This Health Canada-regulated feasibility RCT (ClinicalTrials.gov: NCT05105802) used a parallel-group (1:1), blinded, single-crossover design. Participants in the control group who completed the initial 4-week study and wished to pursue the experimental intervention crossed over (modules 1‐4) during weeks 5‐8, while those in the experimental group could extend their MBI training for an additional 4 weeks (modules 5‐8; Figure 1). To maintain focus on the primary endpoints, only the 4-week outcomes are presented in the main text; week-8 outcomes are reported in Multimedia Appendix 1 [26,45]. Methods and results of the feasibility of the magnetic resonance imaging (MRI) component are presented in Multimedia Appendix 1 [26,45].

This study was approved by the Research Ethics Board (protocol #20/720X) of the Children’s Hospital of Eastern Ontario (CHEO) Research Institute in August 2020. Authorization from Health Canada was granted in July 2022. The study was carried out according to the principles outlined in the Declaration of Helsinki and Good Clinical Practices, and within the laws and regulations of the Tri-Council Policy Statement.

Before consent was sought, the study was introduced, and its purpose was explained verbally. The consent form also presented the purpose of the research, the procedures, potential risks and benefits, and the use and security of the data. Informed written consent was obtained from the parents, and participants were deemed capable of consenting. Informed assent was obtained from the participants deemed by the research assistant (RA) to be cognitively unable to provide informed consent.

The participants were remunerated with a CAD $20 (US $15) gift certificate (eg, Tango) for completing surveys. Participants received an additional CAD $20 (US $15) gift certificate for completing the MRIs. The gift certificate was sent electronically to their email address. They also received a parking voucher or vouchers for in-person meetings and a letter attesting that they have completed 20 hours of volunteer work (5 hours per week completing the intervention).

Participant information was coded using study identification numbers. Research personnel took all appropriate and customary steps to ensure that data remain secure and that patient privacy and confidentiality are maintained.

The full protocol was published previously [46]. Briefly, participants were recruited from the CHEO emergency department (ED) and the 360 Concussion Care (360cc) clinic, an interdisciplinary tertiary health clinic in Ottawa, from October 2022 to June 2024. Families of adolescents with a concussion were approached by an RA to discuss the research study. Interested families completed a screening form on a tablet with the RA, and if eligible (concussion confirmed by a physician) and enrolled, filled out electronic questionnaires via Research Electronic Data Capture (REDCap) [47 48] to collect patient demographics, diagnostic history, injury characteristics, and a concussion symptoms checklist.

The day after the ED or 360cc visit, an RA called participants to randomize them into either the experimental or control group. They provided instructions for installing AmDTx on the participant’s mobile device of their choice. Participants completed Day 1 surveys by phone or via REDCap and began the intervention that same day. At the 4-week mark, participants attended an in-person appointment to complete the cognitive testing at either the CHEO Research Institute or the Royal Ottawa Hospital (for those enrolled in the neuroimaging component; see below). After this visit, participants in the experimental group were offered 4 additional weeks of psychoeducation modules and mindfulness practices. Control group participants were given the option to cross over and begin the DTx-MBI. Follow-up questionnaires were completed on REDCap at 1, 2, 4, and 8 weeks. A subset from each group was prospectively approached and enrolled for a neuroimaging (Multimedia Appendix 1 [26,45]).

Participants were randomly assigned to one of 2 groups using a 1:1 allocation ratio stratified by sex. (1) Experimental group: DTx-MBI program, or (2) control group: sham digital attention-matched puzzle game. The CHEO Clinical Research Unit provided data management services for this study and retained randomization codes. A statistician who was not involved in the study created and maintained the master list.

Only the RAs responsible for randomization and app setup were unblinded to participants’ group assignments, as this was necessary for intervention setup and delivery. These RAs were blinded to the randomization sequence until the point of group assignment and were not involved in outcome assessment or data analysis.

During the consent process and follow-up, we gave controlled information to participants about both interventions, while keeping them unaware of the hypothesis, design, and treatment assignment (experimental or control). The study hypothesis and the distinction between experimental and control conditions were disclosed only at the 4-week follow-up, at which point DTx-MBI participants could choose to extend their intervention and the Sham participants could cross over to the experimental condition.

Adolescents presenting to the ED or 360cc within 7 days of sustaining a direct or indirect traumatic brain injury were eligible to participate if they: (1) were aged 12 to <18 years; (2) had a diagnosed concussion (by a physician); (3) had 1 highest level of certainty symptom (eg, loss of consciousness, posttraumatic amnesia) or 2 higher level of certainty symptoms (eg, nausea/vomiting, headache) immediately or within 1 hour of injury from the adapted version of the Centers for Disease Control and Prevention tiered framework [49]; (4) had a score of >4 on the Predicting and Preventing Postconcussive Problems in Pediatrics clinical risk score, a validated prediction score to identify those at risk of poorer outcomes at 4 weeks postinjury [3]; (5) were proficient in English.

Patients were excluded if they presented with traumatic brain injuries with any of the following: (1) Glasgow Coma Scale score of ≤13; (2) trauma-related abnormality on standard neuroimaging studies (if performed) [50]; (3) neurosurgical operative intervention, intubation, or intensive care required; (4) multisystem injuries with the treatment requiring hospital admission, operating room, or procedural sedation in ED; (5) severe neurological developmental delay resulting in communication difficulties; (6) intoxication to alcohol or drugs at the time of ED/360cc presentation as per clinician judgment; (7) no clear history of head trauma as the primary event to the concussion (eg, seizure and migraine attack); (8) prior psychiatric hospitalization; (9) inability to obtain written informed consent or assent; (10) legal guardian not present (certain forms needed to be completed by parents or legal guardians); (11) no internet and mobile or tablet access; and (12) was in psychological therapy at enrollment.

A total of 99 out of 124 eligible patients were randomized to DTx-MBI (n=49; median age [IQR]=15.28 [13.66‐16.19] years; 30 [61%] female) or Sham groups (n=50; median age [IQR]=14.92 [13.32‐16.71] years; 30 [60%] female) (Figure 2). A total of 65 participants were recruited from the ED and 34 from 360cc. Demographics and clinical characteristics were comparable between groups among all randomized participants and those who completed the 4-week follow-up (Table 2).

Four out of the 5 prespecified feasibility outcomes met the “green zone” benchmarks.

Given low adherence, we conducted a post hoc analysis to examine factors associated with intervention adherence. In the post hoc exploratory multiple regression, none of the demographic or preinjury factors showed statistically significant associations with activity engagement during the 4-week intervention period (Table 4).

Among participants who completed the intervention, weekly adherence rates (ie, meeting the threshold of ≥10 minutes of “meaningful” app engagement for 4 days over a 7-day week) declined from week 1 (DTx-MBI: 22/44, 50%; Sham: 31/45, 69%) to week 4 (DTx-MBI: 17/44, 39%; Sham: 17/45, 38%). Similarly, median weekly engagement time decreased from 40.9 minutes (IQR 20.2‐89.9) to 17.1 minutes (IQR 0‐76.8) in the DTx-MBI group and from 75.2 minutes (IQR 26.2‐108.1) to 12.6 minutes (IQR 0‐60.0) in the Sham group over the 4 weeks (Figure 3).

This feasibility RCT indicated that a larger-scale trial evaluating the effectiveness of DTx-MBI in youth with a concussion compared with an attention-matched sham intervention is viable, with 4 out of the 5 prespecified feasibility benchmarks—eligibility, recruitment, credibility, and retention—being met or exceeded. Only adherence fell slightly below the target threshold, with engagement levels comparable to other digital health interventions in youth [28], suggesting that minor adjustments could improve adherence in the large-scale efficacy RCT.

The high eligibility rate suggests that the inclusion criteria were well aligned with the characteristics of the target population, and the high recruitment rates underscore the strong relevance and suitability of this intervention for the target population. Moreover, retention at 4 weeks approached 90% in both groups, reflecting strong participant commitment and the feasibility of our follow-up strategy. The high credibility and satisfaction across both intervention and control groups support the acceptability of the app-based interventions, regardless of their content. This reinforces findings from our earlier open-label preliminary study [51,80]. In the mixed methods evaluation (n=7), credibility, expectancy, and satisfaction ratings were even higher, although this may have been influenced by the small sample size, open-label design, and participants’ awareness that they would be asked for feedback during the 4-week interview [51]. This blinded RCT strengthens the evidence for the acceptability of the DTx-MBI through a more rigorous randomized design and the inclusion of an attention-matched comparator. Although treatment expectancy was above the scale midpoint for all participants, those in the MBI-DTx reported a qualitatively higher expectancy that the treatment would be helpful, suggesting greater perceived relevance or believability of the experimental intervention. The treatment expectancy ratings in the Sham group were just below the threshold considered “good” in this study, whereas the experimental group met the benchmark. In the efficacy RCT, we will address this potential imbalance by providing participants in the Sham group with additional education about how cognitive training may promote recovery through neuroplasticity during enrollment.

Notably, in this study, no adverse events related to app use were reported, supporting the intervention’s safety. Furthermore, the optional MRI component also met the feasibility benchmarks, with high enrollment and scan completion rates, suggesting that a neuroimaging component can be successfully integrated into future trials.

Adherence, while below the 70% benchmark, remained within an acceptable range (nearly 60%), with slightly higher adherence in the DTx-MBI group across the 4 weeks. Weekly adherence declined over the 4 weeks in both groups, with median weekly engagement time decreasing substantially from week 1 to week 4 in both groups, with a more pronounced decline in the Sham group. This reduction may reflect the natural waning of novelty or motivation over time. This pattern is consistent with Eysenbach’s “Law of Attrition,” which posits that drop-off in usage is an expected and inherent feature of eHealth interventions, as users often disengage over time regardless of intervention quality or intent [81]. The drop-off in app engagement around weeks 2 and 3 may also coincide with the typical course of concussion recovery, a point at which participants may no longer perceive the intervention as necessary [9]. While the DTx-MBI group began with lower adherence rates than the Sham group, they sustained slightly higher engagement by week 4, potentially indicating greater long-term relevance or perceived efficacy or benefit of the mindfulness-based content. Previous studies in adolescents and adults have reported wide variability in adherence to digital MBIs, ranging from 40% to 92% [28]. In this study, we did not identify any participant characteristics associated with lower app engagement. However, prior work suggests that psychological factors such as resilience and coping style may moderate response to behavioral concussion interventions, suggesting that engagement may depend on characteristics not captured here [82]. Future efficacy trials may benefit from enhanced engagement strategies, such as personalized reminders and motivational interviewing. However, such strategies may reduce scalability and accessibility, as they require additional resources and personalization. Implementing them only when early signs of disengagement emerge (as observed in this study around week 2) may help balance these trade-offs.

It is also important to note that the engagement tracking metrics may under-report actual engagement, as the developer’s system was designed to capture only completed “meaningful interactions.” Sessions that were started but not completed (eg, if a participant began a meditation recording but exited before finishing) were not logged. Likewise, activities and assessments (snapshots, journaling, etc) that lasted at least 3 minutes but were not explicitly submitted by the participant once they were done, were not logged. The same goes for the sham app. Based on the developer’s user testing, these incomplete sessions and unsubmitted assessments may account for roughly 10%‐20% of potential meaningful interactions. As such, the engagement data presented here represent only fully completed meaningful interactions, and actual usage may have been modestly higher than reported.

Although not powered for efficacy, both groups seemed to show reductions in symptom burden and improvements in quality of life at the 4-week assessment. The Sham group appeared to report higher sleep and school quality of life at 4 weeks, while these scores declined slightly in the DTx-MBI group. At 8 weeks, among participants who continued with the DTx-MBI, sleep and school quality of life increased, whereas they showed minimal change in the Sham participants who crossed over to the MBI intervention. Moreover, performance on the NIH Toolbox Pattern Comparison task, a measure of processing speed, was modestly higher in the DTx-MBI group compared with the Sham group, although this difference may reflect pre-existing group differences rather than a potential intervention effect since no baseline measures were available for the NIH Toolbox. These descriptive comparisons were not formally tested and are presented only to provide preliminary signals to inform future fully powered trials. Accordingly, these preliminary findings should be interpreted cautiously and warrant further investigation in a fully powered trial to determine the intervention’s efficacy and clarify its impact across recovery outcomes.

Despite overall strong feasibility metrics, this study has some limitations. First, estimates of recruitment, eligibility, adherence, and retention were based on a modest sample size, which may limit precision. Second, the control group’s treatment expectancy was qualitatively lower and just below the benchmark for being considered “good,” which may suggest that some participants suspected they were not receiving the genuine cognitive training intervention, indicating partial unblinding or differential perceived value, despite the use of an active sham. This may have been further compounded by the nature of the control condition, which consisted of math puzzles without concussion-specific content. However, it should be noted that all participants received standard-of-care concussion guidance regardless of group allocation, which is itself expected to support natural recovery and limits the interpretation that any observed differences reflect expectancy alone. These factors will need to be carefully addressed in the design and interpretation of the efficacy trial. Last, as expected for a feasibility study, it was not powered to detect efficacy, and outcome findings should be considered exploratory and interpreted accordingly. This is particularly relevant for weeks 5‐8, which included only a small subset of participants who elected to continue.

There are a few established preventive nonpharmacological interventions for PSAC in pediatrics [83]; however, psychological approaches are increasingly explored. This feasibility trial supports the implementation of a larger RCT to rigorously evaluate the clinical efficacy of the DTx-MBI in pediatric concussion when compared with an attention-matched cognitive sham delivered on the same digital platform. With refinements to improve adherence, especially in weeks 2 to 4, we will be positioned to evaluate whether this intervention reduces symptom burden and enhances multifaceted recovery outcomes, including quality of life, mood, cognition, and neurophysiology, in youth with a concussion.