Introduction

The World Health Organization defines mental health as “the state of mental well-being that enables people to cope with the stresses of life, realize their abilities, learn well and work well, and contribute to their community” [1]. Globally, mental health disorders affect about 970 million people, a burden that has increased since the onset of the COVID-19 pandemic due to heightened stress, isolation, and disruptions to healthcare systems [2]. Despite this growing burden, low- and middle-income countries allocate less than 2% of their health budgets to mental health services [3]. The World Health Organization has acknowledged that mental health systems are underfunded and inadequately resourced, particularly across the African region [2].

In Africa, mental health conditions now affect approximately 116 million people, more than double the 53 million affected in 1990, yet mental health remains a neglected public health priority [4]. Government spending on mental health across the continent averages less than 50 US cents per capita annually, far below the WHO-recommended US$2 for low-income countries [1]. The region also bears the highest suicide rate globally, with about 11 deaths per 100,000 people per year, compared to the global average of 9 per 100,000 [4]. Contributing factors include the shortage of psychiatrists, economic instability, widespread stigma, and limited access to care [4].

Within this broader African context, Ghana serves as a critical case for examining structural determinants of mental health. Despite enacting a Mental Health Act in 2012, Ghana allocates less than 3% of its national health budget to mental health [5]. With fewer than 100 psychiatrists serving over 30 million people, mental health care remains inaccessible for many, especially in rural regions [6]. Around 2.3 million Ghanaians suffer from mental health conditions, particularly depression and substance use disorders. Due to stigma and inadequate service coverage, individuals often seek treatment in non-clinical settings such as prayer camps [6].

This study focuses specifically on Ghana’s Upper West Region, where climate change and food insecurity increasingly affect the lives and mental well-being of smallholder farmers. In recent years, there has been mounting attention on the relationship between mental health, food security, and agricultural resilience. Climate impacts, such as droughts, floods, pests, and disease, threaten the productivity and viability of smallholder farms, defined as farms operating on less than two hectares of land, primarily for subsistence or local markets [7,8]. These disruptions reduce agricultural output and increase food insecurity, which can intensify psychological stress, anxiety, and depression among farmers [7,8].

The Food and Agriculture Organization defines food insecurity as the lack of regular access to sufficient, safe, and nutritious food necessary for an active and healthy life [7]. In 2023, over 2.3 billion people globally (28.9%) experienced moderate or severe food insecurity, including more than 864 million who faced severe shortages [8]. Africa remains disproportionately affected, with 58% of the population suffering from moderate or severe food insecurity in the same period [8].

Consequently, smallholder farmers, whose livelihoods depend heavily on ecological conditions, are particularly vulnerable to food-related stress [9]. Prior research has demonstrated strong links between food insecurity and poor mental health, especially anxiety and depression [9–11]. However, while the individual relationships between food insecurity and mental health are increasingly documented, there is limited empirical exploration of how food insecurity interacts with climate change resilience, defined as the perceived capacity to cope with or adapt to climate stressors, in shaping mental health outcomes [1].

To guide this analysis, we draw on the Social Determinants of Mental Health (SDMH) framework, which recognizes that complex social, economic, and environmental forces influence mental health. Within this framework, we position food insecurity and climate change resilience as intermediate social determinants, shaped by broader structural conditions (e.g., poverty, service access, gender norms), and predictive of individual mental health status. While there are multiple pathways linking climate and mental health (e.g., displacement, loss of income, physical health impacts), our study centers on the pathway in which climate stressors lead to food insecurity and poor resilience, which then impact mental health as shown in Fig 1. Against this backdrop, the study investigates the independent and interactive effects of food insecurity and self-reported climate change resilience on mental health among smallholder farmers in Ghana’s Upper West Region. We hypothesize that increased food insecurity and lower resilience will independently and interactively influence poorer mental health outcomes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Conceptual framework linking climate stressors, food insecurity, resilience, and mental health.

https://doi.org/10.1371/journal.pclm.0000793.g001

Study context

This research was conducted in Ghana’s Upper West Region (UWR), situated in the northwestern part of the country between longitudes 1°36′ to 3° West and latitudes 9°48′ to 11° North [26] as shown in Fig 2. The UWR covers an estimated 18,476 km², approximately 12.7% of Ghana’s total land area, and has a population of 901,502, including 440,317 males and 461,185 females [26].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Modified map of UWR showing the study area.

https://doi.org/10.1371/journal.pclm.0000793.g002

The region experiences a semi-arid climate characterized by a single rainy season from May to September, with annual rainfall ranging from 840 mm to 1,400 mm [27]. However, climate variability has increased significantly in recent decades. The UWR has witnessed more erratic rainfall, prolonged dry spells, and rising average temperatures, estimated to have increased by approximately 1°C over the past 40 years including other parts of Ghana [28,29]. Climate projections suggest the region will face continued warming of 1.5°C to 2.5°C by 2050, with shorter growing seasons and more frequent extreme weather events, including droughts and floods [30].

Socioeconomically, the UWR has the third-highest Multidimensional Poverty Index (MPI) in Ghana at 0.348, compared to the national MPI of 0.112 [31]. About 65.5% of the population live in multidimensional poverty, with the highest levels reported in Wa West (61.9%), Wa East (48.7%), Lambussie (44.2%), Nadowli-Kaleo (40.6%), and Daffiama-Bussie-Issa (38.7%) [31]. Roughly 80% of the region’s population engages in smallholder farming, producing crops and raising livestock primarily for subsistence [27]. Women, in particular, contribute through processing and trading shea nuts, which play a vital role in household food security and income generation [32].

However, the adaptive capacity of smallholder farmers in the UWR is increasingly constrained by environmental, infrastructural, and institutional challenges [32]. Around 22.8% of people in the region experience food insecurity, with vulnerable groups, such as women and the elderly, most affected [33]. Climate change has intensified food, water, and livelihood insecurities, forcing many farmers into cycles of economic instability and psychological distress [33].

While national programs such as Planting for Food and Jobs (PfJ) and One Village One Dam (1V1D) were introduced to support agricultural resilience, these initiatives often lack long-term sustainability, infrastructure, and equitable distribution, limiting their effectiveness in the UWR [25]. Inadequate roads restrict access to markets, and limited communication infrastructure hinders the timely delivery of weather forecasts and agricultural advice, exacerbating feelings of vulnerability and helplessness among farmers [31,33]. Given these intersecting challenges, the UWR presents a critical setting for examining how food insecurity and climate resilience shape mental health. A comprehensive approach that integrates mental health services into resilience and food security is essential to supporting the well-being of farming communities in this vulnerable region.

Methodology

Data collection

Ethics statement.

This study was reviewed and approved by the University of Western Ontario Non-Medical Research Ethics Board (NMREB) under Project ID 124838. Informed consent for this study was obtained for all participants through an implied consent process, as approved by the NMREB and adheres to inclusivity in global research (see S1 Text). This study is based on a cross-sectional survey conducted with 1,033 smallholder farmers engaged in the cultivation of cereals, legumes, and root and tuber crops in Ghana’s Upper West Region. Data collection took place between July 4, 2024, and July 4, 2025.For this analysis, the survey data were accessed on March 4, 2025.

Before participation, each respondent was presented with a detailed Letter of Informed Consent outlining the study’s purpose, procedures, potential risks and benefits, confidentiality protections, and their rights as participants (e.g., right to withdraw at any time without consequence). Consent was implied through the voluntary completion and submission of the survey.

To ensure ethical transparency and community trust, the consent process was documented and witnessed in the presence of each participant’s spouse, another adult household member, or a local community opinion leader (such as an Assembly member or Unit Committee member), all of whom were also informed about the study’s objectives and ethical considerations. No personal identifiers or signatures were collected to maintain participant anonymity. This approach was deemed appropriate given the minimal risk involved in the research and was consistent with approved ethical protocols. To protect participants’ privacy, all survey responses were collected anonymously and maintained in strict confidence.

The survey assessed farmers’ production and food loss dynamics for cereals, legumes, and root and tuber crops, with an emphasis on food loss rates, postharvest management services, and agroecological approaches. It also included information on relevant demographic, socioeconomic, agricultural, climate resilience, suicidal ideation, and mental and physical well-being indicators.

A structured questionnaire was used to guide all interviews. The instrument was developed based on existing validated tools and adapted to the study context. It included both closed and open-ended questions and was pre-tested to ensure clarity, reliability, and cultural relevance.

A multi-stage sampling approach was employed. First, a non-probability purposive sampling technique was used to select five districts, Nadowli-Kaleo, Daffiama-Bussie-Issa (DBI), Lambussie, Wa East, and Wa West, due to their high concentration of smallholder farmers, multidimensional poverty scores, and socio-political conditions. These districts rank among the poorest in the region, making them particularly vulnerable to climate-related agricultural risks. Within each district, farming communities were randomly selected to ensure diverse representation.

Given the large and undefined nature of the target population, Cochran’s formula for sample size determination in an infinite population was utilized to ensure a representative sample [34]. A systematic random sampling technique was then applied to select house units, starting with a randomly chosen house at the entrance of each community and surveying every fifth household thereafter until the desired sample size was met. Within each selected household, the primary male or female smallholder farmer, aged 18 or older, responded on behalf of the household. The sample size was calculated using Cochran’s formula for an infinite population [34].

where:

=Required sample size

Z= 1.96 (for 95% confidence level)

p =0.5 (assumed proportion, as the actual proportion is unknown)

e =0.03 (margin of error)

Substituting these values:

Thus, the required sample size was 1,067 households. To account for potential non-response and data inconsistencies, the sample size was further adjusted using the formula:

where r is the expected non-response rate (assumed at 5% or 0.05):

Thus, the final target sample size was 1,123 households. However, due to household availability and response rates, the actual number of surveyed households was 1,033. Despite being slightly below the ideal sample size, this number represents over 96% of the calculated requirement and remains statistically robust for the study’s objectives. The questionnaire was administered verbally through face-to-face interviews in local languages (Dagaare, Sissale, Brifor, or Waale), conducted by trained enumerators. Each interview lasted approximately 45–60 minutes, depending on the respondent’s pace and the complexity of their responses. Responses were recorded directly into Qualtrics using tablets and smartphones, enabling real-time digital entry, encrypted cloud storage, and regular quality checks throughout data collection.

Measurement of study variables

The study variables were selected guided by the SDMH framework. The dependent variable for this study is ‘Self-Rated Mental Health’. In context, we referred to “Poor Mental Health” as a state in which smallholder farmers recognize significant challenges in their psychological well-being, often reporting feelings of anxiety, depression, and dissatisfaction influenced by various external stressors (e.g., economic, environmental, or social pressures) and personal experiences. Smallholder farmers rated their mental health on a six-point scale: 1 (excellent), 2 (very good), 3 (good), 4 (satisfactory), 5 (poor), and 6 (very poor). This measure is particularly relevant as it captures individuals’ perceptions of their mental state, which may differ from clinical assessments. Responses were recoded into a new ‘Self-Rated Mental Health’ variable for more effective data analysis. In this recoding, three ordered variables were generated: good mental health [“excellent and very good” corresponds to a score of 0], satisfactory mental health [“good” and “satisfactory” correspond to a score of 1], while poor mental health was coded as [“poor” and “very poor” corresponding to a score of 3]. Blomstedt et al. have also used this method [35].

This study measured two focal independent variables: food insecurity and climate resilience. Food insecurity was assessed using the standard nine-item Household Food Insecurity Access Scale (HFIAS), a widely validated tool that captures household-level experiences with food availability, accessibility, affordability, and nutrition over a 4-week recall period [36]. To ease participants into the sensitive topic of food insecurity and to build conversational rapport, two preliminary questions were introduced at the beginning of the food security module. These questions asked whether the household had regular access to clean drinking water and cooking fuel. These items were not part of the validated HFIAS instrument and were not included in any scoring or analysis. Their inclusion was strictly for contextual purposes, recognizing that in many rural Ghanaian communities, access to water and fuel are foundational to daily food preparation. By beginning the interview with these more general, less emotionally charged topics, enumerators were able to foster trust and ensure participants felt comfortable before moving on to HFIAS questions. Following these introductions, the nine HFIAS items were administered to capture the frequency of specific food insecurity experiences, including anxiety over food availability, reductions in dietary quality and variety, and limitations in food quantity and meal frequency. Each item used a four-point Likert-type scale to indicate how often the experience occurred in the past four weeks: “never” (0), “rarely” (1–2 times), “sometimes” (3–10 times), or “often” (more than 10 times). In line with Coates et al., responses were summed to generate a total household HFIAS score ranging from 0 to 27, which was then used to categorize each household into one of four standard Household Food Insecurity Access Categories (HFIAC): food secure (score = 0), mildly food insecure (score = 1), moderately food insecure (score = 2), or severely food insecure (score = 3) [36]. This structured scoring approach preserved the validity of the HFIAS while remaining sensitive to the lived conditions of smallholder farming households in Ghana. This structured method captures the nuances of food insecurity across smallholder contexts, which is also reflected in other scholars’ work [37,38].

We also assessed smallholder farmers’ resilience to climate change using their self-assessed ability to cope with and recover from environmental stressors [39]. Farmers were asked: “How would you rate your ability to handle stress related to climate events such as floods, droughts, storms, dry spells, erratic rainfall, pests, or disease outbreaks?” Response options ranged from 1 (very poor) to 6 (excellent). These were subsequently grouped into three categories: poor resilience (1–2, coded as 3), satisfactory resilience (3–4, coded as 2), and good resilience (5–6, coded as 0). This subjective approach was selected to reflect how farmers themselves perceive and internalize their resilience, which is a key aspect of the Social Determinants of Mental Health framework. While objective indicators like access to credit, climate information, or organizational membership offer important structural insights, self-perception captures how individuals experience their own capacity to adapt, recover, and maintain psychological stability in the face of climate adversity. This method, consistent with prior studies by Pienaah et al. and Oriangi et al. [32,40], ensures that both psychological and structural dimensions of resilience are considered in relation to mental health outcomes.

Following other studies on climate change resilience, food security, and mental health in the study context [32,35,37–40], we controlled for twenty-six (26) independent variables at various levels (see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Contextual study variables and coding.

https://doi.org/10.1371/journal.pclm.0000793.t001

Data analysis

The analysis employed a three-stage approach: univariate, bivariate, and multivariate. Univariate analysis described sample characteristics. Given the ordered nature of the dependent variable, “Self-Rated Mental Health,” the study utilized ordered logistic regression (OLR) at the bivariate and multivariate stages. The bivariate analysis predicted each independent variable based on “Self-Rated Mental Health.” At the multivariate level, three separate ordered logistic regression models were estimated. Food Insecurity-Mental Health Model: This model independently examined the relationship between food insecurity and self-rated mental health, controlling for relevant covariates. Climate Resilience-Mental Health Model: This model independently assessed the effect of climate resilience on self-rated mental health using the same set of covariates. Food Insecurity-Climate Resilience-Mental Health model: this model combined both food insecurity and climate resilience to estimate their interactive effect on mental health, accounting for potential synergistic influences. All models controlled for covariates at the individual, household, and farm/community levels. Odds ratios (OR) with 95% confidence intervals were calculated, where OR > 1 indicated a greater likelihood of poor mental health, and OR < 1 indicated a lower likelihood. The OLR formula is as follows [41].

where (Y_ij ≤ 1) represents the likelihood of an event occurring, (1 − P(Y_ij ≤ 1)) indicates the likelihood that the event will not happen, α_jk is the coefficient element, X_ijk denotes the determining variables, k = 1 shows the initial, p-1 represents the final explanatory variables, and α₀, and Ω-1 are intercept components. V_ij characterizes the different terms in the model [41]. To ensure the robustness of the regression model, the Variance Inflation Factor (VIF) assessed multicollinearity, revealing that all variables had VIF values below 2.0, indicating minimal multicollinearity. The model’s reliability was further evaluated using R-squared, Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC), which suggested a good model fit. This multi-level analysis was conducted using Stata 19.

Results

Univariate results

Table 2 presents the univariate descriptive statistics. Approximately 50.1% of smallholder farmers rated their mental health as good, 40.4% as satisfactory, and 9.6% as poor. Food insecurity is a significant issue, with 36.0% of smallholder farmers being food secure, 17.9% mildly insecure, 37.1% moderately insecure, and 9.0% severely insecure. Additionally, 57.3% of smallholder farmers reported poor resilience. Nearly 49.0% lacked access to credit, and 81.0% did not receive remittances. The average household size was 7.7 members, with 30.0% of households classified as the poorest. Most smallholder farmers used human labor (82.3%) and relied on their own experience for climate information (79.9%). Drought was reported as a major climatic stressor by 68.3% of respondents.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics of measured observations.

https://doi.org/10.1371/journal.pclm.0000793.t002

Bivariate results

Table 3 presents the bivariate predictors of poor mental health. Smallholder farmers who experienced mild food insecurity (OR = 2.2, p < 0.001, CI: 1.5-3.3), moderate food insecurity (OR = 2.9, p < 0.001, CI: 2.2-3.9), and severe food insecurity (OR = 3.1, p < 0.001, CI: 2.0-5.0) were more likely to report poor mental health compared to their food-secure counterparts. Smallholder farmers who rated their resilience as satisfactory (OR = 12.3, p < 0.001, CI: 6.9-15) or poor (OR = 6.8, p < 0.001, CI: 3.8-11.7) were also more prone to poor mental health than those with good resilience. Also, smallholder farmers aged 40–49 (OR = 1.8, p < 0.001, CI: 1.2-2.6) and those aged 60 or older (OR = 4.5, p < 0.001, CI: 2.7-7.7) had a significantly higher risk of poor mental health than younger farmers. Muslim smallholder farmers had a lower risk (OR = 0.4, p < 0.001, CI: 0.3-0.5), while African Traditionalists (OR = 2.0, p < 0.001, CI: 1.2-3.2) were at an increased risk of poor mental health compared to Christians. Additionally, widowed, divorced, or separated individuals exhibited a higher likelihood of poor mental health outcomes (OR = 1.7, p < 0.01, CI: 1.1-2.5) compared to married individuals. Each additional increase in debt owed in the past year (Gh¢) was associated with poor mental health (OR = 1.00, p < 0.001, CI: 1.0-1.5). Furthermore, poorer smallholder farmers (OR = 1.6, p < 0.01, CI: 1.0-2.5) and the poorest (OR = 1.9, p < 0.001, CI: 1.3-2.7) faced greater risks of poor mental health. Households led solely by a female head in decision-making (OR =1.8, p < 0.001, CI: 1.2-2.7) were more likely to report poor mental health, while households with joint male-female decision-making (OR = 0.4, p < 0.001, CI: 0.3-0.5) showed a lower likelihood compared to those with exclusively male-led decision-making. Moreover, smallholder farmers using machinery or animals (OR = 0.5, p < 0.001, CI: 0.4-0.7) had lower odds of poor mental health outcomes than those relying solely on human labor. Smallholder farmers who received climate information from external government sources (OR = 0.2, p < 0.001, CI: 0.1-0.6) were less likely to experience poor mental health than those who relied on their own sources of information. The effects of climatic stressors on mental health were significant, with droughts (OR = 4.2, p < 0.001, CI: 2.8-6.3), floods (OR = 2.5, p < 0.001, CI: 1.5-4.0), and other stressors like erratic rainfall and pest/disease outbreaks (OR = 2.6, p < 0.001, CI: 1.2-5.5) linked to increased poor mental health. Additionally, participation in GvSP (OR = 0.4, p < 0.001, CI: 0.3-0.5) correlated with a reduced likelihood of experiencing poor mental health. Geographically, smallholder farmers in the DBI (OR = 0.2, p < 0.001, CI: 0.1-0.3) and Wa East (OR = 0.3, p < 0.001, CI: 0.2-0.5) districts were less prone to poor mental health than those in Nadowli-Kaleo.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Pairwise prediction of poor mental health of smallholder farmers.

https://doi.org/10.1371/journal.pclm.0000793.t003

Multivariate results

Relationship between food insecurity and mental health.

The results of the multiple OLR results are presented in Table 4. Food insecurity was significantly linked to poor mental health across all the individual-level, household-level, and farm/community-level. When compared to food-secure smallholder farmers, those facing mild food insecurity were 2.3 to 2.7 times more likely to report poor mental health (p < 0.001). Also, those facing moderate food insecurity were 2.5 to 2.7 times more likely to report poor mental health (p < 0.001). Subsequently, those facing severe food insecurity were approximately 2.0 to 2.2 times likely to report poor mental health (p < 0.001). Additionally, older smallholder farmers, those with more education, and those in monogamous or polygamous marriages, as well as those with higher debt levels, were more likely to experience poor mental health compared to their counterparts. Access to remittances was less associated with poor mental health, and access to credit showed a similar lesser association. Furthermore, smallholder farmers in the poorest wealth category were more likely to face poor mental health than those in the richest wealth category. Joint decision-making was less linked to poor mental health. Smallholder farmers who faced climatic stressors were more likely to experience poor mental health compared to their counterparts. Conversely, smallholder farmers with access to climate information from external government sources, access to GvSP, and those in the DBI district were less likely to experience poor mental health.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Food Insecurity-Mental Health Model: Relationship between food insecurity and mental health among smallholder farmers.

https://doi.org/10.1371/journal.pclm.0000793.t004

Relationship between climate change resilience and mental health.

The multiple OLR results are presented in Table 5. Perceived climate resilience was significantly linked to poor mental health across all the individual-level, household-level, and farm/community-level. When compared to smallholder farmers who rate their climate resilience as good, those who rated their climate resilience as satisfactory were 7.6 to 13.1 times more likely to experience poor mental health (p < 0.001). Also, those who rated their climate resilience as poor was associated with a 3.6 to 6.5 times higher likelihood to experience poor mental health (p < 0.001) compared to those with good climate resilience. Notably, most determinants of the poor mental health at the food insecurity-mental health model (see Table 4) mirrored the determinants poor mental health at the climate resilience-mental health model (see Table 5). Additionally, a decrease in households’ agricultural labour force is associated with poor mental health. Access to credit was less associated with poor mental health. Conversely, smallholder farmers in the Wa East and Wa West districts were less likely to experience poor mental health.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Climate Resilience-Mental Health Model: Relationship between resilience and mental health among smallholder farmers.

https://doi.org/10.1371/journal.pclm.0000793.t005

Interactive effect of food insecurity and climate change resilience on mental health.

Table 6 presents the final model assessing the interactive effect of food insecurity and climate resilience on poor mental health among smallholder farmers, controlling for individual, household, and community-level variables. Food insecurity was a significant and strong predictor of poor mental health across all three models and levels. In the fully adjusted model, farmers experiencing food insecurity had nearly five times greater likelihood of reporting poor mental health compared to their food-secure counterparts (OR = 4.8, p < 0.001, CI: 2.9–7.9). Likewise, poor climate resilience was significantly associated with poor mental health (OR = 3.6, p < 0.001, CI: 2.1–6.3). Notably, the interaction between food insecurity and poor climate resilience was significant (OR = 1.5, p < 0.001, CI: 1.4–1.6), indicating that the adverse mental health effects of food insecurity were exacerbated when resilience was also poor. This points to a compounding psychosocial burden in the presence of dual vulnerabilities. Additionally, at the individual level, older age groups were significantly more likely to report poor mental health. Farmers aged 60 and above had higher likelihood than those aged 18–29 (OR = 2.7, p < 0.001, CI: 1.3–5.4). Remarkably, those with secondary education or higher also showed increased likelihood of poor mental health (OR = 2.2, p < 0.001, CI: 1.4–3.2). Both monogamous and polygamous marriages were associated with increased likelihood of poor mental health (OR = 2.8 and 3.4 respectively, p < 0.001). Also, debt was consistently linked to poor mental health (OR = 1.0, p < 0.001, CI: 1.0–1.0). Household-level findings revealed that remittances had a protective effect (OR = 0.8, p < 0.01, CI: 0.5–0.2). Farmers from the poorest households were also more vulnerable to poor mental health outcomes (OR = 1.4, p < 0.01, CI: 0.8–2.4). Joint household decision-making was significantly associated with lower likelihood of poor mental health (OR = 0.5, p < 0.001, CI: 0.3–0.6). At the farm/community level, exposure to drought (OR = 5.6, p < 0.001, CI: 3.2-9.6), floods (OR = 2.0, p < 0.001, CI: 1.1-3.6), and erratic climate events (OR = 4.3, p < 0.001, CI: 1.7-10.8) increased poor mental health. In contrast, receiving climate information from external government sources (OR = 0.1, p < 0.001, CI: 0.0-0.4) and GvSP membership (OR = 0.3, p < 0.001, CI: 0.2-0.4) were protective. Geographical differences were also evident. Compared to Nadowli-Kaleo, farmers in DBI (OR = 0.3, p < 0.001, CI: 0.2–0.6), Wa East (OR = 0.1, p < 0.001, CI: 0.1–0.2), and Wa West (OR = 0.4, p < 0.001, CI: 0.3–0.7) reported significantly lower likelihoods of poor mental health, pointing to potential spatial disparities in vulnerability and access to resources. The final model demonstrated a good overall fit, as indicated by lower AIC (1534.5) and BIC (1776.6) values relative to the earlier models.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Food Insecurity-Climate Resilience-Mental Health Model: Interactive effect of food insecurity and resilience on poor mental health of smallholder farmers.

https://doi.org/10.1371/journal.pclm.0000793.t006

Discussion

Consistent with the hypothesis, food insecurity and poor climate resilience independently and interactively predict poor mental health among smallholder farmers. After adjusting for demographic, household, and community factors, smallholder farmers experiencing mild, moderate, and severe food insecurity were significantly more likely to report poor mental health compared to their food-secure counterparts. This aligns with prior evidence (e.g., Trudell et al.) linking food shortages to anxiety and depression [10], and supports the SDMH framework, which posits that structural stressors like food insecurity operate as upstream determinants of psychological distress. In contrast, Sweetland et al.’s in their research in Nigeria, Uganda, and elsewhere, found food insecurity to be unrelated to mental distress [42]. Food insecurity contributes to mental health challenges through multiple pathways [43]: nutritional deficiencies affecting cognitive functioning, economic instability creating chronic stress, and uncertainty around food access leading to anxiety, depressive symptoms, and severe mental health issues, including suicidal thoughts, suicidal planning, and repeated attempts [43]. Moreover, inadequate nutrition affects brain function and emotional regulation, with studies in the Democratic Republic of Congo and elsewhere showing an association between undernourishment and increased depressive symptoms [44]. These effects are compounded in regions like Ghana’s Upper West Region, where climate-related crop failures threaten both food and income security [9,33,38,45]. In such settings, resource scarcity may escalate household tensions and reduce individual coping capacity, particularly among severely food-insecure farmers. For example, smallholder farmers suffering economic losses due to crop failure experience heightened mental distress [46]. This financial strain can contribute to helplessness and relationship tensions, whereby in severely food-insecure households, competition for limited food resources can result in increased tension between family members.

Climate resilience also emerged as a critical determinant of mental health. Smallholder farmers who perceive their resilience as poor or satisfactory face higher mental distress than those who report good climate resilience. This corroborates existing studies that link low adaptive capacity to anxiety and depression [46]. As per the SDMH framework, exposure to environmental risks without adequate adaptive capacity can increase feelings of helplessness and hopelessness. Drawing on the SDMH framework, resilience affects mental health via multiple pathways. Individuals with low resilience may show decreased coping confidence and higher stress, leading to lower adaptation confidence to psychological distress. Lack of control over climate risks can contribute to helplessness and increased anxiety. Economic stability and social support also influence resilience; smallholder farmers in UWR from disadvantaged households often exhibit low resilience, which can lead to family tensions and altered gender dynamics [31]. Resilience also depends on resource access (e.g., irrigation and drought-resistant seeds); hence, limited access increases anxiety, as seen in Zimbabwe, where crop failures trigger depression [47]. Additionally, repeated climate shocks can be perceived as personal failures, negatively impacting mental health [9]. More broadly, social networks and institutional support can shape community resilience. Connected communities exhibit greater resilience, while isolated ones encounter more stress and mental health challenges, as seen in many developing countries [7,8]. Additionally, structural factors, such as government policies, affect resilience by shaping farmers’ access to resources and support systems [7,8]. For example, while Ghana’s Planting for Food and Jobs (PfJ) program was designed to enhance agricultural productivity and food security, its limited coverage and inconsistent input distribution have left many smallholder farmers, especially in remote areas, without adequate support to adapt to climate-related shocks [25].

Importantly, our interaction analysis showed that food insecurity remained a significant predictor of poor mental health even among farmers with low resilience, although resilience moderated the effect. This finding reinforces that food and climate stressors function not in isolation but in a reinforcing cycle of psychosocial distress, heightening anxiety, depression, and emotional strain. For instance, smallholder farmers facing climate-induced crop failures and food shortages experience heightened stress and uncertainty, which, in turn, reduces their capacity to adapt and cope effectively. Addressing these interlinked challenges requires integrated interventions that simultaneously address food security, climate resilience, and mental health, ensuring that vulnerable populations have the necessary resources to withstand environmental and psychological challenges.

Beyond the focal independent variables, several covariates also emerged as significant risk factors for poor mental health. Rather than serving solely as confounders, these variables reflect broader structural, demographic, and environmental vulnerabilities that shape psychological well-being. For instance, older smallholder farmers were more likely to experience poor mental health. In Ghana’s UWR, older farmers face physical limitations that hinder productivity and suffer from exhaustion and health problems, which can impact their mental health. Similar trends in Kanya show psychological distress linked to older Kenyans, declining strength, and increased family reliance [48]. Additionally, older smallholder farmers in UWR are often the primary breadwinners in food-insecure households; they bear the economic burdens, unlike younger smallholder farmers who diversify their income sources (e.g., through trade). In most cases in the UWR, financially insecure older smallholder farmers are more stress-vulnerable and lack access to modern agricultural techniques, which hinders their ability to adapt to climate-smart practices, thereby reducing their resilience and reliance on outdated knowledge [31]. Community factors can also impact mental health in the UWR, as many older smallholder farmers may feel isolated when their children migrate for work, thereby increasing stress.

Education also emerged as a double-edged factor. Challenging the assumption that higher education improves mental well-being, our study found that smallholder farmers in Ghana’s UWR with secondary education or higher often experience poorer mental health than uneducated peers. From the SDMH perspective, this may be explained by unmet aspirations: educated smallholder farmers may feel trapped in low-reward agricultural livelihoods despite their qualifications. Education may also raise household expectations, creating pressure to support extended family financially. In contexts where structural opportunities are limited, this mismatch between aspiration and reality can lead to psychological distress. Similar trends exist in Tanzania, where educated farmers report higher levels of psychological distress [49].

Marital structure also plays a role. Contrary to the expectation that marriage will provide social cushioning against mental distress, we found that both monogamous and polygamous marriages reported poorer mental health than single smallholder farmers. Research further indicates that women in polygamous marriages have a higher likelihood of experiencing depression compared to those in monogamous marriages [50]. Drawing from the SDMH lens, this could be explained in the context of UWR, as financial pressures often weigh heavily on married couples, with many experiencing constraints that lead to increased anxiety and depression. Additionally, household conflicts may arise from differing opinions on finances and responsibilities, and in polygamous families, the complexity of managing multiple relationships can heighten these tensions. Cultural expectations further complicate this landscape, as societal norms impose specific roles that can leave both men and women feeling inadequate and distressed. Polygamous households also tend to have larger family sizes, which can exacerbate financial and emotional strain due to limited resources, making it difficult to cope with daily pressures. Moreover, community expectations can hinder those in challenging marriages from seeking help, thereby allowing unaddressed mental health issues to worsen.

Debt was another strong risk factor. Debt significantly increases poor mental health in smallholder farmers, with even minor increases leading to anxiety, depressive symptoms, and psychological distress in the UWR. This aligns with broader evidence elsewhere in SSA that financial strain, COVID-19, climatic shocks, loan pressure, and chronic economic insecurity can deteriorate psychological resilience [51,52]. From the SDMH lens, debt influences mental health at multiple levels. For instance, at the individual level, debt can lead to frustration and isolation, leading to anxiety and stress. Debt causes financial insecurity, worry, and a sense of helplessness. Smallholder farmers may face difficulties with loan repayments due to poor harvests and market fluctuations, leading to chronic stress and feelings of self-blame. Recent evidence in SSA linked high debt to suicidal thoughts among smallholder farmers, emphasizing the psychological burden of financial instability [52].

It is therefore not surprising that access to remittances was positively associated with better mental health. This aligns with previous work where remittance-receiving households report lower stress, improved health, and better emotional well-being [53]. The SDMH framework explains how remittances ease financial burdens, enhance stability, and reduce stress in farming communities. Remittances serve as a cushion against agricultural risks, thereby lessening dependence on unpredictable yields and market fluctuations [38]. Additionally, remittances may foster stability by reducing conflicts over resources. Mounting studies indicate that remittance-supported households reported greater security, fewer marital disputes, and stronger emotional well-being, highlighting the protective role of financial inflows [54].

Joint household decision-making also emerged as a positive factor for mental health. When both male and female heads shared in farm and household decisions, psychological outcomes were better, likely due to shared responsibilities and mutual support as evident in Uganda and Tanzania [55]. Linking to the SDMH, decision-making impacts mental health through power dynamics and community norms. Unlike sole decision-making, joint decision-making correlates with better financial planning and emotional well-being [56]. Decision-making also affects resource distribution and productivity at the farm level [57]. In UWR, farms with spouses involved may exhibit higher productivity and lower stress levels due to shared responsibilities, such as planting and decision-making regarding farm produce sales. In patriarchal societies like UWR, women’s limited decision-making power can increase stress; hence, their participation in joint decision-making models can enhance emotional wellbeing [56].

Climatic stressors, such as droughts, floods, pests, diseases, and erratic rainfall, notably exacerbate poor mental health among smallholder farmers. Those facing multiple climate shocks report heightened stress, anxiety, and distress due to crop failures, income loss, and food insecurity. Studies show that African populations are more likely to face mental health disorders due to natural disasters, leading to property and life loss [57]. Prolonged droughts or frequent floods lead to decreased harvests, rising debt, and challenges in affording farm inputs, elevating farmers’ stress. Research in Tanzania indicates that repeated climate shocks correlate with chronic anxiety and depression due to inadequate coping mechanisms and financial buffers [49]. Additionally, climate stressors destabilize social structures, food security, access to services, lead to water shortages, damaged infrastructure, and market disruptions, which intensify stress [49,56]. Thus, timely access to climate information, particularly information on mitigating climate shocks, can enhance the mental health and well-being of smallholder farmers.

Consequently, our findings indicate that smallholder farmers who obtained climate information from external non-governmental and governmental sources showed better mental health outcomes compared to those who depended exclusively on self-sourced information. When smallholder farmers obtain climate information from trusted external sources, it validates their experiences and fosters a sense of belonging and community resilience [7,8]. These social connections can act as buffers against stress and uncertainty associated with climate variability, ultimately contributing to improved mental well-being. Studies have shown that access to reliable climate information from external sources was linked to better coping strategies and mental health outcomes among smallholder farmers [7,8]. Similarly, community-based climate adaptation programs enhanced farmers’ confidence in decision-making, thereby reducing anxiety related to climatic changes [7,8].

Our findings point to the urgent need for integrating psychosocial support mechanisms into agricultural and climate adaptation policies. This integration can be operationalized through several pathways. Foremost, agricultural extension services should include mental health education, counseling referrals, and stress management training for farmers. Subsequently, community-based resilience programs should be co-designed with mental health professionals to ensure psychosocial risks are addressed alongside livelihood concerns. Additionally, government and NGO-led agricultural policies (e.g., Ghana’s Planting for Food and Jobs) should include components that strengthen social cohesion, reduce isolation, and provide safe spaces for collective problem-solving. Furthermore, training frontline agricultural officers in basic mental health first aid could enable early identification and support for distressed farmers. Finally, mobile platforms delivering agricultural advisories can also integrate messages on emotional well-being, coping strategies, and where to seek help. These mechanisms are particularly important in settings like Ghana’s UWR, where social and psychological stressors compound agricultural vulnerabilities. A multi-sectoral approach linking agriculture, climate resilience, and mental health can yield more holistic and sustainable outcomes for smallholder farmers.

While this study provides valuable insights into the links between food insecurity, climate resilience, and mental health among smallholder farmers, several limitations should be acknowledged. First, the use of self-reported data, particularly on mental health status, food insecurity, and resilience, may be subject to recall or social desirability bias, potentially affecting the accuracy of responses. Second, although OLR was appropriate for analyzing the ordinal nature of the dependent variable, it assumes proportional odds (i.e., that the relationship between predictor variables and the outcome is consistent across threshold levels). This assumption may not fully capture the complexity of mental health experiences. Third, some covariates included in the models, such as household wealth, access to credit, or exposure to climate stressors, may function as both confounders and mediators in the relationship between food insecurity, resilience, and mental health. Given that the goal was to assess statistical associations rather than causal relationships, we included these variables to characterize risk factors better and adjust for their influence. Finally, the cross-sectional design of the study precludes causal inference. Future research should adopt longitudinal or mixed methods approaches to explore the temporal dynamics and lived experiences underpinning these relationships more deeply.

Conclusion and recommendations

Despite the link between food security, climate resilience, and mental health, agricultural policy in SSA rarely integrates the mental well-being needs of smallholder farmers. A broader policy framework that integrates mental health services into agricultural programs to meet the psychosocial needs of smallholder farmers is crucial. Aside from direct mental health services, indirect initiatives such as expanding drought-resistant crop varieties, improving irrigation, and providing climate adaptation training will further enhance the mental well-being of smallholder farmers. To reduce economic uncertainty and psychological distress during shocks, governments must prioritize subsidized agricultural inputs and recovery support for smallholder farmers facing climate shocks. Access to credit and strengthening of remittance channels are essential to boost financial stability. Training in climate-smart agriculture and expanding climate information services will also be timely for improving overall climate resilience and food security.

Supporting information

Acknowledgments

We thank the smallholder farmers, research assistants, and community leaders for their support during the research process.