Fidelity of Mindfulness Training and Delivery for Adolescent Mental and Metabolic Health

Abstract

Objectives

Adolescent-onset type 2 diabetes (T2D) is increasing, and depression symptoms amplify T2D risk, highlighting the need for interventions addressing mental and metabolic health. This multisite pilot RCT assessed whether facilitators could be trained consistently to deliver group MBI, cognitive behavioral therapy (CBT), and a matched health education condition (HE) to adolescents at risk for T2D with elevated depression symptoms. It was hypothesized that facilitators would achieve ≥ 80% on knowledge, competence, and adherence following standardized training.

Method

Facilitators at four U.S. sites received training in MBI, CBT, and/or HE through didactics and supervised virtual mock sessions. Training fidelity was assessed through pre/post-training knowledge tests and expert ratings of adherence/competence in mock sessions. Intervention delivery fidelity was assessed with expert ratings of virtual pilot group sessions with adolescents. Contamination across conditions was evaluated.

Results

Seventeen facilitators were trained in MBI (n = 11), CBT (n = 10), and/or HE (n = 9). Descriptive feasibility metrics are presented; between-arm comparisons are not made. Knowledge scores improved across conditions (MBI: 76% → 96%, p = 0.01; CBT: 68% → 92%, p = 0.01; HE: 84% → 96%, p = 0.02). Median competence/adherence met or exceeded the 80% threshold for mock (MBI = 80%/85%, CBT = 86%/90%, HE = 82.5%/80%) and pilot (MBI = 90%/100%, CBT = 90%/90%, HE = 97.5%/96%) sessions. Contamination was limited to overlapping MBI-CBT content.

Conclusions

Results support the feasibility of high-fidelity training and implementation of MBI, CBT, and HE across diverse settings, informing future research into integrated adolescent health prevention strategies.

Preregistration

ClinicalTrials.gov #NCT04992299.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s12671-026-02975-8.

Metabolic health concerns are among the most serious, preventable public health issues facing adolescents in the United States (Nadeau et al., 2025; Suglia et al., 2016). In particular, rates of adolescent-onset type 2 diabetes (T2D) are increasing, especially among youth from racial and ethnic backgrounds most affected by social determinants of health (Lawrence et al., 2021). Adolescent-onset T2D is often characterized by accelerated disease progression and high prevalence of co-occurring health complications. After accounting for the effect of obesity on heightened insulin resistance, youth mental health concerns, particularly depression symptoms, also relate to worsening insulin resistance and greater risk of T2D onset (e.g., Shomaker et al., 2011), highlighting a worrying intersection between adolescent mental-metabolic health vulnerabilities (He et al., 2025). The compounding effects of limited treatment options for adolescents with T2D (Zeitler et al., 2012) and historically high rates of adolescent depression in the United States (Brody & Hughes, 2025) has led to an unprecedented need for interventions with high-quality facilitation. It is crucial for interventions simultaneously address adolescent depression symptoms and T2D to prioritize prevention (Kitabchi et al., 2005; Perng et al., 2023; Shomaker et al., 2011).

Mindfulness-based intervention (MBI) may offer value as a preventative approach for adolescents with co-occurring mental-metabolic health concerns. MBI involves training in the cultivation of present-moment, purposeful attention to one’s body, thoughts, and emotions with an attitude of nonjudgment and a sense of equanimity. MBI has demonstrated efficacy for improving adolescent mental health by enhancing emotion regulation and reducing depression and anxiety (Broderick, 2005; Dunning et al., 2019; Leyland et al., 2019). The mindfulness stress buffering hypothesis is that individuals with high stress are the most likely to benefit from MBI because stress affects the pathogenic processes that contribute to diseases like T2D (Creswell & Lindsay, 2014). As such, MBI could offer a strategy to mitigate rising and co-occurring depression-insulin resistance and, consequently, T2D, which is most pronounced among youth exposed to chronic stress (Pinhas-Hamiel & Hamiel, 2020). When timed in adolescence, a sensitive developmental window for intervening to alter mental-metabolic health, MBI is theorized to improve top-down developing cognitive processes (e.g., executive functioning), bottom-up physiological systems involved in stress regulation and immune function (Jawahar et al., 2025; Zhang et al., 2023), and health-related behaviors such as physical inactivity (Schneider et al., 2019).

Despite promising initial empirical findings on the possible value of MBI for adolescent mental-metabolic health and stress-related physiological/behavioral mechanisms, MBI clinical trials remain nascent, especially in adolescent samples, and outcomes show considerable heterogeneity (e.g., Fulambarkar et al., 2023; Piguet et al., 2025). The proliferation of research involving MBI in recent years underscores the critical need for rigorously designed comparative efficacy trials (Crane & Hecht, 2018), and the availability of trials evaluating the process of training facilitators and delivering MBI for adolescents is minimal (Raudenbush & Bloom, 2015; Weinberger et al., 2001). Fidelity of intervention training and delivery implementation comprise central components of translating the theoretical and scientific promise of MBI into rigorous tests of efficacy.

In contrast to traditional CBT interventions involving structured techniques such as teaching cognitive restructuring and behavioral activation to identify and modify distorted cognitions and maladaptive behaviors, MBI is rooted in a guided discovery learning framework that offers opportunities for experiential practices that deepen participant engagement with core mindfulness concepts and teachings. This stance requires distinct facilitator training. Suboptimal delivery of MBI has been linked to diminished outcomes (Baer et al., 2019; Broderick & Schussler, 2021; Dimidjian & Segal, 2015); thus, efficacy may hinge on adherence to manualized protocols as well as facilitator competence. For mental health interventions to be effective, facilitators must be able to adhere to the protocol in terms of content delivery as well as interactions with participants (McAllister et al., 2018). Additionally, training proves to be particularly important to ensure clear responsibilities between the main and co-facilitator as well as preparedness to administer a group-based program (Higgins et al., 2018). As researchers seek to test MBI relative to active comparators such as CBT, the need to thoroughly demonstrate fidelity in both training and delivery becomes even more critical.

Rigor and reproducibility are strengthened by multisite trials that contribute geographic generalizability and are valued for their ability to enhance both external validity and statistical power (Raudenbush & Bloom, 2015; Weinberger et al., 2001) enabling assessment of training and fidelity between facilitators with different backgrounds. Training procedures play a pivotal role in enhancing fidelity, serving as a primary mechanism for enhancing facilitator knowledge, competence, and adherence to intervention protocols. Comprehensive training supports facilitators in developing a strong understanding of CBT or mindfulness principles, alongside practical skills for delivering curriculum-based content with integrity. Adherence—the extent to which facilitators follow prescribed manuals and session structures—is a key indicator of fidelity and is typically considered “high” when facilitators adhere to intervention protocols with 80% or greater accuracy (Gould et al., 2016). Likewise, facilitator competence—the skill with which practices are delivered—is often viewed as essential to intervention success, although some evidence suggests its relative importance may diminish when manual adherence is strong (Huijbers et al., 2017). Structured training models are important to promote both knowledge acquisition and delivery consistency to support multisite trials comparing MBI to other evidence-supported interventions.

Contamination between conditions is a well-documented threat to internal validity in randomized controlled trials, particularly when intervention arms share theoretical foundations or delivery agents (Perepletchikova et al., 2007). CBT and MBI both target cognitive and emotional regulation through overlapping mechanisms such as attention control, distress tolerance, and self-monitoring, increasing the risk of cross-condition exposure (Dimidjian & Hollon, 2010). In contrast, time/attention-matched controls are less likely to introduce therapeutic overlap.

Although factors such as therapeutic alliance and participant engagement demonstrate stronger links to outcomes than adherence or competence of MBI delivery (Horvath et al., 2011; Parsons et al., 2017), protocol adherence and facilitator competence remain widely recognized as indicators of high-quality implementation (Barber et al., 2006; Webb et al., 2010). Accordingly, in this multisite pilot and feasibility trial, these constructs are assessed to ensure treatment fidelity and prevent contamination across multiple facilitators and multiple sites. Establishing that facilitators can be reliably trained and can deliver interventions with consistency is essential before scaling to a fully powered efficacy trial, where variability in implementation could obscure true treatment effects. Demonstrating fidelity at this stage therefore provides an important foundation for increasing confidence that subsequent trial outcomes can be interpreted as effects of the intervention itself rather than site-level or facilitator-level differences. Outcome associations will be examined in a subsequent efficacy trial. The primary aim of this multisite, pilot and feasibility randomized controlled trial was to evaluate multisite fidelity of training and delivery of 6-week group MBI, 6-week group CBT, and 6-week group health education (HE) didactic arms for adolescents at risk for T2D with depression symptoms. The overarching hypothesis was that facilitators could be trained consistently across multiple study sites to achieve adherence, competence, and delivery fidelity in the respective intervention protocols. The training fidelity endpoint was defined as facilitators achieving a minimum of 80% on post-training knowledge tests of MBI, CBT, and HE. Training fidelity also was assessed through expert ratings of facilitator adherence/competence of mock and pilot group sessions. It was expected that mean ratings of facilitators’ mock delivery of all three intervention arms (MBI, CBT, and HE) would meet or exceed 80%. In parallel, with respect to multisite implementation fidelity, it was expected that experts’ mean ratings of facilitators’ adherence/competence in delivering MBI, CBT, and HE pilot group sessions with adolescents at risk for T2D with depression symptoms would meet or exceed an 80% threshold (see Sanchez et al., 2024). Finally, it was hypothesized that contamination in the HE arm would be minimal, given its distinct content and delivery style, while some contamination between MBI and CBT was expected, given overlapping therapeutic mechanisms. MBI and CBT arms were still anticipated to be distinct from one another.

Method

Participants

The BREATHE study was a multisite, pilot and feasibility randomized controlled trial with adolescents at risk for T2D, based upon a body mass index (BMI; kg/m²) ≥ 85th percentile for age/sex and family history of diabetes, and with elevated depression symptoms, based on a 20-item Center for Epidemiological Studies–Depression Scale (CES-D) total score ≥ 21 (Sanchez et al., 2024). The overarching purpose of this pilot trial was to prepare for a future multisite efficacy trial testing MBI, as compared to CBT and HE, for improving insulin resistance in adolescents at risk for T2D with depression symptoms. In the current manuscript, we report the primary aim, which was to assess multisite fidelity of training and implementation of MBI, CBT, and HE. Enrollment occurred at four sites, including Colorado State University, Children’s Hospital Colorado, the Uniformed Services University, and Children’s National Hospital. Facilitators were identified by each respective site lead investigator and were compensated for their time; pilot group participants received intervention free of charge and a financial incentive for attending at least five of the six group intervention sessions. This study is preregistered in the Clinical Trials Registry (ClinicalTrials.gov #NCT04992299).

Procedure

Group Intervention Arms

Intervention training and delivery procedures occurred remotely using HIPAA-compliant videoconferencing. Training was designed to prepare interventionists to facilitate three group arms, each with approximately five adolescents per intervention group for six weekly one-hour sessions. All arms were co-led by a trained PhD-level lead facilitator and a Bachelor’s or Master’s level co-facilitator. To minimize facilitator effects, the study was designed so that the same PhD-level lead facilitators were trained in all three intervention arms and then delivered all three intervention arms in a parallel cohort model. Co-facilitators were trained in only one intervention arm.

MBI

Learning to BREATHE (L2B; Broderick et al., 2021) is a group-based MBI designed for adolescents with the goal of teaching moment-to-moment attention and awareness without judgment. In school-based trials comparing adolescents at pre- and post-intervention, L2B has been found to improve emotion regulation and decrease perceived stress and negative affect among non-clinical samples of high school-aged youth (Bluth et al., 2016; Metz et al., 2013). Pilot trials support the feasibility and acceptability of delivering L2B with adolescents at risk for excess weight gain (Shomaker et al., 2019) and T2D (Shomaker et al., 2017). L2B has also been found to buffer the effects of chronic stress exposure on emotion regulation in adolescents facing multiple social adversities (Miller-Chagnon et al., 2024). Each of six weekly one-hour sessions covers a theme corresponding to the letters making up the word BREATHE (Body, Reflections, Emotions, Attention, Tenderness, and Healthy Habits, with the overall goal of Empowerment). Formal practices include awareness of breath, body scan, mindfulness of thoughts, mindfulness of emotions, mindful movement, gratitude, mindful speaking, and mindful listening. The curriculum also involves at-home tasks assigned to be practiced between sessions. All participants randomized to L2B received digital meditation audio-recordings, a yoga mat, blanket, homework log, and worksheets.

CBT

The Blues Program is a group-based CBT intervention designed to help teenagers learn strategies to identify negative thoughts, practice more realistic, positive ways of thinking, and incorporate more pleasant activities in their lives. The curriculum involves six weekly one-hour group sessions with at-home tasks and it has demonstrated efficacy for preventing depression among adolescents with elevated depression symptoms (Rohde et al., 2018). The Blues Program reduced depression symptoms and markers of insulin resistance in adolescents at risk for T2D with moderately elevated depression symptoms (CES-D > 21; Gulley et al., 2022; Shomaker et al., 2016). Core content includes psychoeducation, activities, and discussions covering the interconnectedness of thoughts-behaviors-mood, cognitive restructuring, and behavioral activation.

HE

The HE didactic control, derived from the HEY-Durham curriculum for middle and high school health education (Bravender, 2005), was a time and attention-matched control arm with the goal of increasing general health knowledge. HE has been implemented as a control condition in prior studies involving adolescents at risk for excess weight gain, T2D and/or depression (e.g., Shomaker et al., 2016). Topics covered in the matched, six weekly one-hour sessions included nutrition and body image, sun safety, substance use, domestic violence, gang violence and conflict resolution, depression, and signs of suicide. Session components related to mental health were descriptive and designed not to overlap with the content covered in MBI or CBT arms.

Intervention Training

Didactic

All facilitators completed training conducted by PhD-level clinical supervisors with advanced training and expertise in the respective interventions. Training included a presentation of theoretical and empirical backgrounds of each intervention, review of manuals, modeling different elements of sessions, and supporting facilitator practice delivering elements of each intervention. The duration of training was 17 hours for MBI and 7 hours for CBT and HE. MBI and CBT training protocols adhered to recommendations provided by the intervention developers; HE training was modeled on the CBT training protocol. MBI training protocols typically require more time to complete than CBT training protocols due to their emphasis on experiential learning and facilitator embodiment of mindfulness practices. The lengths of these training protocols aligned with other literature documenting that MBI training involves multiple hour training sessions delivered over many days or weeks (Francis et al., 2024) and foundational training in CBT may involve 7–10 h (Rakovshik & McManus, 2010).

Mock Groups and Supervision

Facilitators scoring at least 80% on the post-training knowledge evaluation proceeded to facilitation of mock group sessions. Booster training to review key concepts, rewatch training presentations, and re-take the knowledge evaluation was planned for any facilitators who did not initially achieve this threshold score. All facilitators led ≥ 1 mock session per arm with peer facilitators, graduate research assistants, and study staff serving as mock participants. At least three mock participants were present for each mock session; mock participants were study staff and student research assistants who were asked to behave akin to teenagers. Supervisors listened to the recorded mock sessions and provided feedback at an hour-long supervision meeting for each intervention arm. Supervision meetings covered supervisors’ written notes, general feedback relevant to all session leaders, and feedback specific to leaders or intervention activities.

Intervention Delivery

Pilot Groups

Group delivery with adolescents enrolled in the pilot trial began in August 2022 and ended in September 2023. Arms were run in parallel on three separate weeknights, with each arm meeting for one hour per week for six consecutive weeks.

Supervision

Facilitators met weekly with PhD-level clinical supervisors with advanced training and expertise in the respective intervention arms. Supervisors listened to the group audio-recording and provided written and synchronous feedback to facilitators at weekly meetings following each session. Supervisors utilized standardized forms to evaluate intervention fidelity for all intervention arms. Fidelity rating forms were co-developed with the developers of the interventions (MBI and CBT) or modeled on rating forms for other interventions (HE). Topics covered during supervision included facilitator strengths and areas of challenge, manual fidelity with attention to minimizing cross-contamination among arms, and quality of content delivery. If supervisors observed any challenges in facilitator delivery or drift from manualized content, booster training would be provided, as needed.

Measures

Demographics

All facilitators completed a demographics survey inquiring about their credentials, the field of the highest degree completed, and their site affiliation.

Facilitator Knowledge

Facilitator knowledge was assessed through quantitative knowledge evaluations administered to all facilitators prior to initiating didactic training (pre-training survey) and after completing didactic training (post-training survey). A total of three surveys, one specific to each arm (MBI, CBT, HE), were developed by the study team in collaboration with content experts to assess key content covered by each intervention. Questions were a mixture of true/false, multiple choice, matching, and fill-in-the-blank items. Sample questions in the MBI knowledge evaluation included: 1) Which statement about L2B adaptations designed to mitigate potential adverse experiences is correct? and 2) Which definition best corresponds to the guided discovery approach used in L2B?. Sample questions evaluating CBT knowledge included: 1) Select the 2 categories of positive activities that are encouraged in the Blues Program because they are well-established to have a positive impact on mood and 2) A teen gets a bad grade on a test and has the negative thought, “I’m so stupid, I’m probably going to fail high school.” The teen feels sad. According to the Blues Program, which is the most appropriate Positive Counter Thought from the following options? Sample questions in the HE knowledge evaluation included: 1) Write down 3 non-specific therapeutic techniques you can use in HEY-Durham to foster participant engagement and participation and 2) In Session #3 when brainstorming “Ways to Stay Healthy,” a participant mentions that someone could go to counseling or talk-therapy if they’re feeling stressed, because stress relates to health. What would be the most appropriate response from a Health Education facilitator from the following choices?”.

Each knowledge evaluation contained 25 questions. A point was given for each correct answer, with a maximum score of 25. The MBI knowledge evaluation contained seven true/false, one 3-option multiple choice, four 4-option multiple choice, three 5-option multiple choice, and nine matching questions. The CBT knowledge evaluation contained six true/false, one 4-option multiple choice, five 5-option multiple choice, nine matching, and three multiple answer questions. The HE knowledge evaluation contained eight true/false, seven 4-option multiple choice, six matching, and four fill-in-the-blank questions. The fill-in-the-blank questions had multiple right answers, and 1 point was given if any of 3 answers provided were correct. Knowledge evaluations demonstrate content validity, as each item was directly derived from the core didactic content of its respective intervention and developed collaboratively with the content expert for each arm. Research staff with expertise in adolescent behavioral interventions and the theoretical foundations of each study arm also reviewed knowledge items for relevance and clarity. Reviewers examined item content, and items were revised when needed to ensure alignment with the theoretical and procedural foundations of each arm (DeVellis & Thorpe, 2017; Haynes et al., 1995). Multiple item formats (e.g., true/false, multiple choice, matching, fill-in-the-blank) were used to support broad content coverage and reduce reliance on any single response type (Messick, 1994). All knowledge evaluations are provided in Online Resource 1.

Facilitator Competence

Facilitator competence was measured by expert raters who listened to 100% of mock sessions during the training period and 41 randomly selected audio recordings of pilot sessions (28% of all sessions). Raters were blind to site but not mock versus pilot status, as mock sessions included adult study staff distinguishable from adolescent participants. Random selection of sessions was generated in R under a reproducible seed. Sessions were stratified by lead facilitator such that approximately 30% of each lead facilitator’s sessions were selected without replacement. As sessions were selected within facilitator and rounded to whole numbers, the resulting selected sample represented 28% of all pilot sessions. Selected sessions were further randomized by intervention arm (MBI, CBT, HE), and approximately 70% within each arm were assigned to the arm-specific expert reviewer. The remaining 30% were assigned to the expert reviewers from the other two arms as cross-arm (contamination) sessions, such that each arm had at least one session rated by experts of the other arms. Experts were independent from clinical trainers/supervisors. One expert rater was identified for each of the three intervention arms and possessed deep domain knowledge and extensive experience in their assigned condition. The MBI and CBT expert raters had direct involvement in the development of the L2B and Blues intervention programs, respectively. The HE expert rater led the refinement of this curriculum for use in clinical trials of behavioral interventions for adolescents at risk for mental and metabolic health concerns. Expert raters utilized standardized forms to document facilitator competence. The MBI and CBT rating forms were co-created with the intervention developers; the HE rating form was modeled on these forms.

For MBI, the expert rated a total of 11 competence factors with an additional three foundational attitudes factors crucial to administering MBI to youth. Each factor was scored on a scale of 1–5 (1 = low, 5 = high). Competence factors for L2B included: 1) group participation, 2) level of interest in topic and activities, 3) clarity of instructions and explanations of activities, 4) organization, 5) extent to which facilitator followed the script, 6) timeliness of group activities during the session, 7) how comfortable facilitator appeared leading discussions, 8) openness and acceptance from facilitator, 9) engagement, 10) keeping the group focused and on task, and 11) maintaining positive interactions among group members. Foundational attitudes included: 1) engaging students in the learning experience, enhancing curiosity, encouraging student exploration, and mindful practice, 2) maintaining a primary focus on present-moment experiences, and 3) model and emphasize mindful qualities throughout the lesson.

For CBT and HE, experts rated a total of 12 competence factors. Each competence factor was rated on a scale from 10–100 (10 = poor, 100 = superior). Competence factors included: 1) facilitators’ ability to express ideas clearly and at an appropriate pace, 2) organization, 3) keeping group members on task, 4) attempt to provide equal speaking time for all members, 5) soliciting feedback, 6) listening and understanding, 7) communicate acceptance and respect, 8) enthusiasm, 9) warmth, 10) handling special problems that arose, 11) overall tone, and 12) overall rating of competence.

Facilitator Quality of Adherence

While competence ratings captured the skill with which facilitators delivered session content (e.g., clarity, pacing, group management), quality of adherence ratings assessed whether core session topics were addressed and the thoroughness with which they were covered, providing complementary but non-redundant indicators of fidelity. Facilitator quality of adherence was measured by the same expert raters, who listened to 100% of mock sessions during the training period and 41 randomly selected audio recordings of pilot sessions (28% of all sessions). Raters were blind to site but not mock versus pilot status, as mock sessions included adult study staff distinguishable from adolescent participants. Expert raters utilized standardized forms to document facilitator quality of adherence. The MBI and CBT rating forms were co-developed with the intervention developers; the HE rating form was modeled on these forms. Expert raters listened to randomly selected audio recorded mock and pilot sessions and rated adherence on a scale from 10–100 (10 = no adherence, 100 = perfect adherence). Sessions comprised 5–10 segments each; the number of segments was determined by the amount of key content covered in each session. Expert raters assessed fidelity to core components in all segments. For example, in session one of MBI, these included virtual session guidelines, participants’ mindfulness understanding, rationale for practice, and the final “E” in BREATHE. For session 1 of CBT, raters evaluated coverage of the welcome, discussing confidentiality, rapport-building, psychoeducation on thoughts-feelings-actions, strategies for changing thinking/doing, benefits of change, and weekly home exercises. For session 1 of HE, ratings focused on teacher introductions, group norms, rights and responsibilities, icebreakers, introduction to intimate partner violence, related group discussion, the violence cycle, a case example, cultural context, and a content quiz.

Contamination

To examine potential cross-arm contamination, expert raters scored pilot sessions from the other intervention arms using their own arm’s adherence form. This approach provides a structural indicator of whether content characteristic of one arm appeared in another. Each expert was blindly given audio-recorded pilot sessions. For example, the MBI expert rated at least one audio-recorded pilot session for CBT and one for HE using the fidelity/adherence form for MBI. Specifically, the HE expert was given 2 MBI sessions and 1 CBT session. The CBT expert was given 2 HE sessions and 1 MBI session. The MBI expert was given 2 CBT sessions and 1 HE session. A pilot session with low contamination was indicated by a low fidelity rating using the fidelity assessments described above (10 = no adherence, 100 = perfect adherence).

Data Analyses

All statistical analyses were performed in R, version 4.4.0. Facilitators’ pre-training survey scores were compared to post-training survey scores with paired Wilcoxon signed-rank tests, under the assumption of a Type I error rate of 0.05. Multiple comparison adjustments were not applied as the three arms were analyzed independently without direct comparison. While facilitators were affiliated with four different sites, we did not account for clustering due to site differences as virtual trainings were the same across facilitators. Adherence and competence observations are not independent as lead facilitators delivered sessions across intervention arms and contributed to both mock and pilot sessions. Given the descriptive nature of these analyses, we did not model facilitator-level clustering but rather we report the median and range of the number of sessions rated within facilitators. Median (interquartile range) as well as range of scores were summarized to confirm that all post-training scores were ≥ 80% accurate. For all arms, competence scores were expected to fall at or above 80%. MBI competence scores were linearly transformed from the initial rating of 1–5 (1 = low, 5 = high) to a 20–100 score by multiplying each score by 20 to align the 80% metric across modalities, where an MBI score of 4 corresponds to a transformed score of 80. As this transformation only standardizes reporting and does not establish psychometric equivalence between intervention arms, no direct adherence or competence comparisons between arms are made. Session adherence and competence ratings were summarized across all sessions and stratified by mock versus pilot sessions. De-identified data are publicly available through openICPSR (Project 251,503; https://www.openicpsr.org/openicpsr/project/251503/version/V1/view).

Results

Seventeen facilitators took part in training to deliver MBI (n = 11), CBT (n = 10) and/or HE (n = 9). Facilitators had BA/BS (n = 3, 17.6%), MA/MS/MPH (n = 6, 35.3%), and PhD/PsyD (n = 8, 47.1%) degrees across various fields of study including Psychology (n = 13, 76.5%), Communications (n = 1, 5.9%), Linguistics (n = 1, 5.9%), and Anthropology (n = 1, 5.9%). All lead facilitators held doctoral degrees in psychology (i.e., PhD or PsyD). Facilitators were affiliated with at least one of four enrollment sites: Colorado State University (n = 3, 17.6%), Children’s Hospital Colorado (n = 1, 5.9%), Colorado State University and Children’s Hospital Colorado (n = 2, 11.8%), Uniformed Services University (n = 7, 41.2%), and Children’s National Hospital (n = 4, 23.5%). Four lead facilitators conducted pilot sessions, with a median of 3 [range: 1–5] HE sessions, 4 [range: 1–6] CBT sessions, and 2.5 [range: 2–4] MBI sessions rated per facilitator. Six lead facilitators conducted mock sessions, with a median of 3 [range: 2–5] HE sessions, 3 [range: 2–4] CBT sessions, and 2 [range: 1–3] MBI sessions rated per facilitator. An additional 13 co-facilitators were present for a median of 3.5 rated mock sessions across arms.

Post-Training Knowledge Competency

Across all intervention arms, facilitators demonstrated significant increases in knowledge competency between the pre-training survey and post-training survey, with 100% of facilitators achieving ≥ 80% on the post-training knowledge survey across all three arms (Fig. 1). On the pre-training knowledge survey, 7 (63.6%) MBI, 8 (80%) CBT, and 2 (22.2%) HE facilitators scored below 80%. The median within-facilitator increase in the percentage of correct answers was 20% (IQR: 6%, 26%) for MBI [median pre-training = 76% (IQR: 70%, 82%); post-training = 96% (96%, 100%), p = 0.01], 24% (14%, 28%) for CBT [pre-training = 68% (66%, 70%); post-training = 92% (86%, 94%), p = 0.01], and 8% (4%, 16%) for HE [pre-training = 84% (80%, 92%); post-training = 96% (92%, 96%), p = 0.02]. Only one facilitator scored below 80% on the post-training knowledge survey and required booster training. Reported post-training scores reflect scores after the post-training booster was administered.

Fig. 1.

Distribution of the percentage of correct answers during pre- and post-training knowledge assessments for MBI (Mindfulness-Based Intervention), CBT (Cognitive Behavioral Therapy) and HE (Health Education). Boxplots display the median (center line), interquartile range (box), range (whiskers), and outliers (dots) of the data. *: p < 0.05, **: p < 0.01 indicating statistically significant differences between groups based on paired Wilcoxon Signed Rank tests. Post-training survey results represent scores after one facilitator required a post-training booster

Competence and Quality of Adherence Ratings

The training protocol was successful in ensuring facilitator competence across all three arms in both mock and pilot trial phases. Median facilitator competence was 80%, 86%, and 82.5% for mock MBI, CBT, and HE sessions, respectively (Fig. 2A). Competence was at least 80% for 11/12 (91.7%) MBI mock sessions, 16/18 (88.9%) HE mock sessions, and for all 18 CBT mock sessions. For pilot delivery sessions, competence ratings were equally high or higher. Specifically, median facilitator competence was 90%, 90%, and 97.5% for pilot MBI, CBT, and HE sessions, respectively.

Fig. 2.

Distribution of competence (2A) and quality of adherence (2B) ratings for mock and pilot sessions of MBI (Mindfulness-Based Intervention), CBT (Cognitive Behavioral Therapy) and HE (Health Education). Boxplots display the median (center line), interquartile range (box), range (whiskers), and outliers (dots) of the data

The training protocol was also successful in ensuring facilitator quality of adherence across all three arms. Median quality of adherence was 85%, 90%, and 80% for mock MBI, CBT, and HE sessions, respectively (Fig. 2B). Adherence was at least 80% for 16/18 (88.9%) HE mock sessions. All 12 MBI mock sessions and 18 CBT mock sessions had an adherence score at or above 80%. For pilot sessions, quality of adherence ratings were equally high or higher. Median facilitator quality of adherence was 100%, 90%, and 96% for pilot MBI, CBT, and HE sessions, respectively.

Intervention Condition Contamination

For MBI pilot sessions, median facilitator adherence was 100% (IQR: 95%, 100%). From contamination ratings, the MBI pilot sessions covered 40% (n = 1) of CBT topics and a median of 0% (n = 2; 0% coverage for both sessions) of HE topics. Crossover existed between the MBI and CBT arms with the CBT “Hopes and Goals”, “Changing Doing (Planning Future Fun)”, and “Changing Thinking (Planning Ahead)” domains having some coverage in MBI. For CBT pilot sessions, overall facilitator adherence was 90% (86%, 92%). From contamination ratings, the CBT pilot sessions covered 0% (n = 1) of HE topics and a median of 43% (n = 2; 40% and 45% coverage from separate sessions) of MBI topics, where, similarly, crossover existed between the CBT and MBI arms with the MBI “Stream of Gratitude” and “Awareness of Breath/Short Loving Kindness” domains having some coverage in CBT. Finally, for HE pilot sessions, overall facilitator adherence was 96% (90%, 100%), with no overlap with either MBI (n = 2; 0% coverage for both sessions) and CBT (n = 1) topics from contamination ratings. Despite some crossover between MBI and CBT topics, the median adherence was 10% for all contamination ratings.

Discussion

Mental and metabolic health challenges, particularly depression and T2D, are rising among adolescents in the U.S. and often co-occur in ways that compound current and future health risks (e.g., Brody & Hughes, 2025; Lawrence et al., 2021; Shomaker et al., 2011; Suglia et al., 2016). Given the limited treatment options for adolescent-onset T2D and evidence linking depression to insulin resistance, there is a critical public health need for preventative interventions that address adolescent mental-metabolic health (e.g., Perng et al., 2023; Zeitler et al., 2012). MBI may help reduce the link between depression, insulin resistance, and T2D, especially in youth experiencing elevated levels of stress (Pinhas-Hamiel & Hamiel, 2020), and MBI training and delivery fidelity is integral to this potential (e.g., Broderick & Schussler, 2021).

This multisite pilot RCT aimed to prepare for a future efficacy study comparing group MBI, CBT, and didactic HE for improving insulin resistance in adolescents at risk for T2D with depressive symptoms. A key goal was to assess the feasibility of implementing high-fidelity training and delivery across sites. We hypothesized that facilitators could be trained with ≥ 80% knowledge acquisition and protocol-specific competence, and that mock and pilot sessions would show ≥ 80% fidelity. We also expected minimal contamination in HE and some overlap between MBI and CBT due to shared therapeutic elements.

Facilitators across all three intervention arms showed meaningful gains in knowledge following training, with all facilitator trainees achieving or surpassing the target knowledge threshold. This suggests that didactic protocols effectively equipped facilitators with the necessary foundational knowledge to deliver each intervention. The training protocol also supported facilitator competence and adherence across all three intervention arms, even when learning and delivering the three modalities in parallel. Facilitators demonstrated strong performance during both mock and pilot sessions, with consistently high ratings that met or exceeded benchmarks for fidelity. Moreover, findings indicate that facilitators remained adherent to their assigned intervention protocols, with minimal overlap or contamination in the HE arm and some contamination between the MBI and CBT arms, as expected due to fundamental therapeutic similarities noted in these two conditions. This finding suggests adequate intervention arm differentiation was maintained throughout pilot delivery. The present findings also align with qualitative data obtained from post-intervention focus groups with adolescents in this sample that highlighted perceived differences between MBI and CBT content and facilitation (Thompson et al., in press).

The current findings are consistent with prior literature demonstrating that facilitator training can achieve high levels of competence and adherence in psychosocial interventions, particularly in academic or controlled settings (e.g., Chang et al., 2025; Jordans et al., 2022; Martin et al., 2021). This study offers a rigorous methodological approach to quantifying training and delivery fidelity outcomes across three distinct group-based programs, MBI, CBT, and HE, within a multisite adolescent sample with co-occurring mental-metabolic health risk. These results provide support for the feasibility of maintaining treatment differentiation and quality across modalities and multiple national sites in a virtual telehealth setting. The current pilot and feasibility trial establishes a foundation for broader, real-world dissemination of evidence-based programs addressing adolescent depression and T2D risk and offers a replicable model for cross-site training and implementation quality assurance.

Limitations and Future Directions

This study should be considered with recognition of several limitations. First, the sample was limited to adolescents experiencing both elevated depressive symptoms and risk for T2D, which may constrain generalizability to fidelity with broader adolescent populations. Second, although the trial involved multiple sites, the sample size was insufficient, by design, to evaluate the potential influence of facilitator characteristics (e.g., training background, clinical experience) on fidelity. Further, all facilitators were based in academic settings and had familiarity with clinical research, potentially limiting applicability to real-world or community-based implementation contexts. Similarly, the delivery of all three interventions occurred in a virtual telehealth setting and therefore, fidelity of delivery may not directly transfer to in-person programs. Additionally, the study did not systematically measure prior experience with MBI or CBT delivery nor MBI facilitators' personal mindfulness practice, which has been speculated to be a contributor to competence in delivering MBI (e.g., Crane et al., 2012).

With respect to the arm-specific knowledge tests for facilitators, a comprehensive psychometric validation was not conducted, including test–retest stability or convergent and divergent validity. Internal consistency reliability of the knowledge evaluations could not be estimated due to the small number of facilitators per arm. Future research could include a comprehensive validation of the knowledge tests for the MBI, CBT, and HE interventions administered in this pilot with a larger sample of facilitators (see Online Resource 1). A key limitation of this study is that competence and adherence ratings were completed by a single expert rater per arm, each of whom had prior involvement in developing their respective intervention. This design precludes estimation of inter-rater reliability and introduces the potential for allegiance or leniency effects, so fidelity findings should be interpreted as preliminary indicators of training and delivery quality rather than definitive estimates of competence/adherence. Although session recordings were retained, double-coding was not feasible within the present study because these recordings are being used in a planned companion project comparing expert raters and non-expert raters. This future work will provide the reliability evidence needed to more fully validate this rating approach and offer an additive contribution to the field of mindfulness training and delivery fidelity. Along these lines, as competence, adherence, and cross-arm contamination ratings were completed by the same expert reviewer, using instruments that the reviewer helped develop, this analysis has the potential of common method bias (Podsakoff et al., 2012). Consequently, some consistency across ratings may reflect shared rater effects rather than true fidelity performance. While the use of single arm-expert raters provided standardized evaluation across sessions, future studies would benefit from independent raters, with an assessment of inter-rater reliability, to reduce common method variance. Moreover, as competence and adherence are not uniformly rated across arms, direct comparison of arm intervention cannot formally be tested. As such, between arm differences may be attributable to scoring construct rather than true competence or adherence differences. Another design caveat to the presented analyses is that lead facilitators contributed to multiple rated sessions within each arm. As such, the adherence and competence data are not independent as lead facilitators led an unequal number of rated sessions. Consequently, reported distributions should be treated as non-independent descriptive summaries as they may partially reflect facilitator-specific delivery metrics in addition to intervention-specific fidelity. Finally, interpretation of cross-condition contamination is limited by the structure of the rating forms and the small number of cross-form sessions rated. Because each arm’s adherence form captures arm-specific content, low cross-form scores may indicate either true absence of contamination or simple non-overlap in form items. These findings should be viewed as preliminary indicators of arm differentiation rather than definitive contamination estimates.

Despite limitations, this study incorporates several methodological and practical features relevant to future implementation efforts. The multisite design provides information about feasibility across diverse settings, and the structured training approach offers a reproducible method for assessing facilitator competence and adherence. Examining three intervention modalities highlights differences in protocol-specific training needs and treatment differentiation. Facilitators met performance benchmarks, and cross-arm contamination was limited, supporting the integrity of intervention delivery. Situating this work within the broader context of adolescent mental and metabolic health reflects the ongoing need for scalable, prevention-focused interventions.

This study provides evidence that structured training may prepare facilitators to deliver MBI, CBT, and HE with fidelity, demonstrating that MBI achieves fidelity benchmarks comparable to those of CBT and HE when supported by standardized training. MBI facilitators showed significant knowledge gains and consistently met competence and adherence benchmarks across mock and pilot sessions, meeting criteria for “high” fidelity (Gould et al., 2016). Preliminary evidence of limited cross-condition contamination further supports the effectiveness of the training in cultivating delivery integrity. These outcomes reinforce the central role of comprehensive training in advancing implementation quality, particularly for MBI, where both protocol adherence and facilitation skill are foundational to intervention efficacy (Baer et al., 2019; Broderick et al., 2021; Dimidjian & Segal, 2015). By demonstrating that MBI can be trained and delivered successfully across multiple sites and differentiated from CBT with only moderate overlap despite shared therapeutic mechanisms, this study strengthens the infrastructure for future multisite efficacy trials and expands the field’s understanding of how best to scale mindfulness-based approaches with fidelity.

Present findings provide crucial infrastructure for advancing the unique promise of MBI for adolescent mental and metabolic health concerns. MBI requires a distinct set of facilitation skills rooted in mindfulness principles such as present-moment awareness and nonjudgment. The success of the training protocol in cultivating these competencies—evidenced by strong knowledge gains, high competence and adherence ratings, and contamination limited to conceptually adjacent MBI-CBT elements—confirms that MBI may have the potential to be delivered with integrity at scale when supported by structured training. These results are especially relevant given the heightened vulnerability of adolescents to stress-related metabolic dysfunction and the theoretical grounding of MBI in stress-regulation processes. The primary contribution of this study is to inform training and delivery considerations, thereby expanding the translational potential of MBI to address pressing adolescent health concerns.

Future research should prioritize fully powered comparative efficacy studies to evaluate the health impacts of MBI, CBT, and HE in adolescents at risk for mental and metabolic health problems. Incorporating longer-term follow-up assessments will be critical for understanding the durability of intervention effects and potential for T2D prevention. Expanding to more diverse adolescent populations and facilitator types—such as school- or community-based providers—can enhance real-world generalizability. Additionally, future studies should explore mechanisms of change (e.g., mindfulness skills, emotion regulation) and examine facilitator-level variables, including personal mindfulness practice and training format, to better understand what drives successful delivery. Hybrid effectiveness-implementation designs may also support scalable, context-sensitive dissemination of these interventions in under-resourced settings. Finally, future research is needed into the development and refinement of scalable training models that are both efficient and effective across varying levels of expertise to support broader dissemination and equitable implementation of evidence-based practices.

This multisite pilot trial offers foundational evidence that structured training can support the high-fidelity delivery of MBI, CBT, and HE interventions to adolescents with co-occurring mental-metabolic health risks. The consistent attainment of competence and adherence benchmarks, coupled with limited, expected cross-condition contamination, underscores the feasibility of implementing these protocols across sites. This study advances the science of MBI by demonstrating that facilitators can be effectively trained across sites to deliver this intervention, which requires a skillset that is distinct from other behavioral health interventions, with fidelity while simultaneously delivering alternative behavioral health interventions. These findings pave the way for future research to evaluate efficacy, mechanisms, and delivery models of MBI for adolescents at risk for depression and T2D. Results of this study contribute to the infrastructure for scalable, evidence-based interventions that address the urgent need for integrated mental-metabolic health prevention.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contribution

SP and MC wrote the main manuscript text. SB conducted analyses and prepared Figs. 1–2. All authors contributed to study conceptualization and reviewed the original manuscript and all revisions. MC and NS contributed to project administration. EM, MK, MTK, and RLT supervised research activity planning and execution and contributed to funding acquisition. LS led study conceptualization, provided mentorship of research activity including manuscript preparation and revision, and led funding acquisition. SP, MC, SB, and LS revised manuscript and created supplemental materials to address reviewer and editor comments.

Funding

This work was supported by the National Center for Complementary and Integrative Health and the Office of the Director of the National Institutes of Health (Grant #s U01AT011008, R01AT011008).

National Center for Complementary and Integrative Health,U01AT011008

Data Availability

The de-identified dataset supporting the findings of this study are publicly available on openICPSR (Project 251,503; https://www.openicpsr.org/openicpsr/project/251503/version/V1/view).

Declarations

Ethics Approval

The Colorado Multiple Institutional Review Board approved all procedures. This study was performed in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments.

Informed Consent

Informed consent was obtained from parents/guardians of youth who participated in pilot group sessions.

Conflict of Interest

The authors declare that they have no conflict of interest.