Although the final response rate will be reported after data collection, we anticipate a high cooperation rate due to the use of community-based recruitment strategies and engagement with AYA networks to build trust and encourage participation.

Data analysis

Descriptive analysis

The study will begin with a descriptive analysis of participant characteristics to summarise demographic and contextual variables collected, including age, gender, educational level, HIV status, previous SRH service use and perceived stigma. As part of the demographic survey, the information on previous SRH service use will assess whether previous exposure to these services shapes AYA preferences in the DCE. Since HIV services are often integrated within broader SRH platforms, previous encounters with SRH services may influence AYA’s expectations around confidentiality, provider attitudes and service accessibility, all of which are relevant to their HIV care decisions. In addition, the survey will collect information on participants’ household economic status. This will include questions on self-reported income, household assets and perceived financial well-being to assess socioeconomic status. Categorical variables will be presented as frequencies and percentages, while continuous variables (if applicable) will be summarised using means and standard deviations (SDs) or medians and interquartile ranges (IQRs), depending on the distribution. To ensure methodological rigour and comprehensive reporting, the study protocol adheres to the Discrete Choice Experiment Reporting Checklist (DIRECT), which provides structured guidance for reporting all essential components of DCE studies in health research (online supplemental file 6).⁶⁷

Choice data analysis and model estimation

The primary analysis will focus on estimating AYA preferences for HIV and related SRH service attributes using discrete choice models. The choice data will be analysed using conditional logit, mixed logit and latent class models to assess trade-offs, preference heterogeneity and willingness to pay (WTP) estimates where applicable. To explore variations in preferences, interaction effects will be incorporated into the models to assess whether demographic or contextual factors influence choices. The results will highlight how different subgroups prioritise service attributes differently, providing actionable insights for targeted intervention strategies. If a cost attribute is included, WTP estimates will be derived to assess the monetary value participants assign to different service attributes. This will help determine how much AYA are willing to trade off between different features, which is essential for pricing and subsidy decisions in youth-friendly service models. Model performance will be assessed using standard model selection criteria, including log-likelihood values and information criteria (AIC/BIC) to compare goodness-of-fit, likelihood-ratio tests to determine the best model specification, cross-validation techniques to assess predictive accuracy and tests for assumption violations to ensure that the estimated preferences are valid and reliable. All analyses will be conducted using STATA and Ngene for experimental design.^{59 63 68}

Patient and public involvement

AYA living with HIV have been involved in the design and development of the DCE in three ways: (1) qualitative interviews and focus group discussions exploring barriers and enablers to access HIV and related SRH services; (2) sensemaking participatory study to describe how they navigate complex choices to access HIV and related SRH services; and (3) piloting of DCE to assess understanding of choices and feasibility. Phases 1 and 2 informed the content of the DCE, and phase 3 permitted refinement of the DCE.

Ethics and dissemination

Ethical considerations

This study will adhere to ethical guidelines for research involving human participants and has received approval from the National Committee on Bioethics of Chad (#010/MESRS/SE/SG/2024). Given the sensitivity of HIV and related SRH services, special measures will be taken to protect patient confidentiality and ensure voluntary participation. All participants will receive detailed information about the study including its objective, procedures, potential risks and benefits. Written informed consent will be obtained from all participants aged 18 and above. For AYAs aged 15–17, parental or guardian consent will be obtained, along with the AYA’s assent to participate. In cases where parental consent may pose a barrier (eg, risk of stigma or disclosure), alternative procedures following Chad’s ethical research guidelines on adolescent health will be explored.

All survey responses will be anonymised and stored securely. No personally identifiable information will be collected. Data will be stored on password-protected servers, and access will be restricted to authorised research team members. Participants will be reminded that their responses will be kept confidential and used solely for research purposes. Given that discussions on SRH and HIV may cause psychosocial distress, participants will be provided with referral information for counselling and support services. Additionally, trained research staff will be available to address concerns and provide appropriate referrals if needed.

Dissemination plan

The dissemination phase of this study is essential to ensure that the findings from the DCE inform AYA-centred policy and service design in Chad. Insights generated through the DCE will provide data-driven guidance on how to optimise HIV and related SRH service delivery for AYA. Dissemination efforts will target multiple audiences, including policymakers, healthcare providers, researchers and civil society organisations.

Findings will be disseminated through peer-reviewed publications to contribute to the global evidence base on AYA-friendly HIV and related SRH service delivery. The study results will also be presented at international and regional conferences to facilitate cross-country learning. In Chad, policy briefs will be developed and shared with national stakeholders, including the Ministry of Public Health, the National AIDS Control Programme and youth networks. Stakeholder engagement workshops will be held to communicate findings in accessible formats and stimulate dialogue about implications for national programming and budgeting.

Future research and implementation planning

In parallel to dissemination activities, the DCE findings will serve as the empirical foundation for the next phase of research—a participatory codesign process aimed at developing feasible and contextually appropriate interventions. This process will include engagement with AYA, healthcare workers and policymakers and will be structured using the Delphi method to build consensus on priority service design features.⁶⁹ To support future implementation, the Consolidated Framework for Implementation Research (CFIR) will be used to identify barriers and facilitators to integration of AYA-responsive models within Chad’s health system.⁷⁰,⁷² These next steps are planned as part of a separate implementation research phase that builds directly on the results of the current study.

Data availability statement

A de-identified dataset generated from the DCE will be made publicly available on reasonable request following the conclusion of the study and publication of primary results. The dataset will be stripped of all personal identifiers and stored in a secure institutional repository. Access will be granted to researchers who provide a methodologically sound proposal and agree to comply with data use agreements consistent with participant consent and ethical approval obtained from the National Committee on Bioethics of Chad.

Discussion

To the best of our knowledge, this study is the first of its kind to apply a DCE in Chad to understand the healthcare preferences of AYA living with HIV. The DCE is a rigorous, theory-driven and empirical method grounded in economic and decision sciences, designed to quantify the relative importance individuals place on specific service attributes.^{31 32} Unlike traditional surveys, DCEs rely on experimentally designed choice sets to stimulate real-world decision-making and allow estimation of trade-offs individuals are willing to make.^{31 32} This study leverages those strengths to systematically elicit AYA preferences for HIV and related SRH services under realistic constraints.

While DCEs do not in themselves produce clinical or health outcome data, they serve as essential evidence-building tools for designing more acceptable, person-centred interventions.^{31 32} This novel application is particularly essential in Chad, where AYA engagement in HIV service design remains limited, and interventions often fail to reflect AYA-specific needs. When implemented effectively, DCE-informed services have the potential to enhance uptake, satisfaction and retention, outcomes that are linked, in turn, to improved health trajectories.^{72 73} Therefore, although the DCE is not ‘evidence-based’ in the clinical trial sense, it generates actionable, context-specific preference data that supports evidence-informed policymaking. In this way, the study provides empirical foundations for codesigning AYA-friendly health services that are more likely to be used and sustained.^{73 74}

A strength of this study is its methodological rigour, particularly the systematic integration of qualitative research into the DCE design. The study was preceded by extensive qualitative research, ensuring that the attributes and levels used in the choice tasks were derived directly from the experiences, priorities and challenges expressed by AYA themselves, rather than being externally imposed. This grounded approach enhances the ecological validity of the study, making the findings more applicable for real-world interventions. Another key strength is its purposive, maximum variation sampling approach, which ensured representation across provinces with diverse HIV prevalence rates, health system characteristics and urban-rural distribution. By balancing both the number of AYA living with HIV and HIV prevalence, the study will capture diverse healthcare experiences across urban and rural settings. This approach allows for generalisable insights, ensuring that the resulting interventions are applicable to a wide range of AYA across Chad, including those in hard-to-reach areas where access remains a challenge. Additionally, the use of a D-efficient fractional factorial design strengthens the study’s ability to generate robust estimates of preference weights, while the blocking strategy minimises participant efficiency, ensuring that trade-offs in service attributes are accurately captured, allowing policymakers to prioritise high-impact service modifications that will improve HIV service uptake and retention among AYA.

Despite its strengths, the study has some limitations. First, stated preferences in a DCE may not always translate into actual behaviour in real-world healthcare settings. While DCEs provide valuable insights into decision-making trade-offs, they are still based on hypothetical scenarios rather than observed service utilisation patterns. This means that external factors, such as stigma, social norms or healthcare provider attitudes, could still influence healthcare choices in ways that are not fully captured in the experiment. Another limitation is sample size constraints in certain provinces, particularly in provinces with lower numbers of registered AYA living with HIV. While adjustments were made to ensure representativeness, some provinces with lower HIV prevalence or smaller AYA populations may still be under-represented, potentially limiting the generalisability of findings to all AYA living with HIV in Chad. Future research could consider expanding the sample size or conducting follow-up qualitative studies to validate preference trends in under-represented provinces. Lastly, while the study ensures AYA-friendly survey administration by providing materials in both French and Arabic, literacy and comprehension challenges may still affect some participants’ ability to fully engage with the DCE tasks. To mitigate this, trained facilitators and pre-survey comprehension checks will be incorporated, but some residual bias due to misinterpretation of choice tasks may persist.