The pilot study used mixed methods, with a pragmatic interrupted time series (ITS) design. Given the logistical challenges of enrolling a control group in this setting, an ITS design was selected as a rigorous quasi-experimental approach for examining intervention effects using a single population (19). Although the design was limited to four time points given feasibility constraints, this design was appropriate for a pilot study focused on examining preliminary signals of intervention effects and implementation outcomes. When combined with qualitative data, the ITS design provides valuable early evidence to inform a future, adequately powered trial (20, 21).

This study utilized quantitative survey data from adolescent MindSKILLZ participants to estimate the effect of the intervention on mental health and wellbeing. Qualitative data collected in focus group discussions with adolescent participants as well as young adult MindSKILLZ Coaches (i.e., program facilitators) was utilized to develop a deeper understanding of the experiences of intervention participants and implementers. Key informant interviews also were conducted with Department of Health personnel at the county level and with LVCT Health staff. To understand the program's acceptability and potential for sustainability, we drew on the implementation outcomes framework (22). We defined acceptability as the perception of stakeholders that the program was likable and satisfactory (22) and sustainability as the potential for the program to be continually delivered with adaptation as needed (23). Our goal was to collect initial data on acceptability and sustainability to inform future iterations and scale-up of MindSKILLZ. Acceptability is at the core of all implementation activities (24) and there is increasing recognition of the importance of planning for sustainability from a project's outset (25, 26).

MindSKILLZ utilizes Grassroot Soccer's sport- and play-based approach to learning about critical health topics. Grassroot Soccer has implemented sport-based health programming for youth since 2002 and utilizes a positive youth development approach that emphasizes fun and activity, relatable caring ‘Coach' mentors from the community, and evidence-based program content. Grassroot Soccer ‘SKILLZ' programming has shown effectiveness in improving adolescent health knowledge, building resilience and self-efficacy, and increasing uptake of health services (27–29).

The MindSKILLZ program was conceptualized by GRS, drawing guidance from key resources in adolescent mental health, such as the Inter-Agency Standing Committee (IASC) Guidelines, UNICEF's Helping Adolescents Thrive Toolkit, and relevant systematic reviews (9, 12). Key principles of trauma-informed mental health care were also integrated into MindSKILLZ, such as the importance of safety, trustworthiness, peer support, and empowerment (30).

The MindSKILLZ curriculum was then adapted and contextualized in a co-design workshop in Nairobi, Kenya in November 2022, a meeting that brought together members of the LVCT Health youth advisory committee, GRS and LVCT Health staff, representatives from County Departments of Health and the Ministry of Education. A total of 35 participants took part (21 female, 14 male) in activities across three days. Youth advisory committee members played an integral role in workshop activities, as participants discussed mental health challenges facing adolescents in Kenya, prioritized mental health topics to address in the program, developed a theory of change, and refined the program's activity framework. Engagement of government stakeholders ensured alignment with county and national level priorities for adolescent mental health. During the workshop, GRS and LVCT also discussed and agreed upon the duration and frequency of program sessions, based on experience delivering adolescent health programming. Input from Department of Health and Ministry of Education representatives, available time, resources, and ease of delivery were also considered in determining these decisions about program delivery during the workshop and in follow-up discussions.

The resulting MindSKILLZ program includes 12 sessions, covering topics including basic mental health knowledge, positive coping skills, identification of emotions, seeking care and support, identifying one's strengths, and standing up to peer pressure. A full list of session topics is included in Supplementary material, and Table 1 includes brief descriptions of several key MindSKILLZ activities.

A comic book-style print ‘magazine' was also developed to reinforce key content from in-person MindSKILLZ sessions (sample pages available in Supplementary material). GRS had used print magazines as a supplemental resource in previous programming, and team members in Kenya agreed it would be a useful addition. Program materials were field-tested with young people and translated into Swahili.

Coaches play a key role in the delivery of MindSKILLZ and other GRS programs, as they facilitate program sessions and act as positive, relatable role models for adolescent participants. As such, recruitment and training of Coaches was an important part of the intervention process. Coaches were recruited in Nairobi and Mombasa Counties by LVCT Health, with guidance from GRS. Positions were advertised via existing youth networks, such as the Ministry of Health's Youth Champions for Health Network, with candidates required to have a high school education, basic facilitation skills, and passion for working with youth and/or community development. Once hired, Coaches underwent a 5-day training to equip them with the necessary skills and knowledge to deliver MindSKILLZ: one training was held in Mombasa, and one was held in Nairobi, with Ministry of Health Officials present at both trainings.

In addition to providing key mental health information and skills of young people, the MindSKILLZ program also connected young people to additional mental health support beyond the program sessions. Coaches referred participants in need of additional support to the LVCT one2one platform (a youth hotline and digital health platform that includes chatbots, peer mentors, and professional counselors) or appropriate county health services.

The conceptual framework for the MindSKILLZ program in this pilot study is presented in Figure 1, illustrating how key program inputs and intervention activities are expected to lead to intended outcomes and ultimate impact. Key inputs for MindSKILLZ include engagement at the community level to form strong partnerships with mental health service providers (A), and engage local leadership (B). Meaningful youth engagement (C) across program design, implementation, and evaluation follows positive youth development principles central to GRS work. At the intervention components level, the MindSKILLZ program content (D) aims to improve multiple areas of knowledge, skills, and attitudes of youth, measured as intermediate outcomes (E), where the shaded outcome of mental health literacy is presented in this study. The mental health capacity building of Coaches (F) prepares them to deliver intervention sessions and develops them as positive role models and social support for adolescent participants. The longer-term outcomes measured by the study (G) are at the level of individual experiences of mental health challenges, which further input on the impact level of adolescent wellbeing to thrive (H). Additionally, referrals and linkages to mental health support services and their uptake (I), are considered intermediate outcomes. In this paper, we present the shaded intermediate and long-term outcomes with additional presentation of intermediate outcomes planned in future publications.

Data were collected in Mombasa and Nairobi Counties between August 2023 and March 2024. Quantitative and qualitative data collection were conducted by trained research assistants in sites purposively selected by LVCT through engagement with county and sub-county focal points to represent a variety of implementation settings (e.g., urban informal settlements, more affluent areas, etc.) and to include areas with high numbers of adolescents aged 10–14 years. In Nairobi County, the selected study sites were in Ruaraka and Westlands sub-Counties; participants were recruited in both school- and community-based settings, with Coaches engaging community health promoters and teachers for support. In Mombasa County, the selected study sites were in Likoni, Jomvu Changamwe, and Kisauni sub-counties; participants were recruited only in community-based settings. Inclusion criteria for adolescents were: aged 10–14 and enrolled in a MindSKILLZ intervention group in a selected study site. Within each selected study site, all participants in the selected intervention groups were invited to take part in the study. For MindSKILLZ Coaches, LVCT Health staff, and county health staff, the inclusion criteria were: involved in MindSKILLZ intervention planning or implementation and an adult (aged 18 years +).

Quantitative data were collected from adolescents. Research assistants administered the quantitative survey to participants at Time 1, and three to four weeks later at Time 2. Participants then proceeded through the 12-session MindSKILLZ program and were administered the Time 3 survey immediately after the intervention. The Time 4 survey was administered three to four weeks after Time 3. Across all time points, participants reported on measures of preliminary effectiveness. At Times 3 and 4, participants reported on measures of intervention acceptability.

Qualitative data were collected from all participant groups and included focus group discussions and key informant interviews. We conducted eight focus group discussions in total. Three focus groups with adolescent participants were held in Nairobi and two were held in Mombasa (n = 33 participants total). Two focus groups were held with Coaches in Nairobi and one was held in Mombasa (n = 23 total participants). We conducted two key informant interviews with County Health Department staff, one in Mombasa, and one in Nairobi, and three key informant interviews with LVCT Health staff (two in Nairobi and one in Mombasa). Adolescent focus group participants were purposively selected to represent both school- and community-based implementation groups, while key informant interview respondents were selected based on their involvement in MindSKILLZ program implementation and knowledge of the program. Data were collected post-intervention (i.e., after the Time 3 survey), and respondents received a transportation reimbursement for their travel to a central location.

Ethical approval for this study was obtained from Amref Health Africa's Ethics and Scientific Review Committee (ESRC-PP1389/2023). Adult participants provided written informed consent and adolescents (all minors under age 18) provided written informed assent with written parent permission.

The quantitative survey administered to adolescent participants drew on validated scales where possible and was pre-tested with adolescents prior to the first administration. Surveys addressed demographic characteristics (e.g., gender, age) and the following constructs.

Mental wellbeing: The WHO-5 is a brief, 5-item scale measuring mental wellbeing in the prior two weeks. It is among the most widely used scales for measuring psychological wellbeing, has been translated into over 30 languages, and has demonstrated acceptable psychometric properties, including among adolescents (31, 32). Scores range from 0-low wellbeing to 100-high wellbeing. A change in 10 points on the scale is considered clinically meaningful. Adolescents were classified as having greater levels of difficulty (i.e., poor wellbeing and possible depression) if they had scores of ≤ 50 (33).

Focus group discussions used a semi-structured guide, developed by the research team and field-tested with adolescents. The guide for adolescents included questions on program content and activities, likes and dislikes, recommendations for program improvement, and reported changes in behavior, knowledge, and attitudes around mental health. For example, ‘What are some of the things that you learned in MindSKILLZ sessions,' and ‘Can you give any examples of how mental health challenges can impact peoples' lives?' The guide for Coaches included questions on their views on program content and activities, their observations of changes in participants, and recommendations for improving MindSKILLZ implementation. For example, ‘How well do you think participants learned about mental health and wellbeing in MindSKILLZ,' and ‘What did you personally learn about mental health, if anything, through your work as a Coach?' The key informant interview guide with County Health Department and LVCT Health staff included questions about program acceptability and potential sustainability and scalability. For example, ‘How resource-intensive is MindSKILLZ, compared to other programs addressing the same needs?'

For quantitative data, descriptive statistics were used to summarize and tabulate data at each time point. To assess preliminary intervention effects on the outcomes of interest, we used piecewise or ‘segmented' fixed effects linear regression, a common method for interrupted time series (ITS) designs. The introduction of the MindSKILLZ intervention immediately following Time 2 (out of four time points) served as the “interruption,” creating a pre- intervention and post-intervention segment. We assessed whether there was an immediate effect of MindSKILLZ (i.e., a change in outcome level) by comparing the period prior to its introduction (i.e., from Time 1 to Time 2) with the period following implementation (i.e., from Time 2 to Time 4).

We ran fixed effects segmented regression models for each outcome in the full sample and then stratified analyses by sex, site, and age. For mental wellbeing, depression, and strengths and difficulties, we examined intervention-related changes among participants reporting greater difficulties at baseline. Specifically, we examined the proportion of participants reporting greater difficulties across the four time points and calculated the percent change between Time 2 (immediately before) and Time 3 (immediately after) the intervention, statistically compared the difference using a z test. We then used fixed effects segmented regression to examine changes following the intervention among those reporting greater difficulties at Time 1.

For all fixed effects regression models, we checked for autocorrelation for each model, and where autocorrelation was present, we accounted for it using robust estimates of variance. These analytic approaches were selected to minimize bias from stable confounders and underlying temporal trends, while acknowledging the exploratory nature of this pilot study.

For qualitative data, trained research assistants transcribed all recorded focus group discussions and interviews and translated them from Kiswahili to English where needed. Independent reviewers verified the quality of translated transcripts. Using NVivo, two research assistants coded the qualitative data, first developing a coding framework, then using a combined inductive and deductive approach and creating themes based on the framework (39). Themes were presented to the larger study team and agreed upon through discussion.

Surveys at all four time points were completed by 256 participants. Five participants whose dates of completion were not consecutive were dropped, leaving 251 participants in the final sample. Just over half of participants were female (56.2%). Participant age was relatively evenly distributed from 10 years to 14 years, with the greatest proportion of participants aged 10 years (24.3%). Most participants were in grades 4–6 at school (63.4%), followed by grades 7–9 (25.1%), then grades 1–3 (11.6%) (Table 2).

Across all measures, scores trended in the desired direction but did not reach statistical significance (Table 3). Mental wellbeing, depression, mental health stigma, support seeking, emotional symptoms, and conduct problems all moved in the desired direction from Time 2 (pre-intervention) to Time 3 (~9–12 weeks post-intervention). No notable differences were observed by gender or age. Importantly, desired changes showed signs of persisting and growing stronger for mental wellbeing, depression, and mental health support-seeking based on the mean scores at Time 4.

For mental wellbeing, average scores on the WHO-5 in the full sample ranged from 74.1 (Time 1) to 80.0 (Time 4). These scores are in the range representing the presence of mental wellbeing (i.e., 51–100). When examining average scores separately by site, participants in Mombasa appeared to show greater improvement in WHO-5 scores post-intervention compared to those in Nairobi, but these differences did not reach statistical significance (Figure 2).

For the depression measure (PHQ-9), average scores in the full sample pre-intervention were in the ‘mild depressive symptoms' range (i.e., 5–9 points). These scores decreased to the ‘no depressive symptoms' range (i.e., 0–4 points) post-intervention (scores of 4.4 at Time 3 and 3.7 at Time 4 on a range of 0–4). On the Mental Health Literacy Scale, average scores were generally in the middle of the response range for mental health stigma (between 22 and 25 on a scale of 5 to 45) and toward the upper end of the scale for support seeking (between 14.5 and 16.7 on a scale from 4 to 20). Each scale showed improvements of a few points, respectively, before and after the intervention. These differences were not statistically significant (Figure 3).

On the Strengths and Difficulties Questionnaire (Table 4), in the full sample, participants were in the ‘close to average' range for emotional symptoms (scores of 4 on a range of 0–4), conduct problems (scores of 2 to 3 on a range of 0–4), and prosocial behavior (scores of 8 on a range of 7–10). Score changes from pre- to post-intervention were small but in the desired direction.

The proportion of participants reporting mental health difficulties on the WHO-5, PHQ-9, and SDQ declined from pre-intervention (Times 1 and 2) to post intervention (Times 3 and 4) with the exception of prosocial behavior difficulty, though differences from Time 2 to Time 3 did not reach statistical significance (Table 4). On the WHO-5, we observed a 46% decrease in the proportion of participants showing poor mental wellbeing and possible depression from Time 2 (immediately pre-intervention) to Time 3 (immediately post-intervention).

Participants showing mental health difficulties at Time 1 suggest greater changes in their scores for mental wellbeing, depression, and strengths and difficulties between Time 1 and Time 4 compared to the full sample of participants (Figure 4). However, interestingly, for each of these measures, the most notable jumps in these scores occurred from Time 1 to Time 2, prior to the introduction of MindSKILLZ. Some of these changes between Times 1 and 2 were clinically significant (e.g., for mental wellbeing and depression).

MindSKILLZ participants described enhanced coping skills and improved management of anger and stress, changes that were also observed and conveyed by Coaches and key stakeholders. In focus group discussions, adolescent participants described learning through MindSKILLZ about mental health concepts, increasing their understanding and changing their attitudes about the meaning of mental health:

Adolescent respondents accurately explained the concept of mental health when prompted and provided accurate examples of common mental health challenges. Adolescents were also asked how they would advise a friend experiencing mental health challenges, and most responded that they would encourage their friend to join MindSKILLZ. If the friend would not join, many adolescents explained that they would teach their friend about what they had learned in MindSKILLZ, particularly the ‘Take 5' mindful breathing exercise. They also provided multiple examples of where a friend could seek help, such as by calling 1190 (LVCT's one2one hotline) or going to a hospital, in addition to consulting with MindSKILLZ Coaches.

MindSKILLZ participants explained that the program improved their ability to recognize and make sense of their own emotions as well as those of other people. The program helped them understand their actions and reactions to certain situations, as well as those of peers. Adolescents in both Nairobi and Mombasa Counties described valuable skills they gained from MindSKILLZ, including skills in anger and stress management and communication:

Adolescents reported feeling more confident in coping with stress and handling situations they were previously afraid of. The use of breathing exercises and the “power hand” technique were particularly popular skills that had a positive impact on the adolescents:

MindSKILLZ Coaches and key stakeholders echoed observing positive changes in adolescent participants, providing numerous examples from their intervention groups. For example:

MindSKILLZ Coaches in both Nairobi and Mombasa described how MindSKILLZ had a positive impact on them as individuals. They explained that it helped build their mental health knowledge and provided them with practical skills in managing anger and stress. For example:

One Coach in Nairobi also explained that their MindSKILLZ training helped them support a family member experiencing a mental health challenge. The Coach described learning about where to seek help and viewing the mental health problem in a similar light to a physical health problem, that there was no shame in seeking care.

Coaches highlighted that MindSKILLZ helped them manage strong emotions and resolve conflicts with others in a healthy manner. For example:

Coaches also described how their role in MindSKILLZ helped them develop more patience, and a greater appreciation for cooperation and partnership through working with their coaching partners:

MindSKILLZ was highly acceptable to participants, reflected in both the quantitative and qualitative results. On quantitative surveys, mean scores were 2.7 (standard deviation = 0.7) on the scale from 0-low to 3-high. An overwhelming majority of adolescents reported that they liked MindSKILLZ ‘a lot' (91%) and that they liked attending the sessions ‘a lot' (91%) (Table 3). Participants also reported that they found the skills they learned in MindSKILLZ to be useful (86% reporting ‘a lot'). Participants felt very comfortable raising questions to their MindSKILLZ Coach (79% reporting ‘a lot') and a large majority said that they understood the way things were explained (81% reporting ‘a lot') (Table 5).

Qualitative findings revealed that the activities in MindSKILLZ were highly acceptable to adolescents, which they found engaging and interactive. Coaches, LVCT staff, and County Health staff observed that the play-based learning medium benefited participants. Adolescent participants reported that the games they played with their peers were a lot of fun, and recalled their favorite activities vividly:

Participants provided positive feedback on the program, including use of near-peer mentor MindSKILLZ Coaches:

County health stakeholders familiar with the program highlighted how much participants seemed to enjoy MindSKILLZ:

Most LVCT and County Health staff felt that MindSKILLZ had the potential to be sustained and scaled. They described how MindSKILLZ required minimal resources for implementation, which can help facilitate its adoption and integration into various settings. These respondents emphasized that only two Coaches were needed to facilitate MindSKILLZ Program classes comprising a group of 25 children, which was considered a good Coach-participant ratio. This efficient resource utilization, according to the respondents, was useful in allowing for the enrolment of more participants without necessarily incurring additional costs. For instance:

Contrasting perspectives emerged about the program's potential sustainability and scalability from a few Coaches and one LVCT staff member in Nairobi. One Coach highlighted the unexpected cost of paying for intervention venues in areas where they had not yet established rapport with community leaders, who often later gave venue space at no cost. The LVCT staff member highlighted how additional resources (i.e., additional funds) were allocated for the pilot, noting that scaling the program could be costly: