Impact statement
In many low-resource settings across Africa, families affected by psychosis first turn to traditional healers and religious leaders for help, often long before reaching a clinic. Yet, mental health policies and programs are typically designed around biomedical services alone, overlooking the providers that community members use. This lack of communication and referral pathways between informal and formal providers means that efforts to expand access to psychosis care may fail to reach people with severe mental illness (SMI). This study asked caregivers, health workers, traditional healers, religious leaders, local leaders and people with lived experience of SMI and their caregivers in rural Eastern Uganda how community-based psychosis care could realistically work in their setting. The findings reveal that while participants broadly supported collaboration between traditional and biomedical providers, they did not envision equal partnership. Biomedical providers accept referrals from traditional healers but will not refer back, creating an imbalance that could undermine trust and long-term cooperation if left unaddressed. These insights have practical value for anyone designing or implementing mental health programs in pluralistic health systems. These insights suggest that, within the workshop process used here, collaboration was most readily specified at the level of concrete, discrete functions – such as referral, risk recognition and counseling support – rather than as a blanket partnership. This functional approach may provide program designers a realistic framework for building collaboration that respects professional and contextual boundaries while still expanding access to care. More broadly, the study demonstrates the importance of engaging the full range of community members and care providers early in program development, not only to improve design but also to anticipate the implementation friction that often derails mental health interventions in low-resource contexts. The findings are relevant to policymakers, program planners and non-governmental organizations working to close the psychosis treatment gap across Africa and similar settings worldwide.
Background
SMIs, like schizophrenia and bipolar disorder, are leading contributors to global disability (GBD 2019 Mental Disorders Collaborators, 2022). The burden of SMI is disproportionately concentrated in low- and middle-income countries (LMICs), where large treatment gaps persist due to limited availability of biomedical mental health services (World Health Organization, 2021). The most recent global mapping indicates little progress: mental health continues to receive only about 2% of government health budgets, with median spending of $0.04 per person in low-income countries and persistent workforce shortages (World Health Organization, 2025). In LMICs, it is estimated that up to 85% of people living with SMI do not receive evidence-based treatment (Rathod et al., Reference Rathod, Pinninti, Irfan, Gorczynski, Rathod, Gega and Naeem2017). In Uganda, mental health services are constrained by professional shortages, limited infrastructure and centralized services, making access challenging for rural populations (Cooper et al., Reference Cooper, Ssebunnya, Kigozi, Lund and Flisher2010; Kigozi et al., Reference Kigozi, Ssebunnya, Kizza, Cooper and Ndyanabangi2010), with those with SMI facing high levels of stigma (Lee et al., Reference Lee, Christ, Mbabazi, Dabagia, Prendergast, Wykoff, Dasari, Safai, Nakaweesi, Aturinde, Galvin, Akena, Ashaba, Waiswa, Rosenheck and Tsai2024a). These constraints are amplified in rural districts, where families may contend with long travel distances, out-of-pocket transport costs and limited availability of psychotropic medicines and psychosocial treatment at local facilities, producing discontinuous care pathways (Lee et al., Reference Lee, Nakaziba, Waimon, Agwang, Menon, Samuel, Dyas, Nkolo, Ingabire, Wykoff, Hobbs, Kazungu, Basiimwa, Rosenheck, Ashaba and Tsai2024b).
In Uganda, individuals with SMI and their families commonly seek care from traditional healers and religious leaders, who are widely available and culturally embedded within communities (Akol et al., Reference Akol, Moland, Babirye and Engebretsen2018; Berhe et al., Reference Berhe, Gesesew and Ward2024). These providers offer explanatory models that resonate with local beliefs and shape care-seeking (Kpobi and Swartz, Reference Kpobi and Swartz2019). Help-seeking is therefore commonly pluralistic: families may move between biomedical facilities, religious leaders and traditional healers, often in sequence and sometimes concurrently, with biomedical care frequently accessed late or inconsistently (Lee et al., Reference Lee, Nakaziba, Waimon, Agwang, Menon, Samuel, Dyas, Nkolo, Ingabire, Wykoff, Hobbs, Kazungu, Basiimwa, Rosenheck, Ashaba and Tsai2024b). Families and communities play a central role throughout this process in treatment adherence and long-term support, suggesting that effective interventions must engage not only providers but also family and community systems (Thornicroft et al., Reference Thornicroft, Ahuja, Barber, Chisholm, Collins, Docrat, Fairall, Lempp, Niaz, Ngo, Patel, Petersen, Prince, Semrau, Unützer, Yueqin and Zhang2019; Fortuna et al., Reference Fortuna, Lebby, Geiger, Johnson, MacDonald, Chefetz, Ferron, George, Rossom, Kalisa, Mestrovic, Nicholson, Pringle, Rotondi, Sippel, Sica, Solesio, Wright, Zisman-Ilani, Gambee, Hill, Brundrett, Cather, Rhee, Daumit, Angel, Manion, Deegan, Butler, Pitts, Brodey, Williams, Parks, Reimann, Wahrenberger, Morgan, Bradford, Bright, Stafford, Bohm, Carney, Haragirimana, Gold, Storm and Walker2023; Lee et al., Reference Lee, Mbabazi, Jevremov, Nakaweesi, Brodey, Wykoff, Sivakumar, Kazungu, Ssekalo, Rosenheck and Tsai2025c). The COSIMPO trial, conducted in Nigeria and Ghana, demonstrated that a shared care model involving traditional and faith healers alongside primary healthcare workers improved psychosis symptoms and reduced disability at six months and was cost-effective (Gureje et al., Reference Gureje, Appiah-Poku, Bello, Kola, Araya, Chisholm, Esan, Harris, Makanjuola, Othieno, Price and Seedat2020). Notably, COSIMPO’s intervention was structured around the healer–provider dyad; it did not include a structured family psychoeducation component, leaving the role of family systems in psychosis recovery largely unaddressed.
Coordinated Specialty Care (CSC) for first-episode psychosis has demonstrated improved quality of life and functioning compared with usual care in some high-income settings (Mueser et al., Reference Mueser, Penn, Addington, Brunette, Gingerich, Glynn, Lynde, Gottlieb, Meyer-Kalos, SR, Cather, Saade, Robinson, Schooler, Rosenheck and Kane2015; Kane et al., Reference Kane, Robinson, Schooler, Mueser, Penn, Rosenheck, Addington, Brunette, Correll, Estroff, Marcy, Robinson, Meyer-Kalos, Gottlieb, Glynn, Lynde, Pipes, Kurian, Miller, Azrin, Goldstein, Severe, Lin, Sint, John and Heinssen2016). CSC models, such as NAVIGATE, rely on multidisciplinary specialty teams rarely available in rural low-resource contexts (Mueser et al., Reference Mueser, Penn, Addington, Brunette, Gingerich, Glynn, Lynde, Gottlieb, Meyer-Kalos, SR, Cather, Saade, Robinson, Schooler, Rosenheck and Kane2015). However, the challenge is less about whether these functions are desirable and more about how they can be delivered when specialty teams are unavailable. Adaptation requires task-sharing with non-specialist cadres, fit with local care pathways, and attention to implementation constraints (Farooq et al., Reference Farooq, Fonseka, Ali, Milner, Hamid, Sheikh, Khan, MM-u-H, Ariyadasa, Khan and Ayub2024).
Building on these realities, we convened participants with lived experience of SMI, their caregivers and those who currently provide, or plausibly influence, care for people with SMI in rural Buyende District, Uganda to co-design a coordinated care approach. We used four evidence-informed domains commonly included in coordinated care models (psychosocial support, medication management, supported education/employment and family involvement) as a structured discussion framework. Our objectives were to identify (1) locally feasible delivery agents and role divisions for each domain within a task-shared model, (2) how the pluralistic local chain of care could be engaged safely and realistically and (3) anticipated barriers and enabling conditions for implementation.
Methods
Study site
This study took place in Buyende District, a rural district in Eastern Uganda with 403,486 people and 83,497 households (Uganda Bureau of Statistics, 2024). The district is largely agrarian. This region has documented pluralistic care-seeking for both physical and mental illnesses (Lee et al., Reference Lee, Adusumilli, Kyakulaga, Muwereza, Kazungu, Blackwell, Saenz and Schubert2019a, Reference Lee, Adusumilli, Kazungu, Anywar, Kyakulaga, Katuura, Parikh and Willcox2019b, Reference Lee, Nakaziba, Waimon, Agwang, Menon, Samuel, Dyas, Nkolo, Ingabire, Wykoff, Hobbs, Kazungu, Basiimwa, Rosenheck, Ashaba and Tsai2024b).
Study design
We used a qualitative, community-based participatory research design operationalized through three iterative co-design workshops conducted over five weeks in May–June 2025. The purpose was to elicit participant perspectives on what coordinated care for SMI should look like in a rural, resource-constrained, pluralistic setting and to co-develop feasible role divisions and workflows.
Workshop 1 focused on participants’ conceptualizations of SMI, current care practices and perceived gaps in recovery support. Workshop 2 introduced four evidence-informed coordinated-care domains – psychosocial support, medication management, supported education/employment and family involvement (Mueser et al., Reference Mueser, Penn, Addington, Brunette, Gingerich, Glynn, Lynde, Gottlieb, Meyer-Kalos, SR, Cather, Saade, Robinson, Schooler, Rosenheck and Kane2015) – and elicited participant critique, adaptation strategies and proposed delivery agents for each component. Workshop 3 examined implementation logistics through a mock scenario exercise, in which participants reviewed a proposed workflow and provided structured feedback on referral pathways, role assignments and potential bottlenecks. Workshop guides were developed prior to each session and refined iteratively based on content from preceding workshops.
To clarify the scope of participant input: the four coordinated-care domains and the proposed workflow reviewed in Workshop 3 were pre-specified by the research team. Within this fixed structure, participants shaped who would deliver each function, how roles would be divided, what adaptations would be needed and where the boundaries of collaboration should lie. Several proposals emerged directly from participant input, including the “model families” peer mentorship concept and the use of Savings and Credit Cooperative Organizations (SACCOs) – member-owned community financial institutions that pool savings and provide loans – for socioeconomic reintegration. When participant proposals diverged, disagreements were retained as analytically informative rather than resolved into consensus. In summary, the research team fixed what was discussed (the four domains and a draft workflow), while participants shaped how each function could feasibly be delivered in their setting – for example, by assigning counseling and follow-up roles to Village Health Teams (VHTs) and families, and by setting explicit limits on referral to traditional healers.
Participants and sampling
We recruited 18 participants using purposive sampling through community networks established by the non-governmental organization Empower Through Health (ETH) and community health workers, known as VHTs (Mays et al., Reference Mays, O’Neil, Mworozi, Lough, Tabb, Whitlock, Mutimba and Talib2017). Participants were distributed across two parallel, faith community-anchored workshop series: one based in a church and one in a mosque. Each group (n=9 each) included: a general facility-based biomedical clinician, a facility-based biomedical mental health clinician, a village chairperson, a traditional healer, a religious leader, a VHT member, a person with lived experience (PLE) of SMI and a parent or other caregiver of a person living with SMI.
All participants were aged 18 years or older and resided in Buyende District or the surrounding area. PLEs were eligible if they had a confirmed diagnosis of schizophrenia, bipolar disorder or a related SMI within the past five years. Among the 18 participants, 13 were men and 5 were women (church group: 7 men, 2 women; mosque group: 6 men, 3 women), reflecting leadership role distributions in the local setting.
Attendance across workshops was high. In the church group, the traditional healer was absent for the afternoon session of Workshop 1, and the mental healthcare worker was absent from Workshop 3. In the mosque group, the mental healthcare worker from Workshop 1 was unable to continue and was replaced by a different mental healthcare worker for subsequent workshops; the biomedical worker from Workshop 1 was also unable to attend Workshops 2 and 3 and was not replaced.
Workshop procedures
All workshops were conducted in Lusoga by trained Ugandan research assistants at the master’s or bachelor’s level, with training backgrounds in public health, social sciences and medicine, who were fluent in both English and Lusoga. American research assistants at the bachelor’s level, with training backgrounds in global health and related social sciences, attended as observers and supported technical functions, including audio recording, but did not facilitate discussions. Sessions ran approximately from 9:00 am to 5:00 pm, with a one-hour lunch break from 1:00 to 2:00 pm. All sessions were audio-recorded with participant consent.
Positionality
This study was conducted by ETH, an organization that provides mental healthcare in rural Uganda, as a part of the Global Health Experiential Fellowship (Lee et al., Reference Lee, Kazungu, Ssekalo, Blackwell, Nakaziba, Monnig, Mbabazi, Muwereza, Waiswa, Ingabire, Vega, Brown, Rohrbaugh, Rosenheck, Waiswa and Kohrt2025b). Thus, social desirability bias may have influenced participants to endorse biomedical care pathways. Recruitment through ETH-affiliated VHT networks may also have preferentially included individuals with greater prior exposure to or alignment with biomedical services. Participants may additionally have perceived the research team and ETH’s clinical providers as a unified entity, further shaping responses. To mitigate this risk, the research team clarified its independence from ETH’s clinical operations during informed consent, emphasized that participation would not affect access to services and reiterated this distinction at the start of each workshop session. Finally, post-colonial stigmatization of traditional healing practices may have influenced how participants described or evaluated traditional healers, a dynamic documented in similar Ugandan research contexts (Lee et al., Reference Lee, Coleman, Nakaziba, Terfloth, Coley, Epparla, Corbitt, Kazungu, Basiimwa, Lafferty, Cole, Agwang, Kathawala, Nkolo, Wogali, Richard, Rosenheck and Tsai2025a).
Data management and analysis
Audio recordings were transcribed and translated directly into English by teams of two to three research assistants, each team including at least one assistant fluent in Lusoga. To ensure translation quality, a subset of transcripts was spot-checked against the original audio by bilingual team members to verify accuracy.
We analyzed transcripts using the Framework Method (Gale et al., Reference Gale, Heath, Cameron, Rashid and Redwood2013). Analysis proceeded through five stages: familiarization, thematic framework development, indexing, charting and interpretation. The initial thematic framework was structured deductively around the four coordinated-care domains while allowing inductive themes to emerge from the data. During analysis, we merged psychosocial support and medication management into a single theme, as participants consistently described them through the same distributed delivery: task-shifting across a tiered workforce of prescribers, VHTs and families. One additional cross-cutting theme, the pluralistic care landscape, emerged inductively and was retained because it shaped adaptation decisions across all components.
After a collective coding exercise to establish consensus, teams of two to three research assistants independently coded transcripts and resolved discrepancies through discussion. To enhance credibility, we triangulated findings across the two-workshop series (church and mosque) to identify cross-cutting themes and contextual variations. We also deliberately attended to points of disagreement or contested collaboration in the data (e.g., asymmetries in willingness to refer across provider types), treating these as analytically informative rather than resolving them into artificial consensus. Coding and thematic analyses were conducted using Google Docs. As a final validation step, the coding framework and thematic outputs were reviewed against the charted data by supervisor to confirm that themes were grounded in the source transcripts.
Results
A total of 18 participants took part across 2 parallel workshop series (church-based and mosque-based). Quotes are labeled by speaker role, workshop session (e.g., mosque Workshop 3 = 3M) and gender. Four themes emerged, broadly consistent across both series. Table 1 summarizes participant-proposed adaptations.
Summary of participant-proposed coordinated care adaptations
Table 1. Long description
The table consists of five columns: Theme, Coordinated care domain, Proposed adaptation, Proposed delivery agents, and Key constraints.
* Theme 1: Chain of care (cross-cutting). Domain: All components. Adaptations: Multiple entry points, transport/access infrastructure, and collaboration protocols. Agents: All sectors. Constraints: Pluralistic help-seeking, transport costs, medication stockouts, and referral asymmetry.
* Theme 2: Psychosocial support and medication management. Domain: I R T plus Medication management. Adaptations: Task-shifting to tiered workforce, protecting relational time, and one-way traditional healer referral. Agents: Prescribers, V H Ts (Village Health Teams), caregivers, and religious/community leaders. Constraints: Specialist absence, spiritual cure drift, supervision needs, and economic competition.
* Theme 3: Supported education/employment. Domain: Supported employment and education. Adaptations: Clearance workflows, vocational support in S A C C O s, and family-centered reintegration. Agents: Prescribers, local leaders, families, and S A C C O s. Constraints: School fees/transport and lack of employment specialists.
* Theme 4: Family psychoeducation. Domain: Family education program. Adaptations: Foundational front door component, training cascade, and model family peer mentorship. Agents: Biomedical staff, V H Ts, and model families. Constraints: Message drift, curricula standardization, and supervision requirements.
Theme 1: The “chain of care” is pluralistic, sequential and shaped by barriers and contested collaboration
Subtheme 1.1: Help-seeking is multi-sectoral
Participants described a typical pattern of moving through multiple systems – home remedies, religious support, biomedical care, traditional healing – often as responses to perceived failure within one system.
When a person’s behavior changes is when the family decides to get him to the hospital. They try many things, mostly people try local herbs, some go to church, if they all fail, that’s when they decide to take him to hospital. – PLE, 3M, woman
A participant similarly described community default pathways, emphasizing that biomedical care is often not sought:
If you ask 10 people where they take a person with SMI they will say 7 out of 10 will say traditional healers. 2 of them will say the church and 1 will say the medical center. – Family member, 1C, man
A psychiatric biomedical clinician estimated that:
80% of our patients start from the church, traditional healers, Iman before coming to us – Psychiatric biomedical clinician, 1M, man
This pattern was consistent across both workshop series: help-seeking was described as a patterned sequence shaped by perceived accessibility, legitimacy and prior experiences with care, in which biomedical services were one of several possible entry points rather than the default first contact.
Subtheme 1.2: Structural barriers constrain biomedical continuity
Participants repeatedly emphasized transport costs, poverty and unavailable medicines as reasons for interrupted care. A PLE described transport as a recurring and unsustainable expense:
Because at times we find it hard since we do not have enough money for transport which is a monthly thing and yet this is a lifetime treatment – PLE, 1M, woman
A family member provided a concrete example of the costs associated with seeking services, particularly when symptoms require special transport arrangements:
There was a time when I had to move from Jinja to Kampala and it needed me to hire a special car… so hiring that special car was very expensive. – Family member, 1C, man
Participants also described health system constraints as part of the barriers, including lack of service visibility and missing protocols at lower level facilities:
It will be hard for you to see mental health services on that list… The guidelines that have those treatment protocols are not available. – Psychiatric biomedical clinician, 1C, man
In these accounts, treatment interruption was framed as a predictable consequence of structural constraints rather than individual non-compliance, suggesting that adherence support and transport solutions are core feasibility conditions for any adapted model.
Subtheme 1.3: Collaboration is desired but contested
Participants expressed interest in cross-sector collaboration yet also voiced clear boundaries, especially around biomedical referral to traditional healers. Collaboration with religious institutions was often framed as feasible and already practiced through outreach and reciprocal relationships:
We always have trainings where we got to churches and mosques and train religious leaders…in most cases, we treat patients who usually go to religious leaders and when they get better we usually tell them to go back to the pastor. – Psychiatric biomedical worker, 1C, man
At the same time, participants articulated explicit limits on referral to traditional healers, including professional and evidentiary objections:
So I cannot refer to a traditional healer even though I fail in treating the patient because their treatment is not proven – Psychiatric biomedical worker, 1M, man
Religious leaders also described their own non-referral boundaries:
Like for me I cannot refer my patients to a traditional healer… – Religious leader, 1C, man
Economic incentives were described as complicating collaboration across sectors:
This collaboration can happen although it may not be so easy… Everyone would want to find a way to earn from this. – Family member, 2C, man
A PLE also highlighted that collaboration is strained not only between healing sectors but even within religious communities:
…discrimination among different religious groups, where pastors cannot work with imams. There is also discrimination between traditional healers and medical workers. – PLE, 3M, woman
Participants described collaboration in non-uniform ways, ranging from active cooperation with religious leaders to categorical limits regarding traditional healers, often conditioned by legitimacy concerns and economic incentives. Across accounts, collaboration was discussed as function-specific rather than as a generalized, bidirectional partnership.
Theme 2: Psychosocial support and medication management require distributed, relational delivery across non-specialist cadres
This theme merges Individual Resiliency Training (IRT) and medication management components. Participants consistently described both functions through a common delivery logic: task-shifting across a tiered workforce (facility clinicians, VHTs, family/caregivers and community leaders) with relational trust and continuity as mechanisms supporting engagement and follow-up.
Subtheme 2.1: Task-shifting across a tiered workforce of clinicians, VHTs and families
Participants acknowledged the absence of specialist therapists and proposed training non-specialists:
Medical prescribers are given various training to offer very many therapies like behavior therapy … training these medical workers would highly be beneficial to our community – Psychiatric biomedical clinician, 2M, man
A PLE emphasized missed opportunities within existing systems:
I request that organizations should also go to lower levels to support the VHTs in communities. For example, where I work at (a local health center), there are no organizations that come there to educate people about mental disorders – PLE, 2M, woman
A traditional healer endorsed a VHT-centered task-shifted model for community counseling and linkage:
If VHTs are given the right info, they can give very good counseling services in the community – Traditional healer, 2M, man
Families were similarly described as essential for daily implementation and follow-up. A family member emphasized immediate involvement:
I emphasize that the caretaker should be involved right from the start, because these are the people that follow-up on the patients – Family member, 3C, man
VHTs were positioned as the bridge for community education and linkage, conditional on receiving standardized information from biomedical workers:
Then it is the VHTs who educate the other community members…VHTs would be good but when they are getting information from the biomedical workers and counselors in the communities – General biomedical clinician, 3M, woman
Across participant groups, descriptions converged on expanding care through training and redistributing responsibilities across locally available cadres: facility clinicians for assessment/clinical decisions, VHTs for community counseling, education and linkage, and families for daily implementation, adherence support and follow-up.
Subtheme 2.2: Relational trust as the mechanism of engagement
Participants emphasized relational trust – friendliness, patience and sustained interaction – as central to both psychosocial support and medication adherence. A local leader described provider–patient rapport as directly shaping whether advice is followed:
I think it is important that the patient and nurse befriend each other, because if the nurse is not friendly the patient will not follow whatever he has been instructed to do. If the nurse is friendly the patient will put into practice what they’ve been told. – Local leader, 2C, woman
Multiple participants echoed this view, and religious leaders described the same relational logic for community-based counseling, framing friendship with the patient as what made counsel possible:
When I make this patient my friend, it makes it easy for me to talk to them or counsel them about their illness – Religious leader, 2C, man
Across these accounts, relational trust was described as shaping whether guidance was accepted and sustained.
Subtheme 2.3: Traditional healers as contested collaborators
Traditional healers described referral to biomedical services as appropriate when cases exceed their capacity:
If I cannot manage I refer the patient to empower (ETH, which provides mental healthcare), so there has been collaboration – Traditional healer, 2M, man
However, biomedical providers explicitly refused reciprocal referrals (Subtheme 1.3). Family members described why traditional healers remain relationally credible within the chain of care:
Remember the traditional healers also relates to their patients. They give them herbs and explain everything to them. – Family member, 2C, man
Participants described traditional healers as trusted and relationally accessible for families while also describing them as contested from biomedical perspectives. In these accounts, referral to biomedical care and information-sharing were more commonly described than reciprocal referral in the opposite direction.
Theme 3: Supported education/employment must be reimagined around documentation and community economic institutions
Subtheme 3.1: Letters of support for returning to school and work
Some participants described return to school or work as requiring more than symptom improvement. In their accounts, recovery needed to be made legible to schools, employers and community institutions through formal documentation or local endorsement. Biomedical workers were positioned as especially important in certifying clinical readiness:
The medical prescriber writes a letter that allows them to go back to school – Psychiatric biomedical clinician, 2M, man
Others extended this documentation role to employment, describing health workers as responsible for providing written confirmation that a person had recovered sufficiently to resume work:
In the medical provider’s letter, he should clearly indicate that this patient is physically and mentally fit for all activities at work – General biomedical clinician, 2M, man
Participants also identified local leaders as potential brokers of reintegration, particularly for employment or access to livelihood opportunities:
The other person who can also lead is the chairman LC1 where he can write a letter recommending someone to get a job. – Traditional healer, 3M, man
These accounts suggest that reintegration was understood not only as a clinical or functional process but also as an administrative process. Participants described formal letters from prescribers and local leaders as part of the practical process of returning to work and accessing opportunities.
Subtheme 3.2: SACCOs, savings groups and families as reintegration infrastructure
Participants proposed leveraging existing community financial structures (SACCOs and savings groups) to support livelihood participation and structured re-engagement. A PLE described SACCOs as a pathway to meaningful, locally relevant tasks:
In my own perspective, these SACCOs really could help in the best way. The patient could be given work in a SACCO. Like if he is already well. They can give him a certain task like rearing. Now rearing means keeping, and it can be either keeping pigs, keeping goats, keeping chickens, and so forth. This will help the patient to be engaged. – PLE, 3M, woman
A VHT framed savings groups as mechanisms for ongoing support:
We can form saving groups. The profits from the money he saved can be used to help SMI patients – VHT, 2C, man
PLE also emphasized the centrality of family support for sustaining reintegration:
Family should play a very big role and also help them financially because this is a long process – PLE, 2M, woman
Participants described SACCOs/savings groups and household support as practical mechanisms for livelihood engagement after recovery. These accounts framed reintegration as supported by community and family structures that already exist locally.
Theme 4: Family psychoeducation as foundation, delivered through cascade and peer mentorship
Subtheme 4.1: Families as early detectors, crisis responders and care managers
Participants described family psychoeducation as enabling effective engagement with services and practical care decisions. A PLE emphasized that families need clarity on steps and expectations between patient and clinician:
The family needs to know the steps that are involved between the patient and the nurse – PLE, 3C, man
He also emphasized the family’s responsibility when medicines are unavailable at the
health center:
If the medication is not available and there is a need for buying it, the family needs to take action especially if this patient is young – PLE, 3C, man
A PLE proposed home-based education as a delivery approach, identifying feasible cadres for that role:
We should have people who come to the home and educate them on SMI…VHTs and religious leaders who can educate and help these families. – PLE, 2M, woman
Participants repeatedly positioned families as responsible for navigating care steps, access barriers and practical problem-solving. In these accounts, psychoeducation was framed as enabling families to carry out these roles.
Subtheme 4.2: “Model families” and a task-shifted cascade as the delivery mechanism
Participants proposed peer mentorship, especially “model parents,” as a credible teaching mechanism. A psychiatric medical worker described the concept:
We should have model parents…one who has had a (PLE), and managed to effectively help…a model in the community whereby people can learn from him. – Psychiatric biomedical clinician, 1C, man
Participants emphasized that “model parents” should be trained and aligned under a consistent educator to reduce message drift:
I think VHTs and the model parents should be trained by the nurse – PLE, 3C, man
Participants described a tiered cascade in which knowledge diffuses from clinical staff to VHTs to families. When a facilitator summarized this structure:
So are you trying to propose that the nurse should train the VHT then the VHT trains the model parents and also takes on the responsibility of connecting the families that need their services?
The group responded with unanimous affirmation – Group response, 3C Participants affirmed this proposed cascade structure during the workshop discussion, framing psychoeducation as requiring both expertise at the source (clinical staff) and reach in the community (VHTs).
Discussion
This community-based participatory co-design study examined how participants in rural Buyende District conceptualize coordinated care for psychosis in a pluralistic, resource-constrained setting. We used four evidence-informed coordinated-care domains (psychosocial support, medication management, supported education/employment and family involvement) as a structured framework to elicit locally grounded delivery logic, role divisions and anticipated implementation constraints. Participants described why clinic-centered, specialist team delivery is difficult to operationalize in this context and instead proposed distributing core functions across locally available actors. These findings suggest that adaptation in this setting may be best approached by preserving coordinated care functions while redesigning delivery around local pathways, feasible cadres and clear boundaries for contested collaboration. Positioned against prior literature, most of what participants described confirms existing evidence on task-sharing, collaborative care or peer-delivered support in low-resource settings. The contribution of this study lies in contextual specification – which actors should hold which functions, where the boundaries of collaboration lie in this setting and which existing community structures they viewed as feasible for adaptation – including caregiver-to-caregiver “model family” mentorship and SACCO-based reintegration.
Task-sharing
Participants repeatedly emphasized that non-specialist cadres and community-embedded actors (especially VHTs and families) are plausible anchors for follow-up, education and linkage. This aligns with global mental health evidence that trained non-specialist health workers can contribute meaningfully to mental health care delivery in low-resource settings, particularly when supervision and clear roles are present (van Ginneken et al., Reference van Ginneken, Tharyan, Lewin, Rao, Meera, Pian, Chandrashekar and Patel2013; World Health Organization, 2016). These proposals should also be read against Uganda’s tiered health system, in which VHTs constitute the formal community level (Health Center I), with defined health promotion, referral and follow-up functions (Mays et al., Reference Mays, O’Neil, Mworozi, Lough, Tabb, Whitlock, Mutimba and Talib2017). Outreach from higher level facilities to this community tier has historically been difficult to sustain, and national assessments document chronic underfunding, irregular training and supervision and weak referral linkages for VHT programs (Musoke et al., Reference Musoke, Ndejjo, Atusingwize, Ssemugabo, Ottosson, Gibson and Waiswa2020). Read in this light, the delivery structures participants proposed are largely not new: they would formalize, resource and supervise existing-but-undersupported structures rather than create parallel cadres. Importantly, participants framed distribution of responsibilities not only as a response to specialist scarcity but also to maintain contact and support in everyday environments where transport barriers, competing livelihood demands and intermittent service availability can disrupt continuity. Participants perceived relational continuity and local proximity as key feasibility conditions for sustained engagement. This generates clear hypotheses for piloting: whether VHT-supported follow-up and family-centered routines improve retention in care, adherence and functional outcomes relative to clinic-only strategies.
Pluralistic pathways and collaboration
Consistent with the literature, participants described a pluralistic pathway to care in which families often seek help from religious leaders and traditional healers, sometimes before accessing biomedical services (Lee et al., Reference Lee, Nakaziba, Waimon, Agwang, Menon, Samuel, Dyas, Nkolo, Ingabire, Wykoff, Hobbs, Kazungu, Basiimwa, Rosenheck, Ashaba and Tsai2024b). Their proposals did not converge on a single approach to healer–biomedical integration but instead articulated contested collaboration with clear boundaries and directional asymmetry. Some described traditional healers as referring complex cases to clinics, while biomedical providers expressed reluctance to refer in the reverse direction due to concerns about evidentiary standards and spiritual etiologies. This asymmetry echoes the “co-wives” metaphor reported among Ugandan traditional healers, who described willingness to collaborate with the biomedical system while perceiving the relationship as unequal (Akol et al., Reference Akol, Moland, Babirye and Engebretsen2018). These findings complement the COSIMPO trial, which demonstrated that manualized, structured collaboration between traditional/faith healers and primary care workers can improve psychosis outcomes and be cost-effective in Nigeria and Ghana (Gureje et al., Reference Gureje, Appiah-Poku, Bello, Kola, Araya, Chisholm, Esan, Harris, Makanjuola, Othieno, Price and Seedat2020). However, COSIMPO’s collaborative model was highly structured and context-specific; our findings highlight that in rural Eastern Uganda, collaboration may be experienced as uneven and contested, and a model that relies on traditional healer referrals into the biomedical system while offering nothing in return may risk undermining the legitimacy of the very providers families rely on most, proving unstable if healers perceive the arrangement as extractive. Our data suggest that biomedical providers accepted referral and linkage from traditional healers to clinics, endorsed VHT-mediated information-sharing and were open to training community actors in recognition of risk signs and counseling support, but rejected reverse referral and clinical co-management within the arrangements discussed in these workshops. Within this workshop process, participants therefore predominantly articulated collaboration at the level of discrete functions rather than uniform bidirectional integration. This should not be interpreted as evidence that more fully shared-care or co-treatment approaches, such as COSIMPO, would be infeasible or less effective in rural Eastern Uganda. Such arrangements were not directly elicited or compared in this study.
Family psychoeducation
CSC models include a structured family education component, reflecting the longstanding evidence that family interventions can improve outcomes for schizophrenia-spectrum disorders (e.g., relapse reduction) in many settings (Mueser et al., Reference Mueser, Penn, Addington, Brunette, Gingerich, Glynn, Lynde, Gottlieb, Meyer-Kalos, SR, Cather, Saade, Robinson, Schooler, Rosenheck and Kane2015). Participants in this study similarly positioned families as central – not only emotionally, but operationally – given their roles in day-to-day monitoring, adherence support and practical problem-solving under structural constraints. A context-specific adaptation proposed in the workshops was the use of “model parents/model families” as peer mentors – experienced caregivers who can provide credible, context-specific guidance to other families navigating SMI care. The proposal is not wholly novel: it joins an existing literature on family peer support, caregiver peer mentoring and peer-led family support for caregivers and families of people with mental health conditions (Acri et al., Reference Acri, Hooley, Richardson and Moaba2017), as well as broader lay-delivered mental health approaches, such as the Friendship Bench, in which trained lay “grandmothers” deliver problem-solving therapy for common mental disorders (Chibanda et al., Reference Chibanda, Weiss, Verhey, Simms, Munjoma, Rusakaniko, Chingono, Munetsi, Bere, Manda, Abas and Araya2016). The contribution is contextual: participants in rural Uganda independently identified caregiver-to-caregiver mentorship as a feasible mechanism for extending family psychoeducation, grounding it in experiential knowledge of SMI caregiving and embedding it within a VHT-supported cascade. We therefore frame “model families” as a locally grounded implementation strategy for piloting. “Model parents” may additionally function as a stigma-reduction mechanism through routine contact with families who have navigated recovery, consistent with community-based approaches in which contact and shared public narratives have functioned as stigma-reduction mechanisms (Lee et al., Reference Lee, Rauben, Liu, Kim, van der Velde, Taylor, Walsh, Asasira, Katongole, Hatfield-King, Blackwell, Iheanacho, Christ and Ssekalo2022, Reference Lee, Christ, Blackwell, Kazungu, Krause, Monnig, Iheanacho, Ssekalo and Rosenheck2023). Future pilots can evaluate whether model-family mentorship improves caregiver knowledge, burden, engagement with services and patient functioning. Participants also proposed a psychoeducation cascade (e.g., clinician->VHT->model families), which can be interpreted as a practical diffusion mechanism to extend family education beyond the clinic. This should be treated as a testable implementation strategy. At the same time, the centrality assigned to families should be treated cautiously. Caregiving for SMI carries substantial economic and psychological burden in Africa (Addo et al., Reference Addo, Agyemang, Tozan and Nonvignon2018), and urbanization, labor migration and shifts toward nuclear household structures may erode the extended-family capacity these proposals presume – a tension also visible in Ugandan research on help-seeking for child and adolescent mental illness (Akol et al., Reference Akol, Moland, Babirye and Engebretsen2018). Households without an able caregiver risk being left behind unless VHT- and peer-based supports can substitute for core family functions.
Reintegration
One of the more context-specific findings concerns reintegration. Participants described reintegration not only as clinical recovery or symptom control but also as a process that involves institutional permission-making – including letters from prescribers/health workers and local leadership (e.g., LC1) to facilitate return to work or acceptance back into community roles. This is where it is crucial not to overclaim: the data support that several participants perceived documentation as an enabling or determining mechanism, not that all schools/employers uniformly require it. Participants also proposed reintegration supports through community economic structures (e.g., SACCOs/savings groups) and household-level resource pooling. Whereas CSC models such as NAVIGATE deliver supported education/employment through specialized team roles (Mueser et al., Reference Mueser, Penn, Addington, Brunette, Gingerich, Glynn, Lynde, Gottlieb, Meyer-Kalos, SR, Cather, Saade, Robinson, Schooler, Rosenheck and Kane2015), participants in this setting emphasized locally available economic institutions and administrative workflows as feasible pathways to restore role functioning. As a contribution, this highlights that CSC adaptation may require broadening SEE beyond individual coaching to include practical linkage mechanisms that fit rural economic life and local gatekeeping realities.
Limitations
The sample comprised 18 participants, with only one to two individuals per role in each workshop. This breadth-over-depth design was intentional, prioritizing interaction across the chain of care over saturation within any single perspective, but it limits transferability and means that some perspectives are represented by very few voices. Larger, deeper and more geographically varied sampling might surface heterogeneity that these workshops could not. Accordingly, we present the findings as transferable hypotheses rather than saturated descriptions.
In addition to the limitations inherent to qualitative co-design work in a single rural district, the study context and workshop format may have influenced the perspectives expressed. In particular, because the research team pre-specified the coordinated-care domains, the workshop structure may have privileged function-specific arrangements and constrained consideration of alternative partnership models, including more fully shared co-treatment approaches such as COSIMPO. Workshops were convened in partnership with ETH, which provides biomedical mental healthcare, and this affiliation may have introduced social desirability effects – for example, encouraging favorable framing of biomedical approaches or reducing willingness to critique clinic practices. Workshops were also held at religious centers (church- and mosque-affiliated sites), a pragmatic choice that reflected local help-seeking contexts but may have shaped group dynamics by elevating religious authority and influencing how spiritual explanations and religious roles were discussed. Also, because the sample was male-dominated given the predominance of men in leadership roles, it likely underrepresented women’s gendered experiences. Finally, mixed group discussions included critique of traditional healers, which may have constrained healers’ willingness to speak openly or encouraged defensive positioning, potentially amplifying biomedical or religious critiques relative to healer perspectives.
Several further limitations concern data and team composition. Audio recordings were translated and transcribed directly into English without an intermediate verbatim Lusoga transcription; although a subset of transcripts was spot-checked against the original audio by bilingual team members, this single-step process risks losing idiom and nuance. We also did not systematically document existing outreach and informal coordination structures, such as the health-worker WhatsApp groups used for supervision and communication across levels of the health system in parts of the region (Henry et al., Reference Henry, Winters, Lakati, Oliver, Geniets, Mbae and Wanjiru2016), against which the feasibility of participant proposals could have been benchmarked. Future work could complement these workshops with individual interviews and role-homogeneous groups (e.g., healer-only, caregiver-only and PLE-only sessions) to surface perspectives that may be less accessible in mixed-group settings.
Conclusion
Participants in rural Buyende District proposed a coordinated care approach that preserves core functions (psychosocial support, medication continuity, reintegration support and family psychoeducation) while redesigning delivery around a pluralistic chain of care and a tiered, task-shared workforce. The co-design outputs highlight perceived feasibility of VHT- and family-anchored continuity, function-specific cross-sector collaboration boundaries, peer caregiver mentorship (“model families”) and reintegration supports incorporating administrative endorsement and local economic structures. Future work should pilot these proposals, evaluate feasibility and safety and test whether the distributed delivery model improves engagement and functional recovery relative to usual care.
Open peer review
To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2026.10319.
Data availability statement
Data available upon reasonable request to the corresponding author.
Acknowledgments
This research was supported by the National Institute on Drug Abuse, National Institutes of Health, U.S. Department of Health and Human Services, grant number K24DA061696. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Drug Abuse, the National Institutes of Health or the U.S. Department of Health and Human Services. This manuscript is subject to the NIH Public Access Policy and is available in PubMed Central upon publication.
Author contribution
Conceptualization: Y.J.L., A.C.T. and R.A.R.; Data Curation: Y.J.L., J.K., J.H., F.L., S.G.F., K.H., M.M.M., J.L.L., A.A.H. and R.K.; Formal Analysis: Y.J.L., J.K., J.H., F.L., S.G.F., K.H., M.M.M., J.L.L., A.A.H. and R.K.; Funding Acquisition: N/A; Investigation: J.K., J.H., F.L., S.G.F., K.H., M.M.M., J.L.L. and A.A.H.; Methodology: Y.J.L., A.C.T. and R.A.R.; Project Administration: Y.J.L., J.K., J.H., F.L. and R.K.; Resources: Y.J.L. and R.K.; Software: N/A; Supervision: Y.J.L., J.H., F.L., R.K., R.R. and A.C.T.; Validation: Y.J.L., R.R. and A.C.T.; Visualization: Y.J.L.; Writing – Original Draft: Y.J.L., J.K., J.H., F.L., S.G.F., K.H., M.M.M., J.L.L., A.A.H. and R.K.; Writing – Review and Editing: Y.J.L., J.K., J.H., F.L., S.G.F., K.H., M.M.M., J.L.L., A.A.H., R.K., A.C.T. and R.A.R.
Funding statement
Dr. Tsai reports receiving funding from U.S. National Institutes of Health (K24DA061696-01).
Competing interests
A.C.T reports receiving financial honoraria from Elsevier (for his work as Co-Editor in Chief of SSM – Mental Health) and BMJ Publishing Group Ltd (for his work as Clinical Editorial Advisor for The BMJ).
Ethics and approvals
All research procedures were approved by the institutional review boards of Yale University in the United States (2000034605) and The AIDS Support Organization in Uganda (TASO-2023-222). Consistent with national guidelines, we obtained clearance to conduct the study from the Uganda National Council for Science and Technology (SS1860ES). This study was conducted in accordance with the principles of the Declaration of Helsinki. The consent process included detailed explanation of privacy rights and study procedures, with forms provided in either English or Lusoga depending on patient preference. All participants provided written consent. Reimbursement for time and transport (20,000 UGX per session per participant) was provided per local norms.
Open access
