We employ a matched-pair cRCT design and mixed methods evaluation data collection. For reporting, we follow the Consolidated Standards of Reporting Trials (CONSORT) for cRCT designs  and the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines . We conduct a Hybrid Type II effectiveness-implementation trial  in two regions in Uganda. The school is the unit of randomization because the program is applied at the school-level and builds a “school community” of teachers to promote student mental health. The three-arm cRCT design allows us to simultaneously test PD effectiveness and study the added value of the T-Wellness to address teacher stress, which is a critical challenge to practical task-shifting effectiveness and sustainment. In addition, the Hybrid design, which considers CFIR domains of implementation contexts (listed in Figure 1), allows us to rigorously study other EBI effectiveness-implementation mechanisms, which can further inform decisions about optimal deployment and the generalizability of impact, and may accelerate the introduction of other valuable innovations into practice . To have a more comprehensive understanding of possible underlying implementation and effectiveness mechanisms, we also plan to conduct qualitative interviews and focus groups, which will purposefully select PD/PDT trainers, teachers, and parents. The study has been approved by the Institutional Review Boards of New York University Grossman School of Medicine (i20-00117), Makerere University (REC REF 2020-143), and Uganda National Science and Technology (HS1057ES).
School recruitment and randomization. Primary schools in targeted Kibuli (urban) and Hoima (rural) districts will be identified from governmental school lists. These districts were selected based on MOE leaders’ assessment of high need. To ensure approximately similar school characteristics in three conditions across two geographic regions, a stratified-block randomization procedure will be applied (Figure 2) . A statistician who is unfamiliar with study schools will first match the schools on school size (teacher/student numbers) and school quality/performance (based on MOE data) within regions to ensure similar characteristics in urban and rural regions. Eighteen schools in 6 matched blocks (with blocks of size 3) from each region will be selected. Principals will be invited to attend information sessions hosted by the Ugandan study team During recruitment sessions, principals will be provided with details of study requirements and intervention implementation procedures. They will have an opportunity to ask questions and also complete a questionnaire on school demographics, commitment, willingness to facilitate data collection, and ability to allocate staff time to participate in the study. School principals who express interest and agree to allow teachers’ voluntary participation will be eligible and will be consented. A total of 36 schools will be included for the effectiveness study. Computer-generated random numbers will be used to decide the randomization allocation sequence. Within each block (of size 3), one school will be randomly assigned to PDT, one to PD, and one to Control (receiving child mental health materials approved by MOH/MOE). The intervention and evaluation activities will be carried out in two consecutive cohorts using the two-cohort approach because it allows for building the capacity of TTC trainers to carry out PD in a real-world context and provides time for TTC trainers to develop practice competency. Cohort 1 schools (n=18; 9 urban and 9 rural) will start in 2021, and Cohort 2 schools (n=18) in 2022. Schools from both cohorts will be actively involved for 2 years.
Teacher recruitment. All pre-primary and primary grade 1 to 4 teachers and teaching assistants, serving students between the ages of 3 and 10 years, will be eligible to participate. We include multiple grades because teachers in Uganda teach a wider age range of classes. Teachers’ participation will be completely voluntary, with no consequence for opting out. Based on the 100% enrollment record from our prior cRCT, we anticipate that nearly all eligible teachers will sign up for the study. We anticipate that 540 teachers from 36 schools will participate in the evaluation; and 360 intervention school teachers will receive PD (180 with and 180 without T-Wellness). Based on enrollment from our prior study in Uganda, we anticipate that nearly all teachers will sign up for PD (i.e., > 90% will participate in PD). Based on feedback from stakeholders, we will also recruit two parents from the Parent-Teacher-Associations (PTAs) in intervention schools to be part of the school-based team to support teachers to facilitate parent involvement (e.g., through sharing evidence-based knowledge and parenting strategies with families during parent-teacher conferences).
Parents and child recruitment. Students attending pre-primary and primary grades 1 to 4 (ages 3-10 years) and their parents/primary caregivers will be eligible to participate in the study. Given the large numbers of students in schools, research staff will randomly select 10% of students and families from each school (based on student lists provided by schools) and complete assessments over two years [50, 51]. A total of 1,980 families from 36 schools (averaging 55 families/per school) will participate in the study. Teachers will be informed of the students randomly selected for participation in the assessments and asked to introduce the study staff to the selected families. Primary caregivers from the selected families will be invited to consent for interviews, and for research staff to carry out assessments with their child. Children with parent consent will be asked to assent to the study. Although we only evaluate a subset of the sample, an estimated 13,200 students across 24 intervention schools will be exposed to PD. Teachers and parents who participate in study, will receive a small incentive for their time.
Local PD implementers/trainers and MHP supervisors. A total of 8 Ugandan TTC trainers (4 from each TTC) and 4 MHP supervisors will be recruited and trained to implement and support PD. TTC trainers will be required to have a minimum of university level of education and two or more years of experience in teacher training. MHP supervisors will be clinical psychologists, or mental health counselors with at least master’s degrees and 2 years of experience, or psychiatric nurses (with at least bachelor’s degree and 5 years of clinical experience). They will be recruited from local universities or mental health facilities. TTC trainers and MHPs who agree to participate will be asked to provide written informed consent, which will allow the research team to gather fidelity and competency data (self-reported, audio, or observational data) with their assistance.
Subsample for the Qualitative Study. Subsamples of study participants from the intervention schools will be selected to participate in qualitative interviews or focus groups aimed at better understanding the underlying mechanisms for the effective implementation and sustainment. For each study cohort we will carry out interviews and focus groups with PD/PDT trainers (n=8), teachers and parents (n=40; 20 from PD and 20 from PDT schools across urban and rural sites). Qualitative data will be conducted twice (post PD/PDT intervention and a year after the intervention).
Sample Size and Power
We conducted power analyses for child and teacher effectiveness outcomes, assuming an intention-to-treat (ITT) analysis. The power calculation is estimated primarily based on: i) the statistical analysis approach planned for this study (linear mixed effect models ); ii) the expected magnitude of the effects for the primary child and teacher outcomes from our prior Ugandan PD study (i.e., child mental health d=.39; child social emotion competency d=1.08; teacher practice outcomes-observed d=.55 and self-report d=.32); and iii) detectable effects with 80% power of two-sided significance tests with a=.05. In cRCT designs, the detectable effect sizes depend on the usual study design parameters, as well as the cluster size, N, and the cluster effect, i.e., the intra-cluster correlation coefficient (ICC). Detectable effects also depend on the test used (e.g., a test that accounts for baseline outcome or examines effectiveness with repeated observations is more powerful) . We anticipate that 540 teachers (or 180 per intervention condition) and 1,980 parent-child pairs (or 660 per intervention condition) from 36 schools will participate in the study. We estimate power (for detectable effect size) based on the total sample (1,320 families and 360 teachers for two comparison conditions [1 intervention and 1 control]), as well as based on the sample from one region (660 families and 180 teachers for two conditions) with one or two post-intervention outcome evaluations, and assuming 20% loss of sample by Time 3. Table S2 in the Supplemental file gives the detectable effects for a range of cluster/ICC values and multiple scenarios. In the most conservative scenarios when the ICC=0.05, for 24 school clusters (2-condition comparisons across two regions), the detectable effects are d=.14-.22 for child outcomes, d=.18-.38 for teacher intermediate outcomes; and for 12 school clusters (2-condition comparisons in one region), the detectable effects are d=.19-.42 for child outcomes, d=.25-.55 for teacher intermediate outcomes. This study will have sufficient power to detect impacts that are meaningful and realistically achievable.
Description of Intervention and Scalable Implementation Approach
The PD Implementation Approach. The approach to ParentCorps PD implementation relies on Train-the-Trainer (TTT) and a dynamic multi-layered supervision model . As shown in Figure 3, an experienced clinical team from the ParentCorps Central Office in the US provides comprehensive virtual training and ongoing supervision to four Ugandan MHPs, including two who were previously trained and participated in the previous ParentCorps PD study. The four-person Ugandan clinical team will oversee local implementation efforts including supporting and supervising the TTC Ugandan PD trainers/facilitators. The ParentCorps clinical team from the Central Office in the US will provide virtual training to 8 TTC Ugandan PD facilitators (32 hours over 8 half days) and the Ugandan clinical team will provide live ongoing supervision. Over a two month period (~8 weekly meetings) working with the Ugandan clinical team, the TTC Ugandan PD facilitators will practice, receive feedback and refine aspects of PD for the local context prior to the first round of implementation. Finally, with ongoing supervision from the Ugandan clinical teams (8 live group supervision sessions), the Ugandan TTC facilitators will provide ParentCorps PD (21 hours over 3 days) to the Ugandan primary school teachers and 8 one-hour of coaching session over a 2-3 month period.
Teacher-Wellness Implementation Approach. T-Wellness was co-developed by US and Ugandan investigators, adapted from evidence-supported strategies for teacher stress and burnout management [54, 55]. T-Wellness is a one-day workshop and three group-support sessions (45-60 minutes each) for teachers. MHPs from the Ugandan clinical team will be trained by study investigators to facilitate the workshop and the support sessions to teachers. Teachers from schools assigned to the PDT condition will receive the 1-day workshop right after receiving PD. The three group-support sessions will be integrated in the PD coaching sessions. For implementation quality assurance, the Ugandan clinical team will receive supervision from the study investigators after each group-support session they provide to teachers.
PD. The 3 days of ParentCorps PD aims to help teachers to foster child social emotional learning, strong family-school relationships and safe, nurturing, and predictable classroom environments. There are four elements that the ParentCorps theory of action specifies as essential processes through which PD strengthens teachers’ use of evidence-based practices: building authentic relationships, honoring culture, translating the science of early childhood development, and practicing self-reflection. These essential elements are measurable aspects of the quality of facilitation that complement measures of fidelity to the manuals in explaining the extent to which the program targets change. Specifically, high quality, high fidelity facilitation is hypothesized to support teachers in developing increased capacity as defined by (1) knowledge of evidence-based strategies; (2) awareness of self and child; (3) intentional and responsive interactions; and (4) problem-solving and support-seeking as needed.
(see Table S1 in the Supplemental file for additional contents information).
PDT. The 1-day workshop and 3 group support sessions aim to increase teachers’ self-awareness of their stress and regulation/coping styles, and support teachers to manage stress through practicing evidence-based strategies. Prior to the workshop, teachers will be asked to complete a stress and wellness self-assessment survey using a digital toolkit. A tailored report is generated right after the assessment to share with teachers to promote self-awareness and motivation for change. During the one-day workshop, four key topic areas will be covered: (i) understanding stress and job burden; (ii) self-appraisal and identification of areas for improvement; (iii) cognitive and behavioral strategies; and (iv) teacher-to-teacher support and other additional resources. The group support sessions are to help teachers apply strategies to work toward their wellness goals.
Control Group. Teachers in control schools will receive mental health knowledge and promotion materials. In the second year of participation (after completion of the effectiveness evaluation), control schools will receive a one-day T-Wellness workshop (without PD) and 3 follow-up group-support sessions.
The evaluation design is guided by the implementation outcome framework . The quantitative evaluation measures for teacher and child effectiveness outcomes will be assessed using multiple sources (data collected from objective classroom observation, parents, teachers, and children) and across 3 time points (T1 baseline, before PD; T2 immediately after the PD/PTD, about 3-4 months after T1; and T3, 9-12 months after T2). Research staff responsible for family and observation data collection will be masked to intervention conditions. To ensure masking, we will have an independent assessment team (led by a separate research coordinator), and members will not participate in any intervention activities. We will also train the implementation team on protocol to prevent unblinding. Table 1 lists the measures included in the study. Most of the measures have been used and validated in our previous pilot trial .
Child effectiveness outcomes. The primary outcome is child mental health (externalizing and internalizing problems), and the secondary outcome is social-emotional competency (emotion regulation, executive functioning). Parent and child report data will be gathered. Parents of study students will be interviewed by trained research staff (using English or Luganda, lasting 30-45 minutes). Participating children will be assessed by trained research staff in schools (lasting about 20-25 minutes).
Teacher effectiveness outcomes. The primary teacher outcomes are (i) EBI strategy use, which will be based on objective observation by an independent observation team (primary data source) and teacher report (secondary data source), and (ii) teacher perceived stress and stress management (teacher report). The secondary teacher outcome will be school-home connection, teacher-family relationship, and student-teacher relationship based on parent and student report.
Implementation Outcome Measures
Fidelity will be measured to assess the quality of implementation. Four fidelity measures will be considered, including adherence (the extent to which the TTC facilitators deliver the core intervention content and as per program guidelines), quality of program implementation (assessed based on teacher rating of their training expeirence of Coaches’ competence (knowledge, preparation, ability to control discussion, enjoyable); engagement (assess trainees’ level of PD knowledge improvement from pre- to post-training; and exposure (measured by trainees’ attendance in PD and coaching sessions) . Acceptability and Appropriateness of PD/PDT will also be measured based on teacher report and be assessed post-training and at T2 (after completing the full cycle of PD/PDT) .
Cost Measures. Costs will be measured using an activity-based micro-costing approach  in the intervention and control clusters (school), and in the extended implementation phase again in all clusters from 6 through 12 months (n=36 clusters). Micro-costing entails a three-step approach where we identify, measure, and value resource use for all activities in each study arm. Resource use and cost data will be collected prospectively alongside the trial. All research costs will be excluded. Cost data collection will utilize standardized cost extraction forms and procedures that have been validated in our team’s previous work in Uganda and other LMICs [59-64]. Prior to use, these tools will be tailored and customized to the PD/PDT context. All costs will be adjusted for inflation, discounted to the intervention start year, and presented in US dollars.
Constructs in CFIR domains will be measured to study moderation effect on teacher EBI strategy use outcomes. Selection of CFIR moderators is guided by factors identified in the literature as influential factors for implementation and effectiveness outcomes [46, 65]. Inner setting will include school structural and climate characteristics (classroom size, learning climate, leadership engagement, teamwork alliance); outer setting will include region (urban/rural); process will include partnership quality, fidelity, and cohort (1st or 2nd implementation cohort); intervention characteristics will include perceived PD/PDT acceptability and appropriateness; and Individual teacher characteristics will include teacher years of experience and gender.
To have a more comprehensive understanding of possible mechanisms, we will conduct qualitative interviews and focus groups. Interview guides will comprise semi-structured questions relating to experiences with PD/PDT and sustainment of PD/PDT. Participants will be asked to provide a narrative account of partnership approaches and efforts to implement and sustain PD/PDT, including barriers and facilitators experienced. Qualitative assessment will be conducted twice at T2 and T3. We will also conduct qualitative assessment separately for each cohort, which allows a better understanding of cohort effects and whether quality of implementation improves over time.
All data will be managed and stored in REDCap (Research Electronic Data Capture). REDCap is a secure web application for building and managing surveys and databases for research studies, originally developed at Vanderbilt (www.project-redcap.org) with collaboration from a consortium of worldwide institutional partners. It provides automated export procedures for seamless data downloads to common statistical packages such as Excel, SPSS, SAS, Stata, and R. Access to study data in REDCap will be restricted to the specific members of the study team with authentication. Qualtrics will also be used when collecting data in the field through the offline mobile app function. When using Qualtrics offline mobile app, no identifying information will be collected. Qualtrics mobile app uses Transport Layer Security (TLS) encryption (also known as HTTPS), and data entered into the mobile app cannot be re-accessed in the front-end. Only selected staff members with have access to the data in the back-end through password protected accounts. Data will be entered using only the unique study identification number. Qualtrics data will then be transfer backed into REDCap as our database management system. All final study files for analyses will be captured and finalized ensuring that no personal identifiable information (PII), including students’, parents’ or teachers’ names and contact information, are included. Electronic data entered that include contact identifying information (e.g., master list of consenting information, contact information/address) will be securely saved, and will not be linked to the study data. There will be additional levels of protection and access restrictions to this information.
Data Analyses and Statistical Methods
Preliminary Analyses. Prior to any outcome analyses, we will generate summary statistics for all data, summarizing with means and standard deviations for continuous variables and frequencies for categorical variables. Baseline equivalence between intervention and control schools will be examined. For measures that evaluate similar constructs, composite scales will be created (to minimize number of analyses). In addition, the distribution of study variables and missing data patterns will be inspected. For participants with partially missing data, a multiple imputation strategy using a Markov chain Monte Carlo approach will be applied. We will also sequentially impute data for each wave using predictive mean matching method separately for intervention and control groups to account for the possibility of different missing data patterns by condition [83-85]. Ten data sets will be imputed, and SAS PROC MIANALYZE will be used to combine the results for the final inference .
Analyses for Aim 1. To estimate effectiveness, we will apply intention-to-treat (ITT) analyses and first focus on between-subject comparisons of intervention vs. control (comparing PD to control, comparing PDT to control, and comparing PD to PDT). We will estimate the impact of PD on children and teachers post-intervention (T2, 4-5 months after T1) and at one-year follow up (T3, 12 months after T1). School and class nesting will be considered, and a multiple imputation strategy will be applied to account for missing data. Linear mixed effect models , using SAS PROC MIXED, will be applied to examine short and longer-term impacts. We will first examine the immediate impact by modeling post-intervention outcomes (T2) as a function of intervention, adjusting for T1 outcome measures. The model accounts for correlation between subjects (within-school and -class) by including school- and classroom-level (when appropriate) random intercepts. Next, we will study longer-term effectiveness outcomes (T2 to T3) by applying growth curve models and using repeated assessments over time. In these growth models, we will add time-relevant parameters to the model above, including school-level random slopes associated with time. The post-baseline scores will be modeled as a linear function of time, intervention indicator, and intervention-by-time interactions, adjusting for T1 scores and cohort.
Cost-effectiveness analysis of PD and PDT implementation models will be examined using approaches that have been applied in previous school-based and parenting-based child mental health promotion research [87-90]. The analysis will center on incremental cost-effectiveness ratios (ICERs), where the numerator represents the cost difference between the intervention arms and the control, and the denominator represents the difference in average intervention effects. To that end, the cost-effectiveness analysis of the intervention will involve examining how much the PD/PDT intervention costs to achieve a unit of effect relative to the control group. The effects of the intervention will be estimated using the effect sizes d (standardized mean difference between intervention and control groups) using an ITT approach. For the effectiveness outcomes, we will use an effect size of 0.2-0.4 as a benchmark; this corresponds to a small to medium effect size according to Cohen . We will compute the per-participant cost per 0.2-0.4 SD change for each child effectiveness outcome. Reporting of the cost-effectiveness analysis will follow the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) .
Analyses for Aim 2.We will examine mediation mechanisms for PD and PDT separately. The analysis will be built on the Aim 1 Linear mixed effect models. For PD, we will examine whether impacts of PD on children’s mental health are mediated through improvement in teachers’ EBI practices (primary). For PDT, we will examine whether impacts of PDT on child mental health are mediated through improvement on teachers’ EBI practices and stress management (in cognitive and behavioral domains). The intermediate teacher outcomes will be based on T1 and T2 data, and the child outcomes will be based on T2 and T3 data (to capture changes over time).
Analyses for Aim 3. To study effectiveness-implementation moderation mechanisms, we will test whether impacts of the intervention on teacher effectiveness and EBI practice sustainment outcomes is moderated by CFIR contextual factors. We will apply similar approaches as in Aim 1 and add the moderator and moderator-by-intervention interaction terms in the analysis. T2 implementation and T2 and T3 teacher effectiveness outcome data from the PD/PDT intervention samples will be utilized. Any significant moderators identified will suggest important factors to be intervened on in future implementation to enhance the uptake of evidence-based strategies by teachers, or improve the effectiveness of task-shifting.
Qualitative data analyses. For the qualitative focus group data, we will apply qualitative analysis methods. Interview data will be transcribed and analyzed using Atlas.ti software. To better understand partnership/scalable approaches, coding will focus on themes related to the partnership development process, usefulness of partnership frameworks in formalizing processes, scalable strategies, intervention implementation barriers, teacher stress, and strategies for overcoming teacher stress and other practices (considering CFIR). These analyses will help identify facilitators and barriers for partnership and implementation for carrying out the effectiveness study. For effectiveness-implementation mechanisms, qualitative analysis will focus on themes related to implementation barriers, facilitators, and contextual factors and processes that influence teacher intermediate and child effectiveness outcomes. Coding of qualitative data will follow a constant comparative analysis approach, where data are analyzed for themes that reflect project aims, which are then confirmed by further data analysis, followed by a third review of the data to identify additional themes [93-95].