The trial is a randomized controlled trial with two arms: the Shamiri intervention and a study-skills control intervention. The intervention and control conditions are described in detail later and summarized in Table 1. Interventions will be delivered in schools to groups of 10–13 youths by high school graduates trained to be lay-providers. Assessments will be administered at initial screening, 2-week midpoint, 4-week endpoint, and 2-weeks post endpoint. Academic performance data on grades and national exams will be collected for the school-term before, during, and after study participation.
This study will take place in four secondary schools located in Nairobi and neighboring Kiambu Counties, Kenya. Nairobi County is one of 47 geographical counties that act as units of devolved government in Kenya, a low-income country in Sub-Saharan Africa. Nairobi is a diverse city with a population of at least 3.2 million that is representative of the cultural and socioeconomic diversity in Kenya (20). Kiambu County, a county that is part of the former Central Province region, neighbors Nairobi County and has a population of at least 1.7 million (20), The first author will draw a list of public secondary schools in Nairobi and Kiambu counties. Principals of the schools will be contacted with information about the study and asked to enlist their school to participate. From the pool of interested schools, four schools that have strong demographic, socioeconomic and educational diversity will be selected. In particular, the selection will be done in such a manner that schools that are ranked as national secondary schools (which have a socio-economically diverse student body from all the regions in Kenya) and sub-county secondary schools(which are often resource-poor day single-sex or mixed secondary schools with student body from the local area) by the Ministry of Education, Science and Technology will be selected—more information on the diversity of secondary schools in Nairobi and Kiambu counties can be seen in Ndetei et al., and Osborn et al (21,22). In total, the final school sample will consist of an all-boys national boarding secondary school, an all-girls national boarding secondary school, an all-girls sub-county day secondary school, and a sub-county mixed-day secondary school. This diverse and generalizable sample of high-symptom youth in Kenya will also allow us to conduct robust moderator analyses that can allow us to identify for whom the Shamiri intervention works for.
Participant recruitment and screening.
In all four participating schools, students in Form One, Form Two, Form Three (and in smaller schools Form Four), will be notified about the study by the school principals and other administrators. All students aged between 13 and 18, inclusive, will be eligible for the study if they self-report either elevated depression symptoms or elevated anxiety symptoms—indexed by a score of 15 or higher on the Patient Health Questionnaire (PHQ)–8 (indicating moderate or severe depression) or a score of 10 or greater on the GAD–7 (indicating moderate or severe anxiety) (23,24). No other exclusion criteria will be applied. Of eligible students, approximately 140 - 180 will be randomly selected at each school to participate in the study. See Randomization and Treatment Allocation for more information on selection and randomization.
Students will complete a questionnaire battery to determine study eligibility. Parental consent will be sought for all minor participants. Because many of the students will not have direct access to their parents and because of the logistical difficulties of the study team obtaining written consent from parents who live in many different parts of the country, the Ethics Review Committee allowed school principals and other administrators to seek parental/guardian permission from parents of the students interested in the study using their locally available policies and resources. Only students for which parental permission will be obtained through this process will be allowed to participate in the study. School administrators will then gather these students in classrooms or school halls where they will meet study team. Here they will be further informed about the study and will be informed that their participation is voluntary and that they can opt out. After being provided with an opportunity to ask questions, they will provide informed consent if they wish to participate. Only participants for whom informed consent or informed assent and parental consent (in the case adolescent minors) is received will be allowed to participate. See Figure 1 for a participant flowchart.
Randomization and treatment allocation.
Following baseline assessment and recruitment, youths who meet the inclusion criteria will be randomly allocated within each school to the Shamiri intervention or the study-skills control condition using a computerized random number generator in R Studio. We will use stratified randomization to ensure consistent baseline characteristics (age, form, and gender if mixed school) across groups.
Group leader selection and training.
Both the Shamiri and study-skills groups will be led by trained group leaders who will serve as lay-providers. To qualify as a group leader, an individual must be a Kenyan high school graduate between the ages of 17 and 26, inclusive, and he or she must live in the local community in and around Nairobi. While most group leaders will be either planning to matriculate or already attending a local university in Nairobi, neither college admittance nor attendance is a requirement for consideration. The above group leader characteristics are important for the successful implementation of this protocol for several reasons: 1) Local group leaders’ understanding of the culture, language, and local customs in schools make them better-positioned to deliver the intervention than outsiders. 2) Group leaders who are closer in age to the youths in the program may be better positioned than older adults to connect with youths—this is because Kenyan cultural norms around age relations dictate that youths should primarily listen rather than speak in the company of adults. 3) Lastly, this model is potentially scalable: given that the mean age in Kenya is 19, there is a large pool of youths aged 17–25 from which group leaders could be selected - median age is 19 (27).
We will recruit group leaders through advertisements in high school graduate platforms on various media such as WhatsApp and Facebook. We will also recruit through local universities and organizations in Kenya that work with high school graduates. All interested group leaders will be offered an opportunity to express their interest via a brief online form. Group leaders who meet the above-specified criteria will be invited for a 25-minute interview with members of the study team. This interview will assess past teaching experiences, interest in the project, familiarity with mental health and wellness issues, and interpersonal facilitation skills. Chosen group leaders will then be contacted via email and phone and invited to a mandatory 10-hour training. Due to funding constraints, we intend to recruit 12 group leaders. All group leaders will be compensated $150 plus the cost of transportation for the entire duration of the study.
The group leader training, which will be conducted by members of the research team, will begin with four hours of general communication and group leadership skills such as active listening, noting connections between group members, handling conflicts within the groups, and referring students in need to appropriate school officials if necessary. Group leaders will then be trained didactically in the specific content of each week of the interventions. All group leaders will receive detailed and structured outlines of the content of each intervention and control session. The outlines will include sample wording for the group leaders. The last six of the ten hours of training will consist of role-plays in which the leaders practice leading groups in portions of the protocol. Following the role-plays, leaders receive feedback from their fellow group leaders and from the study team. The leaders are then asked in the full group setting to reflect on their experiences and learnings from the role-plays.
Upon completion of training, group leaders will be randomly assigned to lead either Shamiri intervention or control groups. Given that we will have about 48 groups of up to 15 youths in the entire study, group leaders will be randomly assigned to either the intervention or control condition at each school. That is, the randomization process will be repeated for each distinct school. For instance, one group leader might lead a study-skills control group in School A but a Shamiri intervention group in School B. This randomization should allow us account for group leader differences in competence and leadership style. However, group leaders will be instructed to strictly follow the protocol manual for the condition that they will be leading and not use content from the other condition during their sessions.
The Shamiri intervention consists of three modules: a growth-mindset module (16,17) lasting two sessions; a gratitude module (18) lasting one session; and a virtues module based on value affirmations (19) lasting one session. More detailed information on these sessions is provided below.
Growth-mindset interventions. Growth-mindset interventions target implicit theories about the ability of human attributes to evolve over time. As such, growth-mindset interventions challenge the belief that personality traits and intelligence are fixed and unchangeable. These interventions are designed to strengthen individuals’ beliefs in the malleability of such characteristics (28–30). Meta-analyses of growth mindset interventions have shown robust improvements in academic achievement, particularly among low-income students (31).
Gratitude interventions. Research on trait gratitude suggests that grateful people exhibit more positive states and outcomes than non-grateful people. Gratitude interventions teach individuals to be more aware of and thankful for the positive people and things in their lives, regardless of whether they are tangible or not (18).
Value affirmations interventions. Value affirmation interventions invite people to reflect on specific values that are most important to them. This self-reflection can improve people’s coping skills by reestablishing an awareness of personal worth and integrity (19,32). Due to differences in language and vocabulary, the discussion of values will be presented to Kenyan youths as “virtues.”
Structure of intervention sessions. The intervention consists of four intervention sessions, which are each one-hour long and one week apart. All of the sessions include reading activities, group discussions, and writing activities. Homework exercises are assigned between sessions. In sessions one and two, participants learn about growth, neuroplasticity, and work to build their growth-mindsets. In sessions three and four, participants learn about gratitude and value affirmations respectively.
Session One: Growth. In session one, the group leader begins with group introductions and an ice-breaker activity. Then, the group leader provides a didactic introduction to growth and personal improvement in different areas—social, academic, and emotional—and their benefits. Participants then read an article and watch a short video that explains neuroplasticity: a phenomenon whereby the brain grows and makes new connections as a result of learning and practicing new skills. Both the article and video script are available in supplementary materials; the video is available upon request and online. Next, the participants read three growth testimonials from local Kenyan peers and a growth testimonial from their own group leader (see supplementary materials). Testimonials focus on personal growth in a plethora of domains (e.g. school performance, personality, happiness, relationships, etc.). Participants then have a discussion about the testimonials. To conclude the first session, participants are assigned a take-home activity in which they identify a challenge they faced, describe how they used effortful strategies to deal with the challenge, and reflect on how they grew as a result of the challenge.
Session Two: Growth. In session two, participants begin by discussing their take-home activity. They then brainstorm and discuss effective strategies that they can use to apply the lessons of growth-mindset in their own lives. Next, the group leader will moderate a discussion about problem-solving skills modeled off of the problem-solving STEPS module (33). Afterward, participants are instructed to write a letter to a friend in which they explain the concepts that they have learned thus far—including neuroplasticity, growth-mindset, effective strategies, and problem-solving skills—in their own words. The group leader is instructed to do all of the same activities themselves while the participants do them. Lastly, participants are assigned a take-home activity in which they think of one problem that currently affects them and brainstorm solutions modeled off the problem-solving skills learned in the session. They are encouraged to try one or more of these solutions during the week.
Session Three: Gratitude. In session three, students learn about gratitude. The group leader opens the session with a didactic introduction to gratitude and its various benefits. The participants then discuss gratitude and identify people and things for which they are grateful. After this, participants write a “gratitude letter” (34) to a living person who has changed their life for the better. For homework, participants are instructed to complete a daily “three good things” activity during the following week (35). In this activity, participants identify and write three good things that happened each day and write a short reflection for each.
Session Four: Virtues. In session four, students learn about virtues and their importance in everyday life. While the activities in this section are based on empirically-tested value affirmations exercises, the discussion of values will be presented to Kenyan students as “virtues” due to differences in language and vocabulary. The group leader opens the session with a didactic introduction to what virtues are. The group then enters a discussion about virtues, noting why they are important in guiding daily behavior and helping us find purpose in our lives. Students are then asked to select from a long list several virtues that are meaningful to them. Next, students select the one particular virtue that they feel is most important to them and describe, in writing, why this virtue is important, a time when they have lived up to that virtue, and how they can live in better accord with this virtue in the future (19). There is no take-home activity following this fourth and last session. At this point, participants are allowed to keep the Shamiri booklet, which contains all the lessons and exercises from the four sessions for both the Shamiri intervention and the control.
Using a study skills control group both presents participants with an opportunity to benefit from participation and provides a more rigorous standard of comparison than passive controls (36).
The control condition consists of four modules: a note-taking module lasting one session, an effective-study-strategies module lasting one session, a time management module lasting one session, and a study cycle module lasting one session. All these modules were developed specifically for a pilot trial of the Shamiri intervention (12). More information on these sessions is provided below:
Note-Taking: Developing effective note-taking skills may help students improve their academic performance. The session will cover what the study team will present as the 5Rs of note-taking: Record, Reduce, Recite, Reflect and Review (See online supplementary materials for more information).
Effective Study Strategies: Effective study strategies may help students understand what they are learning and how they can optimize their time spent studying. The session will cover 10 effective study strategies compiled by the study team, including “have a study plan,” “don’t cram all studying into one session,” and “start with the most difficult subject” (See online supplementary materials for more information).
Time Management: Time management skills may help people become more productive and improve academic performance. This session will cover four steps that may help people manage time more effectively: organizing time, assessing time, setting priorities, and scheduling time (See online supplementary materials for more information).
The Study Cycle: The Study Cycle is a 5-step approach to learning that may help students’ academic performances by reinforcing new content and making them feel more confident. This session will cover the 5 steps of the study cycle: preview, attend, review, study and assess (See online supplementary materials for more information).
Structure of the control sessions: To control for non-specific aspects of the intervention, the study skills control group (developed for the present study) will mirror the structure of the Shamiri intervention. Each condition contains the same number of exercises and group discussions, within-session activities are similar in format (e.g., reading, writing, and discussion activities), and between-session take-home assignments require similar effort.
As detailed, the study skills control consists of four sessions, which are each one-hour long and one week apart. Take-home assignments are also assigned between sessions.
Session One: Note-Taking. During session one, group leaders will offer a didactic introduction to the 5Rs of note-taking. Then, participants will complete an exercise to practice the new note-taking method: They will read a short article about climate change and take notes using the 5Rs strategy they just learned. For homework, participants will be asked to complete three tasks: They will 1) identify one particular class during which they used the 5Rs of note-taking, 2) describe how they used the 5Rs to take notes, and 3) reflect on whether their learning improved as a result of using the new note-taking strategy.
Session Two: Effective Study Strategies. To begin session two, participants will reflect on the homework in a group discussion (e.g. What was rewarding or challenging about using the 5Rs? Do you plan to use them in the future?). Then, group leaders will introduce ten effective study strategies. Participants will be asked to discuss these ten strategies, examples of when they have used them, and other strategies that have worked for them. For homework, participants will be asked to complete three tasks: They will 1) identify a specific academic challenge they face over the course of the following week, 2) employ one strategy from the ten discussed in this session to address the challenge, and 3) reflect on whether their learning improved as a result of using the new strategy.
Session Three: Time Management. To begin session three, participants will reflect on their homework assignment in a group discussion (e.g., Which strategies worked for you? Which ones did not? Why? What strategies might you use in the future?) Then, group leaders will introduce the concept of time management and lead a discussion about the four steps to improving it as a skill. For homework, participants will be asked to track and write down how they spend their time on activities like schoolwork, sleeping, and eating. This homework is intended to help participants understand how they are actually spending their time each day.
Session Four: The Study Cycle. To begin session four, participants will reflect on their homework assignment in a group discussion. Group leaders will ask if anybody is willing to share their time management worksheet and discuss with the group. Then, group leaders will introduce the five-step study cycle and discuss it with the students. Finally, group leaders will lead a discussion about how participants can continue practicing skills they have learned from all the sessions in the future. There is no take-home activity following this fourth and last session. At this point, participants are allowed to keep the Shamiri booklet, which contains all the lessons and exercises from the four sessions of the study-skills sessions.
We will assess intervention fidelity—including group leader adherence to protocol manual and group leader competence in delivering the Shamiri and study-skills interventions—using a fidelity rubric (see Table 3) developed by the study team. In this rubric, sessions are broken down into small 5–10-minute chunks that reflect the activities which are outlined in the protocol manual. Raters will listen to and rate these chunks. They will code group leader adherence to protocol manual (i.e. whether a group a leader followed manual and delivered required content classified as 0 = no, 1 = yes) and group leader competence (i.e. effectiveness in communicating concepts and skillfulness of delivery etc., rated from 1 = not-competent to 5 = very competent). Ten percent of the total sessions will be assessed for fidelity.
Raters (N = 2) will be recruited from psychology or social science departments at local universities in Nairobi. The raters will be blind to study purposes. They will first undergo a series of trainings led by the first author, including a didactic training on the protocol manuals of the Shamiri and study-skills groups, a training of the fidelity rubric, and training on the 5–10-minute chunks. The raters will then listen to a randomly selected sample from each of the conditions, and rate them independently. After this, the raters will work with a member of the study team and walk through their ratings. Next, the raters will grade a randomly selected 10% of the audio-recordings of the Shamiri and study-skills sessions.
Primary outcome measures.
Patient Health Questionnaire–8.
We will use the 8-item version of the Patient Health Questionnaire–9, the Patient Health Questionnaire–8, to assess youth depression symptoms. PHQ–8 scores are highly correlated with PHQ–9 scores, and the same cutoffs can be used to assess depression severity (23). The PHQ–8 has previously demonstrated adequate internal consistency in Kenyan adolescents use in Kenyan adolescents (22).
Generalized Anxiety Disorder Screener–7.
The Generalized Anxiety Disorder Screener–7 (GAD–7) (37) is used globally to screen for generalized anxiety disorder in adolescents and adults. The GAD–7 has previously demonstrated adequate internal consistency in Kenyan adolescents (22).
Secondary outcome measures.
Average Term Grade. Academic grades of the participants will be collected for three school-terms: the school-term before the intervention (January to April 2019), the school-term during the intervention (May to August 2019), and the school-term after the intervention (September to November 2019). To determine students’ average term grade, we will calculate a mean term grade from their grades in all the subjects in which they will be enrolled. Although subject enrollment varies across different schools, students are typically expected to enroll in between 6 and 12 subjects. In order to compare students’ grades across different grade levels, academic subjects, and schools, we will convert the academic grades to standard scores (M = 60, SD = 10, chosen arbitrarily and used in rescaling).
Science and Math Score. We will calculate a science and math score for each of the three semesters for which academic grades will be collected. The score will be calculated as the mean grade of the math, biology, chemistry and physics grades. As with the average term grade, we will convert these grades to standard scores with M = 60, SD = 10 used in rescaling.
Humanities and Languages Score. We will calculate a humanities and languages score for each of the three semesters for which academic grades will be collected. The score will be calculated as the mean grade of the English, Kiswahili, history, geography and religious studies grades. As with the average term grade, we will convert these grades to standard scores with M = 60, SD = 10 used in rescaling.
Self-reported perceived academic performance. We will ask students to self-report on how they perceive their academic performance. This will be a one-item instrument that will ask them to self-report their perceived academic performance on a 5-point Likert scale from not satisfactory to excellent. The item has been used in a previous study with Kenyan university students (38).
Kenya Certificate of Primary Education (KCPE) scores. The KCPE is a national examination that all students take at the end of eight years of primary school and determine admission to secondary school (21,22). KCPE scores will be collected for all participants, not as an outcome measure, but as potential moderator of changes in academic performance outcomes.
Multidimensional Scale of Perceived Social Support.
The Multidimensional Scale of Perceived Social Support, MSPSS (39) is designed to measure satisfaction with social support. It consists of three subscales: the “friends” subscale (which measures support from friends), the “family” subscale (which measures support from family), and the significant others subscale (which measures support from significant others). The MSPSS has previously demonstrated adequate internal consistency in Kenyan adolescents (22).
Perceived Control Scale for Children.
The Perceived Control Scale,PCS, (40) includes 24 items related to beliefs about personal control, or the belief that, through effort, one can obtain desired outcomes and avoid undesired outcomes. Example items include “I can make friends with other kids if I really try” and “I cannot stay out of trouble no matter how hard I try.” The PCS has previously demonstrated adequate internal consistency among Kenyan adolescents (22).We will use the social control and academic control subscales as secondary outcome measures.
The Gratitude Questionnaire.
The Gratitude Questionnaire (GQ–6) is a six-item questionnaire that measures grateful disposition—the tendency to experience and act with gratitude in everyday life. The GQ–6 has been shown to have strong psychometric properties in Kenyan youths (41). A previous attempt to use the GQ–6 with Kenyan youths demonstrated weak internal consistency (22).
The EPOCH Measure of Adolescent Well-Being.
The EPOCH Measure of Adolescent Well-Being assesses positive psychological characteristics that might foster well-being, physical health, and other positive outcomes in adulthood. While the EPOCH has been validated with adolescents in Australia and the United States (42), it has yet to be used with adolescents in Kenya and other Sub-Saharan African countries to the best of our knowledge. We will use the Optimism and Happiness subscales of the EPOCH.
UCLA Loneliness Scale—Shortened Version.
The UCLA Loneliness Scale is a widely validated tool for measuring trait loneliness. Participants rate their responses to statements such as “I feel isolated from others” and “people are around me but not with me" on a 4-point Likert scale. The third and newest edition of the UCLA Loneliness Scale has demonstrated high reliability, construct validity, and discriminant validity (43). We will use a shortened version of the scale (8-item) of the UCLA Loneliness Scale, which shows similar validity and reliability to the standard 20-item scale (43). To the best of our knowledge it has yet to be used with adolescents in Kenya.
Program feedback scale.
A feedback scale will be used to collect acceptability and feasibility data from participants. The youths will be asked questions about whether they found the program useful, whether they understood the contents of the program, whether they would recommend the program to a friend, which were the most and least helpful elements of the program, and how they would rate their overall experience. They will also have the option to recommend changes to improve the program for future participants. This scale was developed specifically for this study; however, several items were drawn from prior research (22,44).The scale is available in Table 2.
Timeline for study and data collection.
Participant recruitment will begin in mid-to-late June 2019. The intervention will begin in mid-to-late June 2019 and end in late-July 2019. Data collection will end in December 2019, once final follow-up measures are collected. We intend to report results by Spring 2020. Upon termination of data collection and publication of results, we will make study data publicly accessible. We cannot provide identifiable data about participants and schools.
Data Analysis Plan
Sample size and statistical power.
Optimal Design (45), a statistical simulation program for estimating power and sample size for multi-level models, was used to estimate sample size. Based on a pilot study of the Shamiri intervention (22), which revealed an effect size of d =.32 for depression and an effect size of d =.54 for anxiety, we calculated sample size necessary to detect an effect of d =.3 using an alpha value of p = 0.05 and a power of.80, without consideration of covariates. Using these parameters, 200 participants are required per group for a total of at least 400 participants. Assuming that 30% of willing students at recruited schools will be eligible to participate in the study—based on previous assessment wherein 30% of youths in 5 participating secondary schools in Kenya showed elevated symptoms of depression and anxiety (22), and projecting that up to 10% of students may discontinue from beginning to midway through the intervention—based on a conservative estimate from the drop-out rates of Osborn et al. (12), at least 1500 youths are needed for baseline screening to realize the minimum sample size. Given the participation of 4 Kenyan secondary schools in the present study with 3200 students enrolled, this sample size is believed to be feasible (22).
Analysis of primary and secondary study outcomes.
Outcome measures will need to demonstrate internal consistency with Kenyan youths as indexed by a Cronbach’s alpha of at least 0.70 (46) to be considered for further analyses.
Our analyses will follow an intent-to-treat approach, including all participants initially randomized to an intervention condition. We will run a linear mixed model comparing intervention and control groups on each primary and secondary outcome. Models will be organized to reflect the hierarchical structure of the data. Different assessment points will be nested within participants. All models will include a random intercept that will allow for individual variation at baseline. We will include a random slope that will allow for individual variation in outcome change rates; however, should this result in overfitting (i.e. random effects structure becoming too complex to be supported by the data), we will remove the random slope to allow for more parsimonious models. Time, intervention condition, and their interaction will be included in all models. Covariates will include age (in years), sex. Age will be included because older adolescents are reported to face increased psychosocial stress, which may exacerbate depressive and anxiety symptoms (22,47). Sex will be included because gender differences in internalizing problems have been documented in Kenyan adolescents (21,22,48). We will build two models, one which will include schools a random slope thus allowing for variation in outcome change rates by school, and another with school as a covariate because students in resource-poor schools in Kenya have reported higher anxiety symptoms than students in schools with more resources (22). We will use Akaike Information Criterion (AIC) to determine which of the two models will be a better fit for the data. Significant (p <.05) condition * time interactions in predicted directions will indicate that the intervention condition produced more rapid improvements in outcomes across the full study period (baseline to 4-week endpoint), as compared to the control.
We will calculate effect sizes (ESs) for total pre- to post-intervention change in depressive symptoms (the primary study outcome) using differences in means; these ESs will compare mean gain scores (Cohen’s d) reflecting changes in each outcome from baseline to post-treatment for youths in the Shamiri intervention versus study skills control intervention. Statistically significant, positive Cohen’s d values will indicate greater improvements for intervention group youths versus control group youths.
Participants who miss one or more intervention sessions will be permitted to attend future sessions. Missing item (measurement) and subject-level data will be imputed twenty times using Fully Conditional Specification (FCS), implemented using the multivariate imputation by chained equations (mice) algorithm in R as described by Buuren & Groothuis-Oudshoorn (49).
Study methods and protocols have been approved by the Maseno University Ethics Review Committee (MUERC) reference number MSU/DRPI/MUERC/00727/19.