Impact statement
Despite growing recognition that urban violence and gang recruitment constitute mental health risk factors for youth in low- and middle-income countries, few studies examine how psychosocial interventions for this population function once deployed. This study applies an implementation science framework to Parceros Creadores, a municipal government program in Medellín, Colombia, which deployed novice psychologists to deliver cognitive-behavioral therapy (CBT) sessions to youth aged 14–28 years vulnerable to gang recruitment. Medellín’s more than 350 local gangs govern civilian life in low-income neighborhoods, and youth who lack employment and educational opportunities drift toward gang membership as a default pathway. The program placed sessions at a university campus outside participants’ neighborhood contexts and used a structured protocol that psychologists could adapt to individual clinical presentations, including neurodivergent participants and those with comorbid psychiatric conditions. Psychologists associated 4 design features with positive implementation experiences. These were early-career professional staffing, structured protocols with built-in adaptation space, peer case discussion and government partnership. Feasibility obstacles included youth distrust of government programs, dropout linked to competing survival needs such as food insecurity, and bureaucratic barriers to psychiatric referrals within the public health system. Psychologists perceived that youth adopted CBT techniques in daily life and read this as support for the protocol’s appropriateness, although providers needed to distinguish between cognitive distortions amenable to restructuring and accurate assessments of genuinely adverse environments. These design features may transfer to other Latin American cities where gang presence, youth unemployment, and mental health treatment gaps co-occur, conditions shared by urban centers across Central America, Brazil and the Andean region. The study also identifies a gap in the task-shifting literature: novice professionals with formal training but limited field experience constitute a distinct implementer category from both lay health workers and experienced clinicians, with specific supervision, and support needs that program designers should anticipate.
Introduction
Mental disorders account for 60–70% of disability-adjusted life years among people aged 15–24 years (Patel et al., Reference Patel, Flisher, Hetrick and McGorry2007). At least 1 in 4 young people experiences a mental disorder in any given year, yet treatment coverage for common mental disorders in low- and middle-income countries (LMICs) reaches as low as 7%, producing a treatment gap of up to 93% (Singla et al., Reference Singla, Kohrt, Murray, Anand, Chorpita and Patel2017). The 2018 Lancet Commission on Global Mental Health reported that 1 in 27 people with depressive disorder in LMICs received minimally adequate treatment (Patel et al., Reference Patel, Saxena, Lund, Thornicroft, Baingana, Bolton, Chisholm, Collins, Cooper, Eaton, Herrman, Herzallah, Huang, Jordans, Kleinman, Medina-Mora, Morgan, Niaz, Omigbodun, Prince, Rahman, Saraceno, Sarkar, De Silva, Singh, Stein, Sunkel and Unützer2018). Young people are particularly likely to avoid formal care settings, making alternative delivery strategies essential (Patel et al., Reference Patel, Saxena, Lund, Thornicroft, Baingana, Bolton, Chisholm, Collins, Cooper, Eaton, Herrman, Herzallah, Huang, Jordans, Kleinman, Medina-Mora, Morgan, Niaz, Omigbodun, Prince, Rahman, Saraceno, Sarkar, De Silva, Singh, Stein, Sunkel and Unützer2018).
Latin American countries reproduce these gaps. Youth mental health policies across the region remain fragmented, clinically focused and often imported without local adaptation (Brum Leitão and Zacché Avellar, Reference Brum Leitão and Zacché Avellar2025; López-Gamboa and Pinto-Sosa, Reference López-Gamboa and Pinto-Sosa2025). Programs are evaluated through coverage and budget metrics rather than by examining how interventions function once deployed (Agudelo-Hernández and Rojas-Andrade, Reference Agudelo-Hernández and Rojas-Andrade2021). Colombia illustrates the structural contradiction, as the country’s mental health policy framework recognizes rights-based principles, but administrative barriers and stigma limit access to services in practice (Hernández Holguín and Sanmartín-Rueda, Reference Hernández Holguín and Sanmartín-Rueda2018).
Medellín concentrates these structural conditions and adds a violence-specific dimension. The city’s 2.6 million residents live alongside more than 350 local gangs – known as combos – that occupy virtually every low- and middle-income neighborhood (Blattman et al., Reference Blattman, Duncan, Lessing and Tobón2025). These gangs govern civilian life by mediating disputes, regulating behavior, and extracting revenue through drug sales, extortion, and loan-sharking, among other illicit activities. Gang governance persists because it protects illegal revenues; well-governed residents are less likely to inform on gang members and more willing to shield them from police (Blattman et al., Reference Blattman, Duncan, Lessing and Tobón2025, Reference Blattman, Duncan, Lessing and Tobón2026a). Combos recruit from within their own neighborhoods through social proximity rather than coercion – over 90% of members come from the territory the combo controls (Tobón et al., Reference Tobón, Sviatschi and Cabra-Ruiz2025). Youth who grow up in combo-controlled areas, lack employment and educational opportunities, and have limited spatial mobility are more likely to drift toward combo membership as a default pathway (Blattman et al., Reference Blattman, Rodríguez-Uribe and Tobón2026b). Youth unemployment in Colombia stood at 14.6% in 2025, nearly double the general rate (DANE, 2025). In 2022, youth aged 14–28 years accounted for 43% of all suicides and 70% of suicide attempts in Medellín (Medellín Cómo Vamos, 2024).
Parceros Creadores responded to this convergence of mental health need and violence exposure. We designed the program to provide psychosocial support, vocational orientation, and life skills to youth aged 14–28 years who were out of school, unemployed and exposed to violence. The Secretaría de la Juventud de Medellín – the city agency responsible for youth development and opportunities – funded the program and approved the protocols after continuous discussions with us. The program deployed early-career psychologists trained in cognitive-behavioral therapy (CBT) to deliver individual and group sessions following a structured protocol. A senior clinical psychologist on the team (ML) wrote this protocol. She specializes in cognitive-behavioral therapy and assembled the protocol from core CBT components – cognitive restructuring, behavioral activation, emotional regulation, and distress tolerance – adapted for out-of-school, unemployed, violence-exposed youth rather than drawn from a single existing manual. Its structure followed the Sustainable Transformation of Youth in Liberia program, which delivered CBT to high-risk young men in a low-income, violence-affected setting. CBT-based interventions have experimental support for reducing crime and violence among high-risk populations. Blattman et al. (Reference Blattman, Jamison and Sheridan2017) showed that CBT combined with cash grants reduced criminal behavior among high-risk men in Liberia, with effects persisting over 10 years (Blattman et al., Reference Blattman, Chaskel, Jamison and Sheridan2023). Heller et al. (Reference Heller, Shah, Guryan, Ludwig, Mullainathan and Pollack2017) found that a CBT-informed program in Chicago reduced violent-crime arrests among youth by 45–50%. These trials established that CBT can modify the automatic behavioral responses that contribute to violence, but they test delivery by experienced facilitators in controlled research settings – not by novice professionals within government program infrastructure. This design occupies a distinct position in the task-shifting literature. Most research on nonspecialist mental health delivery examines lay health workers – community volunteers, peers or auxiliary staff with no clinical training (Singla et al., Reference Singla, Kohrt, Murray, Anand, Chorpita and Patel2017). Parceros Creadores deployed a different implementer profile: recent psychology graduates with formal CBT training but minimal field experience, working within a government program. This profile differs both from the lay health workers studied in sub-Saharan Africa and South Asia (Shahmalak et al., Reference Shahmalak, Blakemore, Waheed and Waheed2019; Jacobs et al., Reference Jacobs, Myers, Van Der Westhuizen, Brooke-Sumner and Sorsdahl2020) and from the experienced clinicians assumed in efficacy trials.
Implementation science provides frameworks to examine how interventions function in practice, distinct from whether they produce clinical outcomes under controlled conditions. Most published studies from Latin America evaluate mental health programs through coverage metrics or clinical outcomes without examining the implementation processes that determine whether interventions reach their target populations and function as intended. Proctor et al. (Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) define 8 implementation outcomes that serve as “proximal indicators of implementation processes” and “key intermediate outcomes in relation to service system or clinical outcomes.” This study uses 4 of those outcomes – acceptability, feasibility, appropriateness and fidelity – that map onto the available data. The essays do not support an assessment of a fifth outcome, sustainability. They do document one of its preconditions, workforce-capacity development, which we examine separately.
Our study addresses the following question: how did novice psychologists experience the implementation of a CBT-based psychosocial intervention for gang-recruitment-vulnerable youth, and what do their accounts reveal about the acceptability, feasibility, appropriateness and adaptation demands of this program model?
Methods
Setting and intervention
Parceros Creadores operated in Medellín, Colombia, in 2024. The program targeted youth aged 14–28 years who were out of school, unemployed and exposed to violence. About 450 participants received individual CBT sessions following a structured protocol that included cognitive restructuring, behavioral activation, emotional regulation, distress tolerance and life-project planning. The protocol also included group sessions focused on social skills, assertive communication and conflict resolution. Sessions took place at Universidad EAFIT, a private university in the city – a deliberate choice to place participants in an institutional environment outside their neighborhood contexts. Of the roughly 450 enrolled youth, participants attended an average of 3 sessions, and about one-third completed all 8 sessions of the individual protocol. Dropout concentrated in the early sessions. The program did not collect structured vocational, referral or clinical-outcome data. A separate pre-registered randomized trial measures the program’s longer-term mental health and economic outcomes (AEA RCT Registry, AEARCTR-0013614). The present study examines implementation rather than those outcomes.
The Secretaría hired 14 psychologists, all under age 32, with formal training in CBT but limited postgraduation clinical experience. Professional experience among participants ranged from 3 months to almost 4 years. None had prior experience delivering mental health interventions in violence-affected community settings. These psychologists received supervision through regular peer case-discussion meetings where they reviewed cases, shared adaptation strategies and received guidance from senior staff. The peer meetings functioned as the primary quality assurance mechanism, substituting for the individual clinical supervision that experienced practitioners receive in traditional training models.
We monitored fidelity through several procedures (Bellg et al., Reference Bellg, Borrelli, Resnick, Hecht, Minicucci, Ory, Ogedegbe, Orwig, Ernst and Czajkowski2004). Every psychologist completed 21 h of training on the protocol and on CBT strategies for the presentations they were likely to encounter. ML, who wrote the protocol, delivered this training and gave each psychologist the written protocol afterward. We then administered a competency assessment scored from 1 to 5, and psychologists who scored below 3.5 received an additional individualized hour of training before they delivered sessions. Two psychologists required this additional training. In the weekly group supervision, psychologists raised questions about the protocol and reviewed specific components to maintain adherence. Each psychologist also kept a session checklist that set out the required structure and left space to record questions and difficulties, which the group then discussed in supervision.
Study design
We conducted a reflexive thematic analysis (Braun and Clarke, Reference Braun and Clarke2006, Reference Braun and Clarke2019) of the reflective essays written by the psychologists who delivered the intervention, and we read their accounts through the implementation-outcomes framework of Proctor et al. (Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011). We report the study following the Consolidated Criteria for Reporting Qualitative Research, a reporting guideline rather than a study design (COREQ; Tong et al., Reference Tong, Sainsbury and Craig2007).
Participants
Twelve of the 14 psychologists agreed to participate. Table 1 presents participant demographics. All had completed undergraduate training in psychology with CBT training. None had prior experience delivering mental health interventions in violence-affected community settings.
Participant characteristics
Table 1. Long description
The table contains four columns and twelve data rows.
* P 1: Male, 25 years old, 11 months experience.
* P 2: Female, 30 years old, 1 year 2 months experience.
* P 3: Male, 28 years old, 2 years experience.
* P 4: Female, 28 years old, 2 years 2 months experience.
* P 5: Female, 28 years old, 2 years 9 months experience.
* P 6: Female, 28 years old, 1 year 10 months experience.
* P 7: Male, 31 years old, 2 years 2 months experience.
* P 8: Female, 26 years old, 1 year 7 months experience.
* P 9: Female, 26 years old, 2 years 4 months experience.
* P 10: Female, 30 years old, 3 years 11 months experience.
* P 11: Female, 28 years old, 1 year 7 months experience.
* P 12: Female, 28 years old, 3 months experience.
Source: Authors’ elaboration.
Data collection
Each psychologist wrote a structured reflective essay about their implementation experience. The essay prompt directed psychologists to address 3 domains: (a) observed experiences and changes among the youth participants, (b) professional and personal learning from implementing the program, and (c) reflections on CBT application and program functioning. The prompt was intentionally open-ended to allow psychologists to foreground the implementation aspects they considered most relevant. Essay length ranged from approximately 500 to 2,000 words. The essays were written in Spanish, the psychologists’ native language. Participation was voluntary, and psychologists provided written consent for their essays to be analyzed for research purposes.
Analysis
We analyzed the essays using reflexive thematic analysis as described by Braun and Clarke (Reference Braun and Clarke2006, Reference Braun and Clarke2019). Reflexive thematic analysis positions themes as “creative and interpretive stories about the data, produced at the intersection of the researcher’s theoretical assumptions, their analytic resources and skill, and the data themselves” (Braun and Clarke, Reference Braun and Clarke2019). Themes do not passively emerge from data; we actively construct them through sustained engagement with the material.
Two researchers (GA and ML) conducted the analysis through the 6 phases described by Braun and Clarke (Reference Braun and Clarke2006, Reference Braun and Clarke2019). They familiarized themselves with the data through repeated reading, coded meaningful segments across the full dataset, collated codes into candidate themes, reviewed the themes against the coded extracts and the complete essays and defined and named the final themes. GA and ML held weekly meetings to discuss the coding, compare interpretations, and record analytic decisions in memos. A third researcher (ST) then mapped the inductively developed themes onto the implementation-outcomes taxonomy of Proctor et al. (Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011). This deductive step followed the inductive coding rather than guiding it. The inductive themes corresponded to 5 of Proctor et al.’s (Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) 8 implementation outcomes. Acceptability, feasibility, appropriateness, and fidelity recurred across the essays and organize the results. One theme did not fit those 4. The psychologists’ accounts of their own professional growth pointed instead to a precondition of sustainability, Proctor’s fifth outcome, which is why we examine them separately as workforce-capacity development. No theme fell entirely outside the taxonomy, and the framework guided where we placed themes after coding rather than how we named them.
All analysis was conducted on the original Spanish texts. Quotations presented in the results appear in English translation, with the original Spanish in parentheses. Two bilingual members of the author team (ML and ST) translated the quotations independently and then reconciled the minor wording differences between their versions. The substance agreed in every case.
Reflexivity statement
We designed and implemented Parceros Creadores. The Secretaría de la Juventud de Medellín funded the program and approved the intervention protocols after continuous discussions with us, but neither the psychologists who delivered the intervention nor we are Secretaría staff. This position provided insider knowledge of the program’s design logic, institutional constraints, and operational decisions. It also created a risk that the analysis would favor interpretations consistent with the program’s stated goals. To mitigate this risk, we applied reflexive thematic analysis, which foregrounds the researcher’s interpretive role rather than concealing it behind procedural objectivity (Braun and Clarke, Reference Braun and Clarke2019). We discussed coding decisions and theme development collectively to surface assumptions that individual researchers might not recognize in their own interpretations.
A second feature of our position bears on the data themselves. We designed the program, and the psychologists who wrote the essays were employed to deliver it. This relationship could have shaped what they recorded. Psychologists may have emphasized successes or muted difficulties if they understood the essays as an account rendered to the program’s designers, a more direct threat to the content of the data than bias in our later interpretation. The essays formed part of the program’s reflective practice rather than an external evaluation, which may have eased this pressure without removing it. The essays were not anonymous to us. Participation nonetheless carried no employment consequence, and we had told the psychologists in advance that we would read the essays only after their contracts and the implementation had ended. What a psychologist wrote therefore could not affect their standing in the program. We therefore treat the essays as situated provider accounts shaped by this relationship, not as independent observations of youth outcomes, and we frame the findings accordingly.
Results and discussion
The findings below report what the psychologists perceived and recorded in their essays. They are provider accounts of the implementation experience, not direct measures of youth outcomes or program effectiveness, and we keep what the psychologists reported separate from our interpretation of it.
Acceptability: Therapeutic bond formation in a context of institutional distrust
Acceptability – “the perception among implementation stakeholders that a given treatment, service, practice, or innovation is agreeable, palatable, or satisfactory” (Proctor et al., Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) – emerged as the strongest implementation outcome across the essays.
Every psychologist described youth who entered the program expecting judgment and encountered instead a space where someone listened without conditions. P2 reported that youth told her: “Here I have a pleasant place where a good person waits to talk about me, teach me, and not judge me like everyone else has” (“Aquí tengo un lugar agradable donde una persona buena me espera para conversar sobre mí, enseñarme y no juzgarme como lo han hecho las demás personas”). P8 noted that youth “thought psychology was for people with bigger problems or mental illness” (“pensó que la psicología era para gente con más problemas o con enfermedades mentales, pero veo que no”).
The psychologists described a specific mechanism behind this acceptability. Youth in Medellín’s vulnerable neighborhoods had prior experience with institutional programs perceived as politically motivated and exploitative. P3 described the initial recruitment phase: “Attracting them was not easy: many had convinced themselves that everything offered is done for political purposes and that people would really play with their vulnerability” (“Atraerlos no fue fácil: se habían muchos de ellos convencido de que todo lo que se ofrece se hace con medios políticos y que realmente jugarían con su vulnerabilidad”). In the psychologists’ accounts, the CBT sessions overcame this distrust through sustained, noninstrumental attention. The psychologists identified the therapeutic bond, rather than the cognitive-behavioral techniques, as the main source of this acceptability.
P7 articulated this distinction: “Our gaze, with whoever sits in front of us expecting some change in their life, has to be that of someone who does not want to change them, nor wants to prevent them from feeling discomfort, nor wants to improve them so they become someone else” (“nuestra mirada entonces, con quien se sienta al frente nuestro esperando algún cambio en su vida, tiene que ser la de alguien que no quiere cambiarlos, ni quiere evitarles sentir malestar, ni quiere mejorarlos para que sean otros”).
This context-specific acceptability connects to a broader pattern. Youth distrust of government programs in Medellín is not generic institutional skepticism but a response to neighborhoods where combos function as parallel governance structures – providing order, mediating disputes and managing economic life (Blattman et al., Reference Blattman, Duncan, Lessing and Tobón2025). State-sponsored programs compete with an existing authority structure that youth encounter daily. Parceros Creadores achieved acceptability, in the providers’ accounts, because it offered something the combo governance structure does not: unconditional, noninstrumental attention to individual psychological experience.
This finding resonates with Wallen et al.’s (Reference Wallen, Eberhard and Landgren2021) study of counselors at Zimbabwe’s Friendship Bench, where the therapeutic relationship – not the problem-solving therapy protocol – emerges as the central mechanism of engagement. It also aligns with Campillay-Araya et al. (Reference Campillay-Araya, Di Masso and Muñoz-Arce2023), who found that psychosocial professionals in Chile identified the relational bond, rather than technical procedures, as the foundation of intervention effectiveness. The consistency of this finding across Medellín, Zimbabwe, and Chile suggests that relational engagement may constitute a necessary precondition for the acceptability of structured psychological interventions in vulnerable populations, regardless of the specific protocol deployed.
The psychologists also reported a perceived reduction in mental health stigma among participants. P9 wrote: “Mental health is no longer perceived as a taboo or a topic only for the mentally ill; together we deconstructed the stigma around mental health and prioritized psychological well-being” (“la salud mental ya no es percibida como un tabú o un tema solo para enfermos mentales; juntos deconstruimos el estigma a la salud mental y damos prioridad al bienestar psicológico”). If confirmed, this shift would have direct implications for acceptability at scale. Mental health stigma remains one of the most consistent barriers to help-seeking among youth in Colombia and across Latin America (Hernández Holguín and Sanmartín-Rueda, Reference Hernández Holguín and Sanmartín-Rueda2018). The psychologists perceived that sustained therapeutic contact, rather than explicit anti-stigma messaging, changed these attitudes. This observation is consistent with the possibility that program design itself functions as a stigma-reduction mechanism when it creates positive direct experience with mental health services. Youth-reported and longitudinal data would be needed to test that possibility.
Feasibility: Structural and institutional barriers to sustained participation
Feasibility – “the extent to which a new treatment, or an innovation, can be successfully used or carried out within a given agency or setting” (Proctor et al., Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) – faced 3 categories of obstacles.
Dropout and distrust
Every psychologist described dropout as a persistent problem, concentrated in the program’s early phases. P3 recalled “hung-up calls, forgotten chats, blocked contacts, missed appointments” during recruitment. Youth who lacked current emotional distress particularly tended to disengage. P3 described the initial period as “desolate and silent” (“desolador y silencioso”) before incentives and persistence produced a stable participant base.
Competing survival needs
P11 documented that some youth “could only access one or two meals per day” (“algunos solo podían acceder a 1 o 2 comidas al día”), had intermittent sleep, and lacked routines. When basic material needs compete with therapy attendance, attendance is lost. P7 named this tension directly: “I have said many times that one does not do psychotherapy on a material or sociocultural condition, like unemployment or lack of education” (“Yo he dicho mucho que uno no le hace psicoterapia a una condición material o sociocultural, como el desempleo o la falta de educación”).
Institutional barriers
P1 described cases requiring psychiatric referral – patients with bipolar disorder, schizophrenia and comorbid conditions – where the public health insurance system created bureaucratic delays in accessing medication and specialist services. The program could identify clinical need but lacked the institutional authority to resolve it within the health system.
P4 also identified a specific feasibility parameter: in her account, 1-hour weekly sessions with a week between appointments worked better than compressed formats. She linked session frequency and spacing to both adherence and perceived therapeutic progress, with direct implications for program scheduling in similar settings. P5 added a temporal dimension to the feasibility assessment: in her account, therapeutic gains required sustained engagement over months, and youth who attended fewer than 8 sessions rarely showed the changes she perceived among longer-term participants. This perceived threshold suggests that programs targeting similar populations should design retention strategies to secure participation during the initial phase, when dropout risk peaks.
These feasibility obstacles reflect a pattern documented across LMIC mental health implementation. Faregh et al. (Reference Faregh, Lencucha, Ventevogel, Dubale and Kirmayer2019) argued that mental health programs must account for culture, context and community structures. The Parceros experience specifies what this means in a gang-affected urban setting: institutional distrust functions as a primary barrier to engagement, and programs must offer something qualitatively different from both state bureaucracy and criminal authority. The decision to hold sessions on a university campus – outside the territorial boundaries controlled by combos – addressed a feasibility constraint that neighborhood-based programs would face: the risk that participants associate the program with a specific gang territory or that combo members monitor attendance. Castro-Camacho et al. (Reference Castro-Camacho, Moreno and Naismith2018) encountered analogous spatial considerations when adapting the unified protocol for conflict-affected populations in Colombia, finding that the physical setting of therapy carried symbolic meaning that affected engagement and perceived safety.
Appropriateness: Perceived adoption of CBT techniques in daily life
Appropriateness – “the perceived fit, relevance, or compatibility of the innovation or evidence-based practice for a given practice setting, provider, or consumer” (Proctor et al., Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) – rested on psychologists’ accounts that youth integrated CBT techniques into their daily functioning.
The majority of essays reported that youth adopted specific techniques outside of sessions. P2 described the CBT toolkit as an “emotional first-aid kit” (“botiquín de primeros auxilios emocionales”) that youth used independently. P12 reported that youth learned “not to believe so easily what they thought and to expose themselves to doing things that previously generated fear” (“les ayudó a no creer tan fácil lo que pensaban y exponerse a hacer cosas que anteriormente les generaba temor”). P2 relayed a participant’s assessment: “Thanks to Parceros I can access serious psychology sessions, where I did learn something practical and applicable to my life, to better manage my thoughts, emotions, and decisions” (“Gracias a Parceros puedo acceder a sesiones serias de Psicología, donde sí aprendí algo práctico y aplicable a mi vida, para gestionar mejor mis pensamientos, emociones y decisiones”).
The psychologists described behavioral changes they perceived. P6 described shifts in sleep hygiene, nutritional habits, and physical activity among youth who had previously lacked any daily structure. P5 reported concrete achievements: “We witnessed concrete accomplishments, such as enrolling to finish high school or the decision to begin a detoxification process with professional support” (“fuimos testigos de logros concretos, como la inscripción para terminar el bachillerato o la decisión de comenzar un proceso de desintoxicación con el apoyo de profesionales”). These accomplishments are examples the psychologists reported, not independently verified outcomes.
Changes in self-perception formed a recurring pattern in the essays. P12 reported that youth “used to see themselves as useless, unattractive, undeserving of love, unlovable, and failures” (“solían verse como personas inútiles, poco atractivas, no merecedoras de amor, no queribles y fracasadas”). The psychologists reported that, through cognitive restructuring, youth tested the accuracy of these self-evaluations and developed more realistic self-assessments. P8 summarized the shift: the program “allows broadening the panorama of these young people, about a world beyond the neighborhood, beyond difficult friendships and family, and that one can live beyond surviving” (“permite ampliar el panorama de estos jóvenes, sobre un mundo más allá del barrio, de las amistades y familiares difíciles y que se puede vivir más allá del sobrevivir”).
This reported adoption is consistent with the finding from Hernandez Hernandez et al. (Reference Hernandez Hernandez, Waller and Hardy2020) that CBT’s core mechanisms are transportable across Latin American contexts. Woodard et al. (Reference Woodard, Mraz and Renn2023) reported a parallel finding among nonspecialist providers (NSPs) delivering brief depression treatment in the United States: participants valued CBT’s structured, goal-oriented format because it provided tangible skills they could apply between sessions. The practical, skills-oriented character of CBT aligned with what the psychologists said youth valued: concrete tools rather than open-ended exploration. A caveat, however, concerns the structural determinants of psychological distress. Several psychologists noted that youths’ beliefs about being worthless or incapable of change were not solely cognitive errors – they reflected accurate assessments of social environments where opportunities were genuinely scarce and violence genuinely present. CBT’s focus on testing the accuracy of thoughts works differently when a young person’s belief that “the world is dangerous” is empirically defensible. The psychologists navigated this tension without explicit theoretical guidance, distinguishing between beliefs that could be productively restructured and environmental assessments that required different responses – vocational support, safety planning or referral.
Fidelity and adaptation: Protocol modifications in response to population heterogeneity
Fidelity – “the degree to which an intervention was implemented as it was prescribed in the original protocol or as it was intended by the program developers” (Proctor et al., Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) – existed in productive tension with adaptations that psychologists made to accommodate participant diversity.
P1 described treating a participant with comorbid autism spectrum disorder and bipolar disorder. The standard CBT protocol did not address this clinical profile. P1 adapted the intervention to “potentiate skills through art, train parents in frustration tolerance and emotional validation, and provide a comfortable, safe space.” For participants with consolidated psychiatric diagnoses who had never received treatment, P1 prioritized psychoeducation before introducing cognitive-behavioral techniques.
P4 distinguished between participants who benefited from individual sessions and those who gained more from group work. Group sessions addressed social skills, peer normalization of experiences, and mutual accountability. P4 argued that “the need of each participant must be anticipated” (“debe concebirse anticipadamente la necesidad de cada chico y chica”) when assigning modality – a clinical judgment that exceeded the protocol’s specifications. Several psychologists also incorporated family psychoeducation that was not part of the original protocol, teaching parents emotional validation techniques. This adaptation responded to the observation that youth’s home environments often reinforced the cognitive patterns the intervention sought to modify.
P1 integrated resume writing, interview preparation and job-portal navigation into the therapeutic sessions. P11 described similar integrations, teaching time management and goal-setting skills that addressed both therapeutic and vocational objectives. These vocational components recognized that many participants’ distress was inseparable from unemployment and a lack of economic opportunity. Youth aged 14–28 in Medellín face a labor market where unemployment is nearly double the general rate and where informal or illegal economic activity often represents the most accessible source of income (DANE, 2025; Tobón et al., Reference Tobón, Sviatschi and Cabra-Ruiz2025). The program’s dual mandate – psychosocial support and vocational orientation – reflected an understanding that psychological intervention without economic pathways risks treating symptoms while leaving structural determinants intact.
These adaptations followed the pattern described by Naeem et al. (Reference Naeem, Phiri, Rathod and Ayub2019) for cultural adaptation of CBT: adjustments in delivery that respond to context without compromising the theoretical foundations of the model. The psychologists maintained the core CBT structure – cognitive restructuring, behavioral activation, thought testing – while modifying format, pace and supplementary components.
Piñeros-Leaño et al. (Reference Piñeros-Leaño, Desrosiers, Piñeros-Leaño, Moya, Cañizares-Escobar, Tam and Betancourt2024) used the ADAPT-ITT framework to systematically adapt the Youth Readiness Intervention for conflict-affected Colombian youth, documenting each adaptation through the FRAME-IS reporting protocol. Parceros Creadores followed a less formalized process: individual psychologists modified their approach based on clinical assessment and peer consultation. Both approaches produced contextually appropriate interventions, but the Parceros model relied more on practitioner autonomy than the ADAPT-ITT model did. Future implementations should consider whether structured adaptation frameworks would enhance consistency without eliminating the flexibility that psychologists identified as essential.
Workforce-capacity development as a precondition for sustainability
Sustainability – “the extent to which a newly implemented treatment is maintained or institutionalized within a service setting’s ongoing, stable operations” (Proctor et al., Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) – cannot be assessed directly from provider essays, and we do not assess it here. The essays do, however, address 1 prerequisite for sustainability: the development of professional workforce capacity. What follows examines that prerequisite, not the maintenance or institutionalization of the program.
Every psychologist described the experience as transformative for their professional development. P2 wrote: “I thought I was going to teach, but I ended up learning from them much more, and for that I am deeply grateful” (“creía que iba a enseñar, pero acabé aprendiendo de ellos un tanto más y eso lo agradezco profundamente”). P7 described a reciprocal transformation: “I arrived with the idea of contributing to them, and it was they who demonstrated through their changes, their renunciations, their acceptances, their projects and their happiness, that one can have hope, one can live better” (“Llegué con la idea de aportarles y fueron ellos quienes demostraron con sus cambios, sus renuncias, sus aceptaciones, sus proyectos y sus felicidades, que se puede tener esperanza, se puede vivir mejor”).
The peer case-discussion meetings served a specific workforce-development function. P3 and P11 identified these meetings as essential for managing cases that exceeded their training – participants with suicidal ideation, comorbid psychiatric conditions or histories of abuse. The meetings provided de facto supervision that compensated for the psychologists’ limited clinical experience. Shahmalak et al. (Reference Shahmalak, Blakemore, Waheed and Waheed2019) found similar patterns among lay health workers, where supervision “enhanced their skills and prepared them for any challenges and issues they faced in delivering the therapy.” Parceros Creadores extended this finding to a different implementer category: formally trained but early-career professionals.
Parceros Creadores occupied a position that existing literature does not adequately describe. The dominant task-shifting framework distinguishes between specialist professionals and NSPs, defined as anyone who “has not had specialized professional clinical training in a field closely related to mental health” (Singla et al., Reference Singla, Kohrt, Murray, Anand, Chorpita and Patel2017). The Parceros psychologists do not fit either category. They held professional psychology degrees and had completed CBT training, but they lacked the clinical experience that separates novice from experienced practitioners. Their challenges differed accordingly: they did not struggle with understanding the therapeutic model (a common NSP challenge per Shahmalak et al., Reference Shahmalak, Blakemore, Waheed and Waheed2019), but they did face difficulties managing clinical heterogeneity, maintaining boundaries and adapting protocol to unexpected presentations.
Jacobs et al. (Reference Jacobs, Myers, Van Der Westhuizen, Brooke-Sumner and Sorsdahl2020) identified “task dumping” as a risk when health systems assign mental health responsibilities to workers without adjusting workload, compensation or support structures. Parceros Creadores avoided this pattern through dedicated staffing: the psychologists were hired specifically for this role, with defined caseloads, peer discussion time and access to supervisory guidance. This structural feature – purpose-built positions rather than added responsibilities – appears to have protected against the demoralization and burnout that Jacobs et al. documented among South African counselors expected to absorb mental health tasks into existing roles.
The combination of structured protocol, supervised practice and direct exposure to a population rarely encountered in clinical training produced professionals with competencies that did not exist before the program. These competencies – community-based CBT delivery, adaptation for neurodivergent and psychiatrically complex clients, engagement of institutionally distrustful populations – constitute workforce capacity that outlasts the program itself. P6 described this transformation directly: the experience changed her understanding of the psychologist’s role from a clinical technician applying protocols to a professional who reads context, adapts strategy and maintains therapeutic goals amid conditions that clinical training does not simulate. This workforce development function has implications for mental health human resources planning in Colombia, where Agudelo-Hernández and Rojas-Andrade (Reference Agudelo-Hernández and Rojas-Andrade2021) identified the absence of implementation-oriented training as a barrier to scaling evidence-based interventions.
Limitations
This study has 6 limitations. First, the data consist of self-selected reflective essays written by the implementing psychologists. This single-perspective design captures provider experience but does not triangulate with youth perspectives, clinical outcome data or independent observation. We did not triangulate the reported changes in stigma, behavior or daily functioning against youth-reported, clinical or administrative data. These remain provider perceptions. Second, social desirability may have influenced the essays, particularly given that the psychologists knew their reflections would be reviewed within the program structure. Third, the sample of 12 psychologists from a single program in 1 city limits transferability. Fourth, the essays were written after implementation concluded, introducing retrospective reconstruction bias. Fifth, the study cannot assess clinical effectiveness: the implementation outcomes examined here are necessary preconditions for clinical outcomes (Proctor et al., Reference Proctor, Silmere, Raghavan, Hovmand, Aarons, Bunger, Griffey and Hensley2011) but do not substitute for them. Sixth, the study does not examine the program’s cost structure or its cost relative to alternative service delivery models, which would inform decisions about scalability and resource allocation in municipal budgets.
The absence of youth voice is a particularly consequential limitation. The psychologists reported what the youth said and did, but these are filtered observations, not direct participant accounts. A complete implementation evaluation would require complementary data from the youth participants themselves, using methods designed to capture the perspectives of populations with justified distrust of institutional research.
Conclusions
Parceros Creadores operated in a distinct implementation niche: structured CBT delivery by early-career professionals within a government-sponsored violence-prevention program targeting gang-recruitment-vulnerable youth. The qualitative analysis of provider experiences yields 4 observations about implementing mental health programs in similar LMIC urban settings. Each reflects what the psychologists experienced and perceived, not measured youth outcomes. The clearest contribution concerns the providers themselves. The study documents how early-career psychologists experienced delivering a structured, CBT-informed intervention in a violence-affected, government-supported setting, and what they identified as the demands and supports of that role.
First, the psychologists identified the therapeutic bond, rather than the CBT protocol itself, as the basis of acceptability among a population with deep institutional distrust. Their accounts suggest that programs targeting gang-affected youth may need to invest in relational engagement before structured therapeutic work can take hold.
Second, feasibility depends on addressing the material conditions that compete with therapy attendance. Food insecurity, housing instability, and survival demands constrain participation in ways that program designers cannot resolve solely through scheduling adjustments. Integrated service models that combine psychosocial support with economic and social interventions may improve retention.
Third, CBT’s practical, skill-oriented structure aligned with what the psychologists reported youth valued, but providers need guidance in navigating the tension between cognitive restructuring and the accurate perception of genuinely adverse environments.
Fourth, novice professionals with formal training but limited field experience constitute a viable, underexamined workforce model for LMIC mental health service delivery. Their success depended on specific structural supports – purpose-built positions, peer case discussion, protocol flexibility – that should be treated as design requirements rather than optional enhancements. Programs that deploy early-career professionals without these supports risk the “task dumping” dynamic documented by Jacobs et al. (Reference Jacobs, Myers, Van Der Westhuizen, Brooke-Sumner and Sorsdahl2020) in South Africa.
Three design features may support similar work in Latin American cities that share Medellín’s combination of gang presence, youth unemployment, and gaps in mental health treatment (Atienzo et al., Reference Atienzo, Baxter and Kaltenthaler2016; Luna Manzanero, Reference Luna Manzanero2021). Given the single-site, single-perspective design, we offer this as a hypothesis for further study rather than an established result. The staffing model – young professionals of a similar age to participants – facilitated relational engagement. The structured protocol with built-in adaptation space gave psychologists a therapeutic scaffold without preventing responsive practice. The municipal government partnership provided institutional legitimacy and logistical support that standalone projects would have difficulty sustaining. Future research should complement provider perspectives with direct youth accounts, examine whether structured adaptation frameworks, such as ADAPT-ITT (Piñeros-Leaño et al., Reference Piñeros-Leaño, Desrosiers, Piñeros-Leaño, Moya, Cañizares-Escobar, Tam and Betancourt2024), can enhance consistency while preserving the practitioner autonomy these psychologists identified as essential and assess clinical outcomes through controlled evaluation designs. Comparative studies across Latin American cities that share Medellín’s combination of gang governance, youth exclusion and mental health service gaps would test the generalizability of these design features beyond the Colombian context.
Open peer review
To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2026.10282.
Data availability statement
The dataset, which includes individual reflective essays, is not publicly available due to confidentiality agreements with the participating psychologists.
Acknowledgements
The authors are grateful to Yeni Amaya, Isabel C. Gutiérrez, Lucía Jaramillo, Ricardo Jaramillo, Yesica Mazo, Gregorio Restrepo and Maria J. Sotomayor for coordinating the implementation of the program and facilitating this research.
Author contribution
The authors made the following contributions: conceptualization: G.A., M.L., S.T. and M.V.-A.; data curation: G.A. and M.L.; formal analysis: G.A., M.L. and S.T.; methodology: G.A. and M.L.; writing - original draft preparation: G.A., M.L. and S.T.; writing - review and editing: G.A. and M.L.
Financial support
This research was funded by Universidad EAFIT and Secretaría de la Juventud de Medellín.
Competing interests
None.
Ethics statement
This study was approved by the Institutional Review Board of Universidad EAFIT. The psychologists provided written informed consent for their essays to be analyzed for research purposes. All participant data were anonymized using numeric codes (P1–P12).
Open access
Comments
No accompanying comment.