This sample included 134 leaders and community members (Table 1).

Given the accessibility of the population, a non-probabilistic convenience snowball sampling technique was used. Inclusion criteria required participants to be over 18 years of age, currently living in any of the fifteen municipalities and have experienced some form of violence related to armed conflict.

Several questionnaires were administered to characterize the participants and to evaluate MH, social support, and sense of community.

Sociodemographic Characteristics. A comprehensive set of sociodemographic variables that enables the assessment of the relevant dimensions of inequality within the SDH framework (Bezo et al., 2012) was included in the study. These variables were age, sex, department, group affiliation, income, educational level, religion, and history of receiving psychosocial care.

Mental Health Assessment. Two validated questionnaires were employed to assess MH:

The Patient Health Questionnaire (PHQ-4) (Kroenke et al., 2009) consisting of four items with response options ranging from “not at all” (0 points) to “almost every day” (3 points), reflecting the frequency of symptoms of anxiety and depression in adults over the past two weeks. The Colombian validation of the scale was used (Kocalevent et al., 2014), it showed good internal consistency (α = 0.84), validity was established through confirmatory and discriminant analyses, and the fit indices indicated a very good model fit.

The Depression-Anxiety-Stress Scale (DASS-21) (Antony et al., 1998), which consists of 21 items reflecting depression, anxiety and stress symptoms over the past week ranging from 0 (“Did not apply to me at all”) to 3 (“Applied to me very much, or most of the time”). The Colombian validation of the test by Ruiz et al. (2017) was used, demonstrating high internal consistency (α ranging from 0.92 to 0.95) and a consistent hierarchical factor structure that supports its validity.

Social Support Assessment. Recognizing that social capital is a critical SDH (Bezo et al., 2012), and it is defined as the social support, network size, trustworthiness, reciprocity and social interactions available to an individual (Carrillo & Riera, 2017; Lagaert et al., 2021), we used the Medical Outcomes Study Social Support Survey (MOS) (Sherbourne & Stewart, 1991) to assess social support. The MOS consists of 20 items, the item 1 refers to the size of the social network, while the remaining 19 items assess four dimensions of functional social support: emotional/informational support – ES (8 items), instrumental support – IS (4 items), positive social interaction – PSI (4 items), and affective support – AS (3 items). Response options are provided on a Likert scale ranging from 1 (Never) to 5 (Always). The overall social support – OSS is calculated from the sum of the items. Higher scores indicate greater perceived social support. We used a version validated for Colombia (Londoño et al., 2012), which reported a total Cronbach’s alpha of 0.941, with values ranging from 0.921 to 0.736 for the different components and whose validity was established through confirmatory analysis.

Sense of Community Assessment. To measure the meso-level of social capital, understood as collective efficacy, community support, trust, civic participation, engagement and reciprocity (Carrillo & Riera, 2017), we used the Brief Sense of Community Scale (BSCS). This scale consists of 8 items that evaluates four dimensions: needs fulfillment – NF (2 items), group membership – GM (2 items), reciprocal influence – RI (2 items), and emotional connection – EC (2 items); each item in a 5-point Likert scale ranging from “Strongly agree” to “Strongly disagree” where higher scores indicate a greater sense of community. The overall sense of community – OSC is calculated from the sum of the items (Peterson et al., 2008). The Colombian validated version was used (Palacio-Sañudo et al., 2023), which had a high internal consistency (α = 0.918) and the validity of the test were confirmed through exploratory and confirmatory factor analysis.

Descriptive analyses were conducted for all sociodemographic variables (Table 1). Given that continuous variables violated the assumption of normality (Kolmogorov–Smirnov test), results are reported as medians and interquartile ranges (IQR). Categorical variables are presented as absolute frequencies and percentages. Bivariate non-parametric tests (Mann–Whitney, Kruskal–Wallis, and Spearman correlations) were used to explore associations between sociodemographic characteristics and MH, social support, and sense of community indicators, with a 95% confidence level. These analyses were performed in SPSS version 26.

To explore the complex interrelations among study variables, we conducted a network analysis using the Extended Bayesian Information Criterion Graphical Lasso (EBICglasso) method (Isvoranu & Epskamp, 2023) via the Qgraph package in JASP version 19. This approach estimates regularized Gaussian graphical models (GGM), which model partial correlations between pairs of variables while controlling for all others, assuming no unmeasured common causes. GGMs are well suited for psychological data as they focus on the interplay among observed variables and are robust in small samples with sparse structures (Burger et al., 2023). The EBICglasso procedure enhances the graphical lasso by applying an EBIC-based penalty, balancing model fit and complexity while promoting sparsity—weak edges are set to zero, yielding a parsimonious and interpretable network. In the estimated network, nodes represent observed variables and edges represent conditional associations: blue edges indicate positive relationships, and red edges indicate negative ones (Figure 1). Network properties were evaluated through node expected influence (EI), which accounts for both the number and the sign of connections, and bridge strength (BS), which quantifies the extent to which a node connects different communities of variables. This allowed us to identify key variables with the greatest influence and potential bridging roles across domains such as MH, social support, and community sense.

The sample included 167 leaders, 107 of which identified as female, and 60 as male. Participants ranged in age from 18 to 78 years (M = 43.4; SD = 1.3). As for their department of origin, 90 of them were from Bolívar and 77 from Sucre. Although some of these participants also took part in the quantitative component, due to the anonymity of the data collection it is not possible to determine any overlap. Therefore, the qualitative and quantitative samples are treated as independent, with data that complement each other and contribute to a deeper analysis and understanding of the SDMH in communities of the Montes de María.

To provide context and support the analysis of the quantitative data, thirty-five focus groups and two regional meetings were conducted to further explore the SDMH in the region. These encounters were led by experts from the research team using a question guide about MH, SDH, effects of the armed conflict, adverse experiences, stressors, community support, community relations, and interactions with MH services. The focus groups and regional meetings were audio-recorded, transcribed, and analyzed using NVivo 14.

A content analysis was conducted, with concurrent data analysis and collection and three phases of coding: open, axial, and selective (Schreier, 2014). In the open coding phase, a meticulous line-by-line coding approach was employed, leading to the development of a preliminary codebook, which was continuously reviewed and refined by the research team. Axial coding facilitated the aggregation of data into broader categories, establishing interconnections among them. Selective coding enabled the formulation of overarching categories that elucidated the SDMH within the municipalities.

The analysis adhered to the criterion of saturation (Hennink et al., 2019). To ensure methodological rigor (Korstjens & Moser, 2018), the principles of credibility and transferability were carefully upheld. This was achieved through prolonged engagement in the municipalities over several years, fostering an in-depth comprehension of contextual intricacies. The validity of the analyses was further reinforced by presenting the findings to participants during two regional meetings. Throughout the research process, the researchers engaged in a continuous reflexive exercise.

The primary characteristics of SDH among the 134 leaders and community members from the fifteen municipalities of Montes de María that participated in the quantitative phase are presented in Table 1. Participants were predominantly Afro-Colombian, rural farmers, and indigenous individuals. Additionally, members of the LGBTIQ+ community and 35 participants from other demographic groups, particularly young individuals, were also included. Most participants were women (59%). The age range of the participants was between 18 and 68 years, with a median age of 31 [25–48] for women and 37 [26–47] for men. In terms of socioeconomic status, 71% of the participants earned less than the minimum monthly wage (USD 268), and 41.1% earned less than USD 115.61, a significant percentage given that the poverty line in Colombia for 2023 was set at USD 108. Six participants had no formal education, and eleven had only completed elementary education. The predominant religion among participants was catholicism.

Furthermore, 64.2% had received psychosocial care. Among those 86 respondents, 56.98% received support from only one institution, 16.28% from two, 17.44% from three, and 9.30% from four or more institutions. The most commonly reported providers of psychosocial care were the Psychosocial Care and Comprehensive Health Program for Victims (PAPSIVI) with 46.51%, followed by Health Promoting Entities (EPS) (44.19%), other participants received care from community organizations and nongovernmental Organizations (NGOs). Among those who received psychosocial care, the most common types of interventions were group talks (90%), followed by workshops (27.5%), individual psychological counseling (20%), psychosocial training sessions (5%), and family psychological counseling (2.5%).

Regarding MH outcomes, on the PHQ-4 depression scale 21% (14 men and 14 women) had scores indicating the presence of depressive symptoms. On the PHQ-4 anxiety scale 11% (10 women and 5 men) had scores indicating the presence of anxiety symptoms. Regarding the DASS-21, on the depression scale 62% (n = 83) did not have scores indicative of depression, while 38% had depressive symptoms (mild = 25; moderate = 17; severe = 4; extremely severe = 4). On the anxiety scale 62.7% (n = 84) did not show anxiety symptoms, while 37.3% did (mild = 4; moderate = 24; severe = 7; extremely severe = 11). The stress scale showed that 82% (n = 110) did not have stress, while 18% had stress symptoms (mild = 10; moderate = 10; severe = 3; extremely severe = 1).

Regarding the assessment of social support, the MOS indicated that 23.1% (n = 31) of the participants scored below the established mean for this instrument, while 76.9% (n = 103) scored above the mean. Similarly, the BSCS revealed that 21.6% (n = 29) had a low sense of community, 19.4% (n = 26) had a moderate sense of community, and 59% (n = 79) exhibited a high sense of community.

In the bivariate analyses conducted (Table 1), negative correlations were observed between age and PHQ-4 anxiety and DASS stress. Additionally, age positively correlated with the MOS tangible support scale, needs fulfillment, influence, emotional connection of the BSCS, and the overall index of the sense of community of the BSCS. It was found that Afro-Colombians and the LGBTIQ+ community had significantly higher median scores in PHQ-4 depression (p = 0.04). Farmers and indigenous people exhibited significantly higher median scores in the GM dimension of the BSCS (p = 0.04); furthermore, farmers demonstrated a significantly higher median in the EC dimension of the BSCS (p = 0.05) (Table 1). Significantly higher median scores for stress were identified in the group with a technical education level and among those practicing the Christian religion (Table 1). Conversely, the group that received psychosocial support achieved a significantly higher median in the BSCS dimensions.

The EBICglasso network analysis examined the conditional relationships between MH domains —depression, stress, and anxiety—and sociodemographic variables, perceived social support, and sense of community (Figure 1). The estimated network comprised 23 nodes and 74 non-zero edges out of a possible 253, yielding a sparsity of 0.708. Regarding MH variables, DASS Anxiety (BS = 0.651, EI = 0.346) (Table 2) emerged as a major connector in the network, linking stress and depression with both social support indicators and sense of community dimensions. DASS Depression (BS = 0.979, EI = 1.180) and DASS Stress (BS = 1.110, EI = 0.327) also displayed significant connectivity, suggesting that these MH indicators share strong reciprocal associations with psychosocial resources (Figure 2).

Centrality analyses revealed that dimensions of the BSCS, particularly OSC (BS = 2.149, EI = 2.252), occupied a highly central position in the network (Table 2). This node maintained strong conditional connections with multiple indicators of social support and MH, suggesting that a greater sense of belonging and identification with the community is closely linked to lower distress and higher perceived resources. Other BSCS dimensions, such as EC (BS = 0.548, EI = 0.564) and RI (BS = 0.265, EI = 0.473), also showed substantial centrality, highlighting their structural role in bridging community perceptions with both psychosocial and emotional well-being variables. All social support dimensions were meaningfully integrated into the network, OSS showed the highest connectivity within this group (BS = 1.877, EI = 2.003), reflecting strong conditional links with both MH and sense of community variables: ES (BS = 0.520, EI = 0.512), PSI (BS = 0.379, EI = 0.546), and AS (BS = 0.375, EI = 0.583). Together, these results suggest that diverse facets of perceived support contribute substantially to the cohesion of the psychosocial network, potentially acting as protective factors that link community connectedness with emotional well-being.

These patterns indicate that higher perceived social support and stronger community bonds may mitigate psychological distress, while lower scores in these domains could amplify symptoms through interconnected pathways. The analysis also revealed meaningful links between sense of community and social support variables, indicating that individuals who report a stronger identification with their community also tend to perceive higher levels of available support. Sociodemographic variables, however, showed comparatively weaker integration into the network. For example, age exhibited negative values across all centrality metrics (BS = –1.133, EI = –1.338), suggesting limited direct influence on the main psychological constructs when controlling for other variables.

Psychosocial care demonstrated moderate centrality (BS = -1.231, EI = 0.846). This dimension does not serve as a primary hub in the network, playing a secondary role compared to the most central nodes such as OSC and OSS, suggesting a supportive but not dominant influence within the overall psychosocial structure.

The understanding of MH in Montes de María comes from a contextual and community perspective, that goes beyond the individual and pathological perception of MH: “With mental health and land, we have enough to live”. Consequently, they identified the following elements play an important part in their mental well-being:

“Mental health is the balance of trust a person has in their social context. It is the security that allows them to move freely, the trust and security that lets them enter their land without worrying about mines, the balance that lets them eat food without fearing it’s stolen, that allows them to leave their house at any hour of the night without thinking that a thief is waiting for you. It is that balance, that trust in their context.”

Armed conflict is one of the factors most strongly related to the current MH issues in their communities:

Each one of these victimizing events had individual psychological impacts that are still present in the ones who experienced them. However, a widespread effect that was felt by every member of the Montes de María community, even if they weren’t direct victims of the armed groups, was the fragmentation of the social ties and the loss of trust in others. This is greatly contrasted with the environment that existed before the armed conflict, described as a free, peaceful, united and supportive community, where the resources were shared, and nothing was lacking:

As a result, an entire generation has been affected, which in turn creates an intergenerational effect, related to the continuation and normalization of violence as a part of the culture, the rise in family violence cases in the region and other social issues such as drug addiction and prostitution. There was also a constant reminder by the participants that this violence has yet to cease in the region, as many armed groups have returned, bringing back fears, worries, and restrictions that were common during the worst periods of the armed conflict. This fear and insecurity are experienced especially by the leaders, who risk their lives while working for the rights of their communities, which in turn affects their well-being.

There is a noticeable lack of action from the public forces, which enables the continuation of problems such as drug use; this inaction is also reflected in the negligence towards infrastructure in sectors like education, healthcare, road systems, and adequate housing, which hinders access to basic needs such as electricity and water. The state’s abandonment is particularly evident among victims of the armed conflict, as the reparations plans and the restoration of their rights, including land restitution, have not been fully implemented:

There is also a perceived lack of commitment from institutions and their officials due to the absence of continuous actions or failure to meet their commitments to the territory, which hinders the development of the necessary trust to report victimizing events. This is related to a belief that the government is not going to be able to protect and guarantee the communities’ safety, which creates a constant state of fear. Another reason to mistrust the institutions is related to corruption in the territory. The community perceives a possible association between the outlaw groups with government entities, which is why there is a perception that reporting any victimizing act won’t work or could even have negative consequences.

Another common problem is the lack of resources, unemployment, and lack of opportunities in the region, which worsened after COVID-19. These issues are closely related to feelings of sadness, hopelessness, frustration and stress, and physical symptoms such as sleep issues, headaches, and tachycardia, often due to not seeing a way out of the situation.

This particularly affects young people, who have few opportunities for higher education or employment to generate income, leading to unclear life goals, many dropping out from school or having to move to other parts of the country to search for those opportunities. It is also related to young people making problematic decisions such as drug abuse, prostitution, suicide, or joining armed groups that offer a higher income than they could earn legally:

Additionally, this problem extends to adults who find a lack of resources to support and develop projects, associations, and organizations working in the territory. There is also a lack of guarantees for farmers, who are forced to sell their lands and products at lower prices, leading to economic hardship and constant worry and stress.

There are severe barriers to access MH services for the inhabitants of the region. The leaders recognize that the impacts of armed conflict on their MH should have been addressed soon after the victimizing events, and the fact that in many cases this didn’t happen has worsened the MH outcomes:

As in the network analysis, psychosocial care only showed secondary associations with MH compared to more central nodes. Some participants had received psychosocial support from government programs like PAPSIVI, however, these programs have been highly criticized by the communities, with many people having negative experiences, as on many occasions, they report the support they are given worsens the wounds that the conflict has left or creates new wounds. This is related to multiple problems identified with the services such as a focus on achieving goals and metrics, rather than offering a high-quality attention, an insufficient number of sessions, infrequent attention, unfinished processes and only implementing group interventions, offering few possibilities of having individual interventions when needed.

Besides the programs, other of the main barriers to ensure MH are the non-existence, inadequate functioning, and/or lack of awareness of MH care pathways, which are even less clear in the rural areas, where people must travel extensive distances to access any services. There are also public policies that are not being correctly implemented or are insufficient to deal with all the MH struggles of the community. All these barriers have made community members less likely to seek psychological support. When they are finally able to access MH services, they frequently find poorly trained providers, who sometimes lack sensibility and are likely to make more harmful actions out of unfamiliarity with the population, the issues they are experiencing, and unawareness of their specific needs.

Consistent with the findings from the quantitative component regarding factors related to social support and sense of community, the qualitative component revealed that despite all the victimizing events, several community members have found the strength to move forward and have organized, creating community initiatives to recover what was lost, such as forming community action boards, sports teams, youth groups, and bringing back artistic and cultural practices. This has allowed the communities to integrate and restore relationships that were broken during the conflict, promoting broader support networks and rebuilding the social fabric. It has also made it possible to heal the distrust instilled by armed groups in the territories, and restore the sense of community which are elements of social capital. These spaces were also recognized as a protective factor for MH and social challenges that are common in the region, such as substance use disorders, mood and anxiety problems, and youth crime.

This progress was hindered by the social isolation measures imposed due to COVID-19, which limited social interactions, community gathering spaces and traditional practices in the region, such as mourning rituals and ancestral practices of indigenous and afro-descendant communities. This made it harder to cope with the emotional challenges of the pandemic, as community spaces and interactions were identified within the community as crucial in these processes.

Regardless of the various social determinants that affect the MH of Montes de María’s communities, the community leaders are making constant efforts to fight for the rights of their people and secure better opportunities and resources for them. This advocacy has led to improvements such as road and infrastructure maintenance and expanded access to essential services like water and electricity. They are also advocating for reparation from the State for the violence they were exposed to, for the reduction of the violence that is still present and for access to quality healthcare and MH services within the community.

These resistance initiatives have reinforced forms of social support at a community level, such as trust, mutual support, civic participation and engagement. They have also had positive effects on the leaders’ MH and sense of community, as many of them find great satisfaction in knowing they are making a positive impact in their communities. This sense of accomplishment strengthens their collective belief in their ability to create change and motivates them to continue their efforts to achieve these goals.