Impact statement
This study provides insights into the complex relationships between HIV status, depression, mental health status and pluralistic providers in rural Uganda. By analyzing the interaction among HIV status, mental health status and service use among traditional, religious and biomedical providers, this study identifies primary correlates of service use and highlights the challenge of providing HIV treatment and mental health services to those in need. In rural Uganda, mental health care frequently involves multiple provider types due to limited biomedical options and cultural beliefs about mental health. This limited mental health service availability in the region also contributes to competing demands among patients with physical and mental health service needs. These findings suggest that integrating mental healthcare services into the existing religious and traditional systems may be beneficial due to cultural alignment and social trust. For faith healers, who are most frequently sought for mental health care, a complementary role with biomedical providers including integrated referral and communications systems may be appropriate. Collaboration with traditional healers should also be pursued, involving culturally appropriate messaging and referral networks with biomedical providers. Mental health screening and counseling ought to be integrated into existing primary care clinics and village health teams to determine patient needs and facilitate access to appropriate mental health resources. Ultimately, these findings highlight the importance of integrating treatment for mental health care across pluralistic pathways to best address patient needs, particularly among those with both mental health and physical disease burdens. This study provides insights into the primary correlates of mental health service use and commonly used providers in rural Uganda, supporting integrated care models that could improve mental health access and outcomes in rural Uganda and similar settings across Africa.
Introduction
Human immunodeficiency virus (HIV) has been widely associated with comorbid depression (Bernard et al., Reference Bernard, Dabis and De Rekeneire2017; Zhan et al., Reference Zhan, Ouyang, Zhai and Yang2024). Similarly, people living with symptomatic HIV have been associated with higher likelihood of lifetime depression compared to people living with asymptomatic HIV generally (Ayano et al., Reference Ayano, Solomon and Abraha2018; Seid et al., Reference Seid, Abdu, Mitiku and Tamirat2020). Global meta-analyses have estimated this prevalence at around 39% (Tran et al., Reference Tran, Ho, Ho, Latkin, Phan, Ha, Vu, Ying and Zhang2019) in comparison to approximately 31% among asymptomatic HIV (Rezaei et al., Reference Rezaei, Ahmadi, Rahmati, Hosseinifard, Dehnad, Aryankhesal, Shabaninejad, Ghasemyani, Alihosseini, Bragazzi, Raoofi, Kiaee and Ghashghaee2019). Within these populations, depression is a highly treatable mental illness with many evidence-based interventions (Passchier et al., Reference Passchier, Abas, Ebuenyi and Pariante2018; Karrouri et al., Reference Karrouri, Hammani, Benjelloun and Otheman2021). The combination of HIV and major depressive disorder disproportionately affects low- and middle-income countries, especially in Africa (Abas et al., Reference Abas, Ali, Nakimuli-Mpungu and Chibanda2014) with meta-analyses estimating a current depression prevalence of 39% in Eastern Africa (Molapo et al., Reference Molapo, Mokgalaboni and Phoswa2025). Specifically, Uganda has the tenth-highest national prevalence of HIV globally at 4.9% (World Bank, 2025). Similarly, some studies report current depression prevalence in Uganda as high as 30% among people living with HIV (Kinyanda et al., Reference Kinyanda, Woodburn, Tugumisirize, Kagugube, Ndyanabangi and Patel2011; Nakimuli-Mpungu et al., Reference Nakimuli-Mpungu, Bass, Alexandre, Mills, Musisi, Ram, Katabira and Nachega2012; Kaggwa et al., Reference Kaggwa, Najjuka, Bongomin, Mamun and Griffiths2022b). This high prevalence of depression is accompanied by low availability of resources to provide appropriate treatment (Kulisewa et al., Reference Kulisewa, Stockton, Hosseinipour, Gaynes, Mphonda, Udedi and Pence2019).
Structural barriers to mental health care are particularly noticeable in rural Africa, which contains the majority of the continent’s population (Jobarteh, Reference Jobarteh2023). In Uganda, approximately 69% of adults in Uganda reside in these areas (Uganda Bureau of Statistics, 2024; World Bank, 2024). Despite the predominant rurality of Uganda and many other African countries, most prior research on the intersection of HIV and mental illness has been conducted in urban or semi-urban settings (Owe-Larsson et al., Reference Owe-Larsson, Säll, Salamon and Allgulander2009; Breuer et al., Reference Breuer, Myer, Struthers and Joska2011). Furthermore, biomedical mental health services are largely unavailable due to limited trained personnel in rural areas and low levels of mental health care funding (Atewologun et al., Reference Atewologun, Adigun, Okesanya, Hassan, Olabode, Micheal, Ahmed, Ukoaka, Idris, Oso and Lucero-Prisno2025). Stigma toward people with mental illness is also a significant barrier to mental health care in rural areas (Lee et al., Reference Lee, Christ, Mbabazi, Dabagia, Prendergast, Wykoff, Dasari, Safai, Nakaweesi, Aturinde, Galvin, Akena, Ashaba, Waiswa, Rosenheck and Tsai2024b). Less than 1% of Uganda’s already limited health budget is directed toward mental health (Molodynski et al., Reference Molodynski, Cusack and Nixon2017), and the vast majority of psychiatrists reside in urban settings (Kigozi et al., Reference Kigozi, Ssebunnya, Kizza, Cooper and Ndyanabangi2010). Visits to health clinics can take a day or longer due to long distances traveled. Similarly, the majority of the population in rural Uganda relies on subsistence farming, which is associated with high levels of poverty (Hammond et al., Reference Hammond, Pagella, Caulfield, Fraval, Teufel, Wichern, Kihoro, Herrero, Rosenstock and Van Wijk2023). Individuals in rural settings thus face unique challenges in accessing physical and mental health care (Nicholas et al., Reference Nicholas, Joshua and Elizabeth2022; Ssemata et al., Reference Ssemata, Smythe, Sande, Menya, Hameed, Waiswa, Bannink and Kuper2024). Though several studies have examined the prevalence of depression in rural African settings (Kinyanda et al., Reference Kinyanda, Woodburn, Tugumisirize, Kagugube, Ndyanabangi and Patel2011; Nakimuli-Mpungu et al., Reference Nakimuli-Mpungu, Bass, Alexandre, Mills, Musisi, Ram, Katabira and Nachega2012; Adedeji et al., Reference Adedeji, Ma, Raji, Cha, Rasaki, Hutson, Taiwo, Charurat, Yusuf, Fehintola, Gureje and Morse2023), few studies have directly compared symptomatic HIV, asymptomatic HIV and HIV- adults living in the same rural areas of Africa with respect to both the prevalence of depression and mental health service use. Such comparisons may help clarify which groups experience greater depressive prevalence and whether that prevalence is matched by care-seeking. Classification of HIV status by physical symptoms will provide insights into the interplay between physical and mental health.
Mental health pathways in rural Africa are distinctly different from those of urban areas. Due to both the dearth of biomedical mental health services and traditional cultural preferences, many people seek help from traditional healers and religious leaders for mental illness (Olsen, Reference Olsen2017; Sarkar et al., Reference Sarkar, Bardaji, Baingana, Rivera, Criel, Bunders-Aelen and Grietens2021; Lee et al., Reference Lee, Nakaziba, Waimon, Agwang, Menon, Samuel, Dyas, Nkolo, Ingabire, Wykoff, Hobbs, Kazungu, Basiimwa, Rosenheck, Ashaba and Tsai2024c; Lee et al., Reference Lee, Coleman, Nakaziba, Terfloth, Coley, Epparla, Corbitt, Kazungu, Basiimwa, Lafferty, Cole, Agwang, Kathawala, Nkolo, Wogali, Richard, Rosenheck and Tsai2025a). The perceived etiology of a disease is associated with the treatment sought by an individual (Lee et al., Reference Lee, Mbabazi, Jevremov, Nakaweesi, Brodey, Wykoff, Sivakumar, Kazungu, Ssekalo, Rosenheck and Tsai2025c). Given the predominance of traditional and religious definitions of mental illness, religious and traditional sources of care are often preferred over biomedical sources for many individuals (Burns and Tomita, Reference Burns and Tomita2015). Similarly, mental health service pathways are pluralistic in nature in Africa, creating a complex interplay between traditional, religious and biomedical sources of care (Olsen, Reference Olsen2017). These help-seeking pathways are frequently dynamic as patients and caregivers use a combination of biomedical and traditional or faith-based pathways to access mental healthcare services (Williams et al., Reference Williams, Baldeh, Bah, Dennis, Robinson and Adeniyi2025). The use of multiple types of mental health care in resource-limited settings needs further research to understand provider choices and help-seeking pathways.
Given that the help-seeking process encompasses both formal and informal settings, our study established a conceptual framework to examine prior care-seeking behavior across three specific pathways: biomedically oriented, religious and traditional pathways. Biomedically oriented mental health care referred strictly to formal clinical services delivered by trained medical professionals within an institutional healthcare setting. Engagement with biomedically oriented mental healthcare was hypothesized to be driven by established, regular contact. Religious and traditional mental health care encompassed informal, community-based healing pathways sought from religious leaders and traditional healers, respectively. Engagement with religious and traditional mental health care pathways was hypothesized to be driven by alignment with cultural beliefs, personal values and social trust. Any mental health care was an aggregate of the biomedically oriented, traditional and religious mental health care, capturing overall mental health seeking pathways.
In this paper, we used cross-sectional survey data from rural communities in eastern Uganda to examine the prevalence of depression among samples grouped by clinical HIV status: symptomatic HIV, asymptomatic HIV and HIV-negative or unknown HIV status. We also examined patterns of previous mental and physical health service use from diverse providers including biomedical providers, religious leaders and traditional healers. Finally, we assessed factors associated with mental health service use and provider type. Our investigation of service use and type was designed to illuminate the prevalence of depression and relevant correlates of service use among adults in rural Uganda.
Methods
Study setting
This study was conducted in the rural Buyende District of eastern Uganda with a population of 403,486 people (Uganda Bureau of Statistics, 2024). This region of Uganda has 29 health centers, 19 of which are publicly operated (Republic of Uganda National Ministry of Health, 2026). In Buyende District, over 64% of the population is at least 5 km away from the nearest public health center, creating systemic barriers to healthcare access (Uganda Bureau of Statistics, 2017). The local population is characterized by limited employment and educational opportunities as 85% of adults are subsistence farmers and 45% of adults are illiterate. The study was conducted by a local non-governmental organization, Empower Through Health, which provides mental health and general medical clinical care in this area of Uganda.
Participants and sampling
Trained research assistants proficient in English and Lusoga (the local language) conducted the screenings and interviews. The questionnaire in English was translated by local language experts into Lusoga, and then back-translated to ensure accuracy. Research assistants were fellows affiliated with Empower Through Health, a local NGO. Fellows were trained to obtain written informed consent, administer the questionnaire and apply psychological first aid through 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).
Participants were required to be 18 years of age and both willing and able to provide informed consent. Informed written consent was obtained from participants in their preferred language. Participants who were unable to provide written consent provided a fingerprint in lieu of a signature. Participants were compensated with 10,000 Ugandan Shillings for their time (~$2.70 USD), an amount selected to provide fair reimbursement without compromising autonomy.
The investigators collected a sample of 304 adults based on the fixed 4-week timeframe of fieldwork, approximately half from public HIV clinics (N = 150) and half from randomly sampled members of surrounding communities based on community population registries (N = 154). Participating HIV clinics were Bugaya Health Centre III and Irundu Health Centre III due to their geographical proximity to the community study sites. The HIV+ participants were recruited through direct referrals by healthcare providers and staff, who introduced the study and sought preliminary consent for all patients to reduce selection bias. All HIV+ patients of at least 18 years of age were eligible to participate after they received their scheduled clinical care.
For the community-based sample, a stratified random sampling approach was employed in villages to recruit HIV-negative and unknown participants. A target quota of 10 HIV-negative participants was sampled from each of 15 nearby villages in Kagulu Sub-county of Buyende District due to proximity to Empower Through Health. To minimize selection bias, community health workers assisted in recruiting a random sample, using official village health records to confirm HIV status. If a randomly selected participant was HIV+ or unavailable to participate, selection continued until 10 participants were gathered. Participants who reported unknown HIV status were classified with HIV-negative participants for the purposes of analysis due to the focus on the subjective experience of HIV status (Payán et al., Reference Payán, Derose, Fulcar, Farías and Palar2019).
Measures
HIV status was determined by self-report of prior test results, allowing classification as HIV+, HIV-negative and unknown. Out of the 154 participants classified as HIV-negative, 15 reported unknown HIV status and 17 reported never being tested for HIV. Two participants reported being HIV-negative despite never being tested. HIV+ participants were separated into symptomatic HIV+ (n = 43) and asymptomatic HIV+ (asymptomatic HIV; n = 107) using the following binary question: “Are you feeling sick now from HIV?”
Participants were screened for depression using the Patient Health Questionnaire-9 (PHQ-9) (Kroenke et al., Reference Kroenke, Spitzer and Williams2001), previously validated for use in Uganda as a reliable depression screener (Kaggwa et al., Reference Kaggwa, Najjuka, Ashaba and Mamun2022a). The Lusoga translation for this study was previously used by the research team (Lee et al., Reference Lee, Miller, Mbabazi, Nakaweesi, Atim, Ning, Kisakye, Wagener, Tsai, Kazungu, Ssekalo, Ashaba, Rosenheck and Tsai2026). A cut-off score of 10 or higher on the PHQ-9 was used to designate moderate depression, with scores of 15 and 20 signifying moderately severe and severe depression, respectively. These groups were combined for the purpose of analysis due to low frequency. Due to differences in the cultural concept of depression, care was taken to characterize depression using associated somatic symptoms. Subjective assessment of physical and mental health used the stem question, “In general, how would you say your mental [physical] health is?” Participants were asked to respond on the following scale: poor, fair, good, very good, excellent. This was adapted from the single-item measure of self-rated physical health (Jylhä, Reference Jylhä2009).
Given the established relationship between loneliness and depression and stigma, the UCLA 3-Item Loneliness Scale was also used to explore loneliness in HIV+ and HIV-negative adults (Hughes et al., Reference Hughes, Waite, Hawkley and Cacioppo2004). A score of six or higher on this scale indicated loneliness. Participants were also asked about sociodemographic characteristics, including age, sex, occupation, religion, education and marital status. Finally, participants were surveyed about prior mental health and physical health care received from diverse providers using the binary yes/no stem question, “Have you seen a [traditional healer/religious leader/health care provider] for [physical/mental] health problems?”
To ensure uniform comprehension, surveys were interviewer-administered aloud in the participant’s preferred language, with research assistants trained to provide standardized conceptual clarifications as defined during fellowship preparation.
Ethics
In line with ethical best practices, any participant who screened positive for moderate or severe depression was provided with information and referrals to appropriate mental health care resources. All research procedures were approved by the institutional review boards of Yale University in the U.S. (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.
Analysis
Chi-square tests and ANOVAs were used to evaluate the statistical significance of differences between HIV groups on categorical and continuous variables (e.g., PHQ-9 score), respectively. These tests examined measures of sociodemographic and clinical characteristics, as well as past use of physical and mental health services. Additional analyses compared those reporting any previous mental health care and those who did not, as well as those who received mental health care specifically from a biomedical healthcare provider and those who did not. Both categorical and continuous measures of depression and loneliness were examined for completeness using the chi-square test statistic and F-statistic, respectively.
Logistic regression models were created to analyze the effect of HIV status and mental health status on prior mental health care sought by modality (general, biomedical, religious and traditional). These were created in a stepwise forward fashion using HIV status, followed by the PHQ-9 score and then the interaction between HIV status and PHQ-9 score. The same analysis was repeated with subjective general mental health status instead of PHQ-9 score. Covariates for the multivariate logistic regression models were selected based on a conceptual, hypothesis-driven approach. This strategy established primary clinical stratification by HIV study group, included baseline behavioral predictors of care-seeking and utilized an interaction term to test distinct care-seeking patterns across the groups. The proportion of explained variance was determined using Nagelkerke’s R 2. For models with quasi-complete separation, Firth’s penalized logistic regression was used to establish model convergence.
To ensure adequate statistical power, the research team calculated Events Per Variable (EPV) for outcome models. The primary model (Any Mental Health Care) had an EPV of 19.0, suggesting sufficient power for clinical prediction models (Meehan et al., Reference Meehan, Lewis, Fazel, Fusar-Poli, Steyerberg, Stahl and Danese2022). Models evaluating specific care sub-types with low baseline endorsement (including traditional healers) were classified as exploratory sub-analyses prone to wider confidence intervals.
Survey data were gathered and exported from KoboToolBox. All statistical tests were conducted in SAS 9.4 software (SAS Institute INC, 2013).
Results
Comparison by HIV status
Participants, classified into symptomatic HIV (n = 43), asymptomatic HIV (n = 107) and HIV-negative/unknown (n = 154) samples, showed several significant differences. There were significant differences across subgroups by sex and marital status, with the HIV-negative sample having higher proportions of male and married participants compared to both HIV+ samples (Table 1). The symptomatic HIV sample was older on average than the HIV-negative sample.
Demographic information by HIV status (N = 304)
Table 1. Long description
The table presents demographic data for a total sample of 304 individuals, divided into three groups: H I V positive with symptoms (43 individuals, 14.1 percent), H I V positive without symptoms (107 individuals, 35.2 percent), and H I V negative or unknown (154 individuals, 50.7 percent).
* Age: Mean age is highest in the symptomatic group (43.3) and lowest in the negative group (38.0).
* Sex: Significant difference (Chi-square 19.6). Females represent 72 percent of the symptomatic group, 80 percent of the asymptomatic group, and 55 percent of the negative group.
* Marital Status: Significant difference (Chi-square 22.99). 79 percent of the negative group are married, compared to 51 percent and 55 percent in the H I V positive groups.
* Religion: No significant difference. Anglican is the most common affiliation across groups (25 percent to 38 percent).
* Education: No significant difference. Over 50 percent in all groups have incomplete primary education.
* Occupation: No significant difference. Farming is the primary occupation for 69 percent to 74 percent of all participants.
Statistical significance is noted for sex and marital status at the p less than .001 level.
* p < .05.; ** p < .01.; *** p < .001.
Depression severity (categorical and continuous) and loneliness were significantly higher among symptomatic HIV compared to others (Table 2). Moderate or severe depressive symptoms (PHQ-9 ≥ 10) were observed in 34% of symptomatic HIV, 12% of asymptomatic HIV and 15% of HIV-negative participants (p < .001). Self-reported general mental health status was also significantly poorer for the symptomatic HIV subgroup with 51% reporting poor/fair mental health, compared to 24% for asymptomatic HIV and 30% for HIV-negative participants (p = .004). Reported receipt of mental health care – overall and by provider type (biomedical providers, religious leaders or traditional healers) – did not differ by HIV status (Table 3).
Health-related measures by HIV status (N = 304)
Table 2. Long description
The table is organized into five columns. The first column lists health measures, followed by three H I V status groups: H I V positive with symptoms (n = 43, 14.1 percent), H I V positive with no symptoms (n = 107, 35.2 percent), and H I V negative or unknown (n = 154, 50.7 percent). The final column provides Chi-square or F statistics.
Key data points include:
* P H Q 9 category: Symptomatic H I V positive individuals show higher rates of moderately severe to severe depression (21 percent) compared to asymptomatic (3 percent) and negative/unknown groups (5 percent). Chi-square is 24.38.
* Loneliness: 42 percent of symptomatic H I V positive individuals report being lonely, compared to 29 percent of asymptomatic and 31 percent of negative/unknown individuals. Chi-square is 2.52.
* P H Q 9 score: Mean scores are 8.8 for symptomatic, 4.8 for asymptomatic, and 5.3 for negative/unknown groups. F statistic is 12.63.
* U C L A 3-item loneliness score: Mean scores are 5.4 for symptomatic, 4.6 for asymptomatic, and 4.6 for negative/unknown groups. F statistic is 3.86.
* General physical health: 61 percent of symptomatic individuals report fair health, while 59 percent of asymptomatic individuals report good health. Chi-square is 51.93.
* General mental health: 40 percent of symptomatic individuals report fair health and 40 percent report good health, while 58 percent of the negative/unknown group report good health. Chi-square is 22.84.
Note: The ^ symbol designates a significant difference between symptomatic HIV participants and all other groups.
* p < .05.; ** p < .01.; *** p < .001.
Prior mental and physical health service use by HIV status (N = 304)
Table 3. Long description
The table is organized into five columns: Prior service use category, H I V positive with symptoms (n=43), H I V positive with no symptoms (n=107), H I V negative or unknown (n=154), and the Chi-square or F statistic.
Key data points for mental health services:
* Traditional healer: 7 percent for symptomatic, 3 percent for asymptomatic, and 3 percent for negative/unknown.
* Religious leader: 23 percent for symptomatic, 24 percent for asymptomatic, and 25 percent for negative/unknown.
* Healthcare provider: 12 percent for symptomatic, 14 percent for asymptomatic, and 12 percent for negative/unknown.
* Any mental health care: 37 percent for symptomatic, 33 percent for asymptomatic, and 31 percent for negative/unknown.
Key data points for physical health services:
* Traditional healer: 19 percent for symptomatic, 13 percent for asymptomatic, and 16 percent for negative/unknown.
* Religious leader: 44 percent for symptomatic, 44 percent for asymptomatic, and 53 percent for negative/unknown.
* Healthcare provider: 100 percent for symptomatic, 100 percent for asymptomatic, and 95 percent for negative/unknown. This category showed a statistically significant difference with a Chi-square of 6.98 (p < .05).
* Any physical healthcare: 100 percent for symptomatic, 100 percent for asymptomatic, and 95 percent for negative/unknown (Chi-square 6.98, p < .05).
* p < .05.; ** p < .01.; *** p < .001.
Patterns of service use
Of the sample population, 31% received mental health care. Religious leaders were responsible for the largest proportion of services provided (24%) followed by biomedical providers (13%) and traditional healers (4%) (Table 3). A notable proportion of participants (25%) reported utilizing more than one provider for mental health care.
A bivariate analysis of correlates identified poor self-rated mental health status, severity of depression symptoms and receipt of physical health care from a religious leader as significantly associated with prior mental health service utilization (Supplementary Tables S1–S3). Additionally, previous utilization of religious leaders or traditional healers for mental health care and higher levels of education were associated with mental health care from a biomedical provider.
Logistic regression models of care-seeking behaviors yielded similar results across provider types (Tables 4, 5). HIV status alone was not a significant correlate of prior service use. The PHQ-9 score was associated with increased odds of mental health service use across all providers with odds ratios between 1.12 (religious leader) and 1.17 (biomedical healthcare provider) for all final models. All final models utilizing HIV status and PHQ-9 score proved significant except for traditional care. No model explained more than 9.2% of the variance in care-seeking behavior as determined from Nagelkerke’s R 2. Among participants with symptomatic HIV, higher PHQ-9 scores were associated with lower odds of prior biomedical (OR = 0.80) and overall (OR = 0.85) mental health service use.
Logistic regression of mental health service use by provider as correlated with HIV status and PHQ-9 score
Table 4. Long description
The table is organized into four main vertical sections based on provider type: Biomedical (n = 39), Religious (n = 74), Traditional (n = 11), and Total (n = 95). Each section reports Beta and O R (Odds Ratio) values for three models.
* Biomedical Section:
- Symptomatic H I V minus: Model 1 Beta -0.07, O R 0.94; Model 2 Beta -0.48, O R 0.62; Model 3 Beta 1.45, O R 4.25.
- Asymptomatic H I V minus: Model 1 Beta 0.15, O R 1.16; Model 2 Beta 0.2, O R 1.23; Model 3 Beta 0.52, O R 1.69.
- P H Q 9 score: Model 2 Beta 0.1, O R 1.11 (p < .01); Model 3 Beta 0.15, O R 1.17 (p < .01).
- Interaction Sym. H I V x P H Q 9: Model 3 Beta -0.22, O R 0.80 (p < .05).
- R-squared values: Model 1 0.0014; Model 2 0.0556 (p < .05); Model 3 0.0832 (p < .05).
* Religious Section:
- Symptomatic H I V minus: Model 1 Beta -0.11, O R 0.89; Model 2 Beta -0.45, O R 0.64; Model 3 Beta 0.46, O R 1.58.
- P H Q 9 score: Model 2 O R 1.09 (p < .01); Model 3 Beta 0.11, O R 1.12 (p < .01).
- R-squared values: Model 1 0.0008; Model 2 0.0467 (p < .05); Model 3 0.0556 (p < .05).
* Traditional Section:
- Symptomatic H I V minus: Model 1 Beta 0.8, O R 2.24; Model 2 Beta 0.45, O R 1.57; Model 3 Beta 0.48, O R 1.62.
- R-squared values: Model 1 0.017; Model 2 0.0483; Model 3 0.0626.
* Total Section:
- Symptomatic H I V minus: Model 1 Beta 0.27, O R 1.31; Model 2 Beta -0.1, O R 0.91; Model 3 Beta 1.18, O R 3.25.
- P H Q 9 score: Model 2 Beta 0.1, O R 1.11 (p < .001); Model 3 Beta 0.15, O R 1.17 (p < .001).
- Interaction Sym. H I V x P H Q 9: Model 3 Beta -0.16, O R 0.85 (p < .05).
- R-squared values: Model 1 0.0044; Model 2 0.0704 (p < .01); Model 3 0.0920 (p < .001).
* p < .05.; ** p < .01.; *** p < .001.
Logistic regression of mental health service use by provider as correlated with HIV status and self-perceived mental health status
Table 5. Long description
The table is organized into four main sections based on provider type: Biomedical, Religious, Traditional, and Total. Each section evaluates parameters including Symptomatic H I V positive, Asymptomatic H I V positive, General mental health status (G M H), and interaction terms like Sym. H I V times G M H.
* Biomedical: Model 1 shows Symptomatic H I V positive with a Beta of 0.15 and O R of 1.16. Model 2 shows G M H has a significant O R of 4.18. Model 3 shows G M H with an O R of 4.90.
* Religious: Model 2 and Model 3 show highly significant correlations for G M H with O R values of 8.70 and 8.96 respectively. R-squared values increase from 0.0008 in Model 1 to 0.2493 in Model 3.
* Traditional: Model 2 shows G M H with an O R of 5.88. Model 3 shows an interaction term Sym. H I V times G M H with an O R of 2.33.
* Total: This section aggregates all providers. G M H is the strongest predictor across Model 2 (O R 11.43) and Model 3 (O R 14.46), both significant at p less than .001. The R-squared value for the total Model 3 is 0.3201.
Statistical significance is indicated by asterisks: one for p less than .05, two for p less than .01, and three for p less than .001.
* p < .05.; ** p < .01.; *** p < .001.
When general mental health status (dichotomized into poor/fair vs. others) was included, all models had significant discriminatory power (Table 5). General mental health status had odds ratios between 4.90 (biomedical mental healthcare provider) and 14.46 (any mental health provider) in final models. No interactions between HIV status and general mental health status were observed among all providers. In comparison to models using PHQ-9 score, subjective mental health status explained a greater proportion of variance in care-seeking behavior from Nagelkerke’s R 2. The most significant model (any mental health provider) explained 32% of the variance.
Discussion
Statement of principal findings
Consistent with global trends, this adult sample from rural eastern Uganda demonstrated that symptomatic HIV status was associated with greater depressive symptoms, loneliness and worse self-rated mental health than asymptomatic HIV and HIV-negative status. While these differences were anticipated, the findings establish a baseline disparity within Buyende District that allows for further analysis of care-seeking pathways. Care-seeking across pluralistic modalities most frequently involved religious leaders. Rather than HIV status or the combination of HIV and depression, severity of depression symptoms, poor self-rated mental health status and receipt of physical health care from a religious leader were associated with greater odds of pluralistic mental health service utilization. Participants with symptomatic HIV were less likely to have sought mental health care from any source as their depressive symptoms increased.
Context with existing literature
In concordance with prior research, we found significantly greater prevalence of depression symptoms among symptomatic HIV+ adults in comparison to asymptomatic HIV+ and HIV− adults in this rural Ugandan sample. Our findings highlight the link between general physical and mental health (Fiorillo et al., Reference Fiorillo, De Girolamo, Simunovic, Gureje, Isaac, Lloyd, Mari, Patel, Reif, Starostina, Summergrad and Sartorius2023). Economic and educational factors, which have been widely regarded as significant risk factors for depression (Bernard et al., Reference Bernard, Dabis and De Rekeneire2017; Pitcairn et al., Reference Pitcairn, Laverty, Chan, Oyebode, Mrejen, Pescarini, Machado and Hone2021; Admassu et al., Reference Admassu, Chen, Logie, Okumu, Mackenzie, Hakiza, Musoke, Katisi, Nakitende, Kyambadde and Mbuagbaw2024), were similar across all groups. This may be a result of the homogenous nature of a rural sample (Uganda Bureau of Statistics, 2024). While we did not find a difference in prior service utilization by sex, prior research has found significantly lower mental health service utilization among women (Semahegn et al., Reference Semahegn, Mehretie Adinew, Tiruneh, Argaw, Bekele, Tesfaye, Fesseha, Tiku, Gebrehiwet, Girma, Belete, Tesfaye, Teferi, Tadesse, Hussien, Austin, Shiferraw, Abebe Woldie, Muluneh, Tefera, Delpizzo, Tamire, Ayehu, Emaway and Makonnen2025). Likewise, some men do not permit their wives to seek mental health care due to underlying traditional masculine beliefs about mental health (Ezeugwu and Ojedokun, Reference Ezeugwu and Ojedokun2020); thus, further research is needed to examine the role of gender in care-seeking pathways. Our findings suggest that poor self-rated physical and mental health status, severity of depression symptoms and structural contexts may play a more significant role in depression prevalence than HIV status, socioeconomic or demographic factors.
Most previous studies of HIV in Africa have been based in urban settings, finding significantly higher levels of depression among symptomatic HIV as compared to asymptomatic HIV and HIV- adults (Owe-Larsson et al., Reference Owe-Larsson, Säll, Salamon and Allgulander2009; Breuer et al., Reference Breuer, Myer, Struthers and Joska2011). Our study replicated these findings. One study that examined the prevalence of depression among people with HIV (no classification by symptom status) in a rural area of Nigeria (Adedeji et al., Reference Adedeji, Ma, Raji, Cha, Rasaki, Hutson, Taiwo, Charurat, Yusuf, Fehintola, Gureje and Morse2023) found the prevalence of depression to be 16.3%, which is comparable to our findings of 19%.
Lower overall mental health service use among symptomatic HIV with increasing depressive symptoms may be reflective of competing treatment demands, with physical health often taking priority over mental health in primary care settings (Kroenke and Unutzer, Reference Kroenke and Unutzer2017). Prior literature has found poor adherence to mental health treatment in individuals with HIV and comorbid mental illness (Soller et al., Reference Soller, Kharrazi, Prentiss, Cummings, Balmas, Koopman and Israelski2011). Additionally, lower levels of social support, associated with loneliness, contribute to less frequent mental health service utilization (Parcesepe et al., Reference Parcesepe, Filiatreau, Ebasone, Dzudie, Pence, Wainberg, Yotebieng, Anastos, Pefura-Yone, Nsame, Ajeh and Nash2022). Symptomatic HIV may also face higher perceived levels of stigma, contributing to lower overall mental health service use.
Across the entire sample, the structural determinants of health, including stigma and healthcare infrastructure, may have reduced prior care-seeking behavior (Nicholas et al., Reference Nicholas, Joshua and Elizabeth2022; Komu et al., Reference Komu, Ngigi and Melson2025). Rural Uganda is characterized by a dearth of mental health providers, limiting biomedical care-seeking behavior. As such, many individuals either delay care or seek religious or traditional sources of care (Molodynski et al., Reference Molodynski, Cusack and Nixon2017). Along similar lines, social stigma may reduce care-seeking behavior out of fear of social judgment (Asiimwe et al., Reference Asiimwe, Nuwagaba-K, Dwanyen and Kasujja2023). Additionally, taking a day off from work to seek mental health care is often not feasible due to tight financial constraints and transportation barriers associated with living on low income (Bruwer et al., Reference Bruwer, Sorsdahl, Harrison, Stein, Williams and Seedat2011).
Depression is often not considered a biomedical condition in Uganda (Akol et al., Reference Akol, Moland, Babirye and Engebretsen2018; Asiimwe et al., Reference Asiimwe, Nuwagaba-K, Dwanyen and Kasujja2023), which may help explain high utilization of religious leaders for mental health concerns. We observed relatively low reported lifetime use of traditional healers compared with some prior estimates, which may reflect differences in measurement, setting or social-desirability bias if researchers were perceived to be associated with biomedical care (Lee et al., Reference Lee, Coleman, Nakaziba, Terfloth, Coley, Epparla, Corbitt, Kazungu, Basiimwa, Lafferty, Cole, Agwang, Kathawala, Nkolo, Wogali, Richard, Rosenheck and Tsai2025a). Notably, participants who endorsed prior “biomedical mental health care” later described encounters that primarily addressed somatic complaints frequently associated with depression. This pattern suggests a potential opportunity: integrating depression screening and brief evidence-based interventions into primary care encounters may improve identification and treatment in settings where mental health services are scarce.
Meaning of the study
Biomedical depression care in rural Uganda remains challenging to access due to distance, time costs, attitudes toward mental health care and limited availability of mental health services (Kisa et al., Reference Kisa, Baingana, Kajungu, Mangen, Angdembe, Gwaikolo and Cooper2016). Mental health care-seeking was common and most frequently involved religious leaders. This suggests that religious institutions serve as a locally available and culturally congruent entry point for support, aligning with our conceptual framework.
Among populations with increasing physical health treatment demands, however, mental health service use decreased. This highlights potential competing demands, higher stigma and lower social support. Patterns of care-seeking observed in this study, particularly high reliance on religious leaders and relatively limited use of biomedical services, are consistent with the possibility of important demand-side barriers alongside supply-side barriers, including limited recognition of depression as a condition warranting biomedical care. Thus, educational interventions involving traditional, religious, biomedical and community stakeholders could reduce stigma and increase mental health care access throughout rural areas.
Given the pluralistic nature of care-seeking behavior in our sample, it stands to reason that future mental health interventions should build upon biomedical, religious and traditional pathways in conjunction. In parallel, integrating depression screening into routine clinical encounters could leverage existing patterns of biomedical utilization – including visits that may involve somatic complaints associated with depression – to improve identification and access to care in settings where specialty care is scarce, especially for populations with competing demands. This could occur at local Health Centre IIIs, such as Bugaya and Irundu, to maximize limited biomedical interaction. For instance, PHQ-9 screening and mental health counseling could be integrated into standard appointments for anti-retroviral therapy among HIV+ participants and general medical appointments for all individuals. Additionally, community health workers could perform regular follow-ups for individuals identified as at risk for depression and additional mental health conditions.
Unanswered questions and future research
Future research should examine how individuals in rural settings decide which provider type to seek care for mental health concerns, including the perceived roles of traditional healers, religious leaders and biomedical providers. Further investigation into the causes of medical pluralism is warranted, particularly given the divergent etiological frameworks maintained by traditional, religious and biomedical systems. In parallel, community psychoeducation and destigmatization strategies delivered through community platforms could help address demand-side barriers to accessing biomedical care and improve recognition of depression as a treatable condition (Lee et al., Reference Lee, Rauben, Liu, Kim, Van Der Velde, Taylor, Walsh, Asasira, Katongole, Hatfield-King, Blackwell, Iheanacho, Christ and Ssekalo2022; Lee et al., Reference Lee, Christ, Blackwell, Rauben, Krause, Monnig, Ssekalo, Iheanacho and Rosenheck2024a). Finally, more granular measurement of biomedical health care, including types of medication and counseling, should be examined to map pathways to evidence-based treatment.
Study strengths
Though previous research has investigated the relationship of HIV and comorbid depression in Africa, our study was one of the few studies to directly compare HIV+ and HIV-negative adults in the same rural setting. The additional presence of HIV symptoms further strengthened our analysis of the HIV+ samples. A variety of demographic and health-related factors were considered to characterize the specific relationship between depression and care-seeking behavior, including education, occupation, religion, marital status and loneliness. Our study acknowledged the pluralistic nature of care in rural areas and examined the use of religious leaders and traditional healers for both physical and mental health care. Lastly, we utilized widely accepted, evidence-based tools for screening and evaluation such as the PHQ-9, which has been previously validated in rural Uganda (Nakku et al., Reference Nakku, Rathod, Kizza, Breuer, Mutyaba, Baron, Ssebunnya and Kigozi2016).
Study limitations
Despite methodological strengths, there were several notable limitations. Our study included subjective self-reported symptomatic HIV, not objective measures of AIDS, such as CD4 counts. Some participants reporting symptoms of HIV may meet criteria for AIDS; however, we cannot determine this based on the data collected, limiting the precision of our findings. Additionally, the specific symptoms of HIV among these individuals were not examined. Some questions in the PHQ-9 address somatic symptoms, including fatigue and sleeping problems, which could be due to HIV, depression or a combination of both. This may have contributed to higher overall PHQ-9 scores among symptomatic HIV. However, symptomatic HIV showed significantly higher scores on depressive symptoms, including little pleasure in daily activities and feeling depressed.
The two comparison groups were also recruited using different sampling frames. HIV+ participants were recruited as a clinic-based sample, whereas HIV-negative participants were drawn from a community-based random sample. Although few demographic differences were identified, this sampling method introduces selection bias and limits direct comparability between groups. HIV+ participants were recruited in a healthcare setting, which would suggest a higher likelihood of prior mental health care within biomedical pathways due to exposure to healthcare professionals. Additionally, HIV+ participants may have been less likely to report religious or traditional mental health care due to the healthcare setting of fieldwork.
The HIV-negative group included 15 participants with unknown HIV status. This classification was made given that not knowing one’s HIV status may be experienced subjectively in a similar manner to being HIV-negative (Folayan et al., Reference Folayan, Ibigbami and Lusher2022). To control this, a sensitivity analysis was run without the 15 participants of unknown HIV status. There was no longer a significant difference in physical healthcare utilization between subgroups; however, all other results remained the same.
Also, there was considerable ambiguity in the measurement of biomedical service utilization. Our study focused specifically on depression, but mental illnesses including anxiety, post-traumatic stress disorder and attention-deficit/hyperactivity are also common in HIV+ populations. The reason participants sought biomedical mental health care was not specified. Additionally, this did not distinguish between medication therapy and counseling. Furthermore, limited clarity around medication made it difficult to determine whether medications were antidepressants or medications for the somatic symptoms of depression.
Self-reported mental health service, associated with substantial risk of recall bias and social desirability bias, likely limited accuracy. Our study was similarly constrained by limited traditional pathway utilization with only 11 participants reported utilizing traditional healer care. While these exploratory models offer valuable preliminary insights into healthcare-seeking patterns, the resulting odds ratios should be interpreted with caution. These models also did not include a variety of sociodemographic and health-related variables, which may have influenced observations. Health-seeking behavior models, such as the Andersen model, were not prospectively used to provide a framework for mental health care utilization. While predisposing and need factors were measured, enabling resources and contextual-level factors were not measured in our study, limiting our evaluation of structural barriers to care. Finally, the cross-sectional design of this study prevented the inference of a causal relationship between depression symptom severity and increased care-seeking behavior.
Conclusion
In rural eastern Uganda, symptomatic HIV was associated with greater depressive prevalence; however, these individuals did not show greater mental health service use, reflecting competing demands, stigma and low social support. Mental health service use was most readily correlated with symptom severity, perceived mental health and pluralistic care. The prominence of religious mental health services supports strengthening depression identification and care across community and primary care entry points to align with cultural beliefs and social trust. Such efforts would address competing demands among symptomatic HIV and increase mental health services more broadly for individuals with poorer self-rated mental health.
Open peer review
To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2026.10334.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/gmh.2026.10334.
Data availability statement
The data that support the findings of this study are available from the corresponding author, Y.J.L., upon reasonable request.
Acknowledgments
The team is grateful for the contribution of Buyende District community health workers, who aided in data collection.
Author contribution
All authors made substantial contributions to either the conception or design of the work or the acquisition, analysis or interpretation of data. All authors drafted or revised the manuscript critically for important intellectual content and approved the final version to be published. All authors agree to be accountable for all aspects of the work and ensure that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Funding statement
Empower Through Health provided funding for this research.
Competing interests
The authors report no conflict of interests.
Ethics and approvals
In line with ethical best practices, any participant who screened positive for moderate or severe depression was provided with information and referrals to appropriate mental health care resources. All research procedures were approved by the institutional review boards of Yale University in the U.S. (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.
Open access
Comments
Yang Jae Lee, MD
Department of Psychiatry and Behavioral Sciences
Department of Global Health
University of Washington
April 27, 2026
Cambridge Prisms: Global Mental Health
Dear Editorial Board of Cambridge Prisms: Global Mental Health,
We are pleased to submit an original research article entitled “Depression and Mental Health Service Use Across Pluralistic Care Pathways Among HIV+ and HIV- Adults in Rural Uganda” for consideration for publication. We confirm that this manuscript has not been published elsewhere and is not under consideration by another journal. All authors participated in the preparation of the manuscript, approved the manuscript, and agreed with its submission to Global Mental Health.
This quantitative study explores how HIV and comorbid depression interact to influence mental health service use across pluralistic healthcare systems in rural eastern Uganda. Through the analysis of 304 structured questionnaires with adults with and without HIV, several key findings emerged. Participants with symptomatic HIV reported significantly greater depressive symptoms, loneliness, and poorer self-rated mental health, but did not have greater odds of mental health service use. In fact, as depressive symptoms increased among participants with symptomatic HIV, likelihood of mental health service use decreased. Across participants, perceived mental health status was the strongest predictor of service use from any provider type. Specifically, mental health service use frequently involved religious leaders.
This work will be of interest to Global Mental Health readers because it highlights how utilization of mental health services is influenced not only by limited service availability, but by the pluralistic nature of services and local beliefs surrounding depression etiology and treatment. The findings shed light on possible areas to target care seeking behavior. These insights align with the journal’s focus on epidemiology, individual care, treatment access, and equity of care.
Should you decide to send this manuscript for review, we suggest the following reviewers: Melanie Abas (melanie.abas@kcl.ac.uk), Charlotte Bernard (charlotte.bernard@u-bordeaux.fr), Erica Breuer (erica.breuer@newcastle.edu.au), Jonathan Burns (burns@ukzn.ac.za), Ahmad Ghashghaee (ahmad.ghashghaee1996@gmail.com), Rose Kisa (kisarose@gmail.com), Etheldreda Nakimuli-Mpungu (ethelnakimuli@chs.mak.ac.ug), Nandini D. P. Sarkar (nsarkar@itg.be), Mark Siedner (msiedner@mgh.harvard.edu), and Peter Waiswa (pwaiswa@musph.ac.ug).
Thank you for considering our manuscript. Please address all correspondences concerning this manuscript to me at leeyjae@uw.edu.
Sincerely,
Yang Jae Lee, MD
Assistant Professor, Department of Psychiatry and Global Health