- Split View
-
Views
-
Cite
Cite
Shari L Wade, Craig Sidol, Lynn Babcock, Matthew Schmidt, Brad Kurowski, Amy Cassedy, Nanhua Zhang, Findings from a Randomized Controlled Trial of SMART: An EHealth Intervention for Mild Traumatic Brain Injury, Journal of Pediatric Psychology, Volume 48, Issue 3, March 2023, Pages 241–253, https://doi.org/10.1093/jpepsy/jsac086
- Share Icon Share
Abstract
The aim of this study was to examine the efficacy of the SMART (Self-Management After Recent Traumatic brain injury) program and potential moderators.
Parallel randomized controlled trial (ClinicalTrials.gov Identifier: NCT03498495) was conducted. Eligibility criteria included treatment for mild traumatic brain injury in the emergency department and age 11–18 years. Participants were assigned equally to SMART (n = 35) or usual care (UC; n = 36). SMART included symptom monitoring and online modules supporting the return to activities and symptom management. Coping and quality of life (QoL) (primary outcomes) and post-concussive symptoms (secondary outcome) were assessed at baseline and weekly for 4 weeks.
Groups did not differ in coping, QoL, or return to pre-injury symptom levels at any time point. Problem-focused engagement (PFE) moderated group differences over time (p = .02). At high PFE, UC participants reported lower QoL at time 1 (effect size [ES] = 0.60); SMART participants did not report a decline at any point. At low PFE, SMART participants reported declining QoL from pre-injury to time 1 (ES = 0.68), whereas UC participants reported an increase from time 1 to 3 (ES = 0.56). PFE also moderated group differences on the Health and Behavior Inventory (HBI) cognitive (p = .02) and somatic symptom scales (p = .05). At high PFE, SMART participants reported a more rapid return to pre-injury levels than UC participants (p = .05). Resilience also moderated group differences in QoL and HBI cognitive recovery.
Effectiveness of the SMART app varied based on preinjury coping styles and resilience, underscoring the potential need to tailor treatments to individual characteristics.
Introduction
Mild traumatic brain injury (mTBI), commonly referred to as concussion, affects millions of children annually in the United States, with international estimates suggesting far higher rates and increasing incidence in recent years (Langer et al., 2020). Symptoms include physical concerns (e.g., balance problems, headaches, blurred vision, and fatigue) and cognitive symptoms (e.g., attention and concentration difficulties, irritability, and memory problems). While most children recover fully within a month following injury, as many as a third experience persistent post-concussive symptoms, making the development of interventions to promote symptom resolution and recovery imperative (Babcock et al., 2013; Barlow, 2016; Eisenberg et al., 2014).
Approaches to the management of acute mTBI recovery have evolved in recent years from complete cognitive and physical rest to more active interventions, including subthreshold aerobic exercise soon after injury (Leddy et al., 2019). Active coping strategies and a resilient mindset (e.g., feeling capable of handling adversity) have also been linked to recovery, with problem-focused coping contributing to fewer symptoms (Woodrome et al., 2011). Taken together, these findings suggest that a stepwise return to activities coupled with changing how one thinks about and copes with the mTBI and its consequences have the potential to improve the recovery trajectory. Support has emerged for the utility of cognitive-behavioral approaches to address expectations and coping (McNally et al., 2018). However, most studies addressing cognitive coping have focused on longer term rather than acute recovery.
mTBI constitutes a physical and psychological stressor; consequently, factors that influence an individual’s ability to successfully respond and adapt to stress are likely to play a role in mTBI recovery. Coping strategies have been conceptualized along the dimensions of engagement (e.g., engaged vs. disengaged/avoidant) and focus (e.g., problem-focused efforts to address the situation vs. emotion-focused efforts to manage emotions; Lazarus & Folkman, 1984). In this regard, limited research suggests that an engaged and problem-focused coping style may contribute to faster or more complete recovery. In adults with mTBI, passive or avoidant coping styles have been associated with poorer outcomes (Maestas et al., 2014; Scheenen et al., 2017). Similarly, a prospective study of children with mTBI and orthopedic injuries found that higher levels of problem-focused, engaged coping were associated with lower post-concussive symptom levels, whereas emotion-focused coping and disengagement/denial were associated with higher levels of post-concussive symptoms (Woodrome et al., 2011). Taken together, these findings suggest that how one routinely responds to life challenges may influence the course of post-concussion recovery, with active and engaged coping styles corresponding to better recovery.
An individual’s response to their mTBI may also be shaped by their resilience prior to the injury. While coping is thought to involve a situation-specific response to stresses or life experiences, some researchers have conceptualized resilience as trait-like confidence and persistence in the face of adversity or challenge (Connor & Davidson, 2003). Thus, resilient individuals perceive themselves as able to bounce back from traumatic or stressful events, a characteristic that has been linked to more rapid and complete recovery after both sports- and nonsports-related mTBI (Bunt et al., 2021; Ernst et al., 2021; Losoi et al., 2015; McCauley et al., 2013; Sullivan et al., 2016). Although limited, extant research in children with mTBI also suggests that psychological resilience may serve as a protective factor, contributing to lower post-concussive symptoms long after injury (Durish et al., 2018, 2019). However, prior research has not examined the relationship of resilience to acute symptom resolution in children more broadly, nor has it considered resilience as a potential moderator of treatment response.
We previously developed and tested the Self-Management After Recent Mild Traumatic Brain Injury intervention (SMART), a self-guided online intervention to promote proactive symptom monitoring, stepwise return to activities, and active management of post-concussive symptoms during the initial weeks post-mTBI. The SMART intervention was designed to engage the adolescent in their recovery and to promote an active, problem-focused approach to coping with the mTBI and its symptoms. We adopted an eHealth approach to improve access and reduce potential internalized stigma (Boydell et al., 2014). Specifically, SMART included daily online symptom monitoring to guide return to activities coupled with eight psychoeducational modules providing information about mTBI and strategies for actively managing and coping with symptoms. We tested SMART in a single-arm pilot with 13 adolescents enrolled in the emergency department (ED) following a recent mTBI. We found high levels of satisfaction and significant reductions in post-concussive symptoms over time, with all participants returning to baseline levels of symptoms (Babcock et al., 2017; Kurowski et al., 2016). Stakeholder feedback from this pilot was used to redesign the intervention prior to the larger randomized clinical trial (RCT) that is reported in the current article (Schmidt et al., 2020).
We report findings from an RCT comparing SMART to usual care (UC). We examined changes in self-reported coping strategies and quality of life (QoL) as primary outcomes and post-concussion symptoms as a secondary outcome. We hypothesized that participants assigned to SMART would report higher levels of problem-engaged coping, hereafter referred to as problem-focused engagement (PFE), and better QoL over time, relative to the UC group, as well as more rapid return to baseline symptoms. In secondary analyses, we examined moderators of treatment efficacy. Given that SMART was designed to promote self-monitoring, active coping, and a stepwise return to activities following mTBI in adolescents, we hypothesized that its efficacy might vary as a function of the adolescent’s characteristic coping style and psychological resilience. Specifically, we hypothesized that participants in SMART with higher levels of PFE and resilience would report more rapid recovery of QoL and resolution of symptoms relative to those in SMART with lower levels of PFE and resilience, as well as those assigned to the UC condition. Given that females have been shown to have slower recovery than males, we also examined sex as an exploratory moderator of recovery (Hannah et al., 2021; Iverson et al., 2017).
Methods
Design
This study was registered with clinicaltrials.gov (ClinicalTrials.gov Identifier: NCT03498495) and approved by the Institutional Review Board. We conducted a two-arm, parallel design RCT with participants allocated equally to the SMART condition or UC. The trial began in May 2019, with recruitment ending in August 2020 due to funding constraints. We followed the CONSORT checklist, and the final checklist is available as Supplementary File 1. Participants, ages 11–18 years, were recruited from the ED of a large, urban midwestern children’s hospital with a level 1 trauma center. We targeted emerging adolescence/adolescence, a period of increasing autonomy, given the intervention’s focus on self-monitoring and self-management/active coping with post-concussive symptoms. Eligibility required an mTBI, defined as a witnessed blow to the head, injury with head acceleration/deceleration, or self-reported injury with the evidence of head trauma (McCrory et al., 2017). Exclusionary criteria included: Glasgow Coma Scale score of less than 13, non-English speaking, no internet access, more than one moderate extracranial injury (i.e., an Abbreviated Injury Severity Scale >2 to that region; Loftis et al., 2018), evidence of other reasons for altered mental status (e.g., ingestions), or a history of neurologic or cognitive conditions (e.g., autism, intellectual disability). Adolescents provided written assent (or consent for those aged 18 years) and parents provided written consent. Data are available on request.
Study Procedures
Trained research coordinators screened potential participants in the ED, confirmed eligibility, obtained informed consent/assent, and asked parents and adolescents to complete patient-reported measures electronically in REDCap using a computer tablet. If the youth was unable to complete data entry due to self-acknowledged cognitive or visual impairment, the coordinators administered the questionnaires and recorded the answers. To understand and control for preinjury differences in QoL and symptoms, parents and youth completed measures regarding preinjury functioning. Participants also rated current symptoms, resilience, and coping strategies, as described in greater detail below. See Supplementary Figure 1 for a chronological depiction of study procedures.
The primary statistician (N. Zhang) generated the random allocation sequence, which was administered through REDCap. Stratified randomization was performed in SAS using a computerized random number generator to assign participants to SMART or UC; stratification on sex was used to minimize any imbalance of sex between the two treatment groups. The randomization allocation table was loaded into REDCap’s randomization module, which generated treatment assignment based on that table once all relevant data (consent, gender, etc.) had been entered. The research coordinator did not have access to the sequence and was unaware of group assignment until it was generated by REDCap.
Neither participants nor the research team was blinded to group assignment. For those assigned to the SMART arm, the coordinator provided the website log-on, demonstrated how to access the program and, when possible (given timing and the patient’s symptoms), walked through the first module with them. Given the nature of the intervention, fidelity was not assessed.
We collected follow-up data regarding UC, coping, QoL, and symptoms at 4 weekly intervals. Participants received text reminders to complete follow-up assessments before, during, and after the scheduled completion date on a set schedule. Participants received $10 at initial enrollment, $15 at week 1, $20 at week 2, $25 at week 3, and $30 at week 4, for a total of $100. No additional compensation was provided for completion of the intervention.
Interventions
Usual Care
UC consisted of standard written instructions about how to manage common post-concussive symptoms, potential high-risk symptoms requiring immediate evaluation, stepwise guidelines for return to activities, sports, and school, and instructions to seek outpatient reevaluation and follow-up. They also received a link to the CDC mTBI site (Centers for Disease Control and Prevention, 2019). We tracked resource utilization at weekly follow-ups as an index of UC.
SMART Intervention
In addition to UC, SMART participants received access to the SMART online program, a web app consisting of eight interactive, multimedia psychoeducational modules partnered with a tool for tracking symptoms and daily activities over time. The learning modules provided information regarding the recovery process and return to activities, as well as training in symptom management strategies (e.g., breathing exercises for pain). Overall, the program’s framework and philosophy promoted resilience and active, problem-focused coping through cognitive reframing, planning and problem-solving around managing symptoms and activity restrictions, and self-advocacy. Upon logging on, participants were directed to assess their symptoms using a digital version of the 22-item Post-Concussion Symptom Scale (Kontos et al., 2012) and given feedback on whether their symptoms were improving or worsening (based on a >5-point change from the average of the prior two measurements). Participants then tracked their activities over the prior 24 h, including sleep, school, physical activities, and mental activities/brain time. Participants were asked to reflect on the relationship between daily activities and symptoms and indicate how they would adjust their activities during the coming day. After assessing their symptoms and activities, participants were directed to the modules. The eight online learning modules provided a program introduction/overview, information about common symptoms and timelines for recovery, step-by-step guidelines for return to activities, and strategies for: self-care and managing symptoms, staying positive, managing attention and concentration challenges, problem-solving, and self-advocacy (Schmidt et al., 2020). Participants were encouraged to monitor their symptoms daily and to complete a new, 10-minute learning module after tracking their symptoms. However, they could track their symptoms without reviewing additional learning modules. Parents were provided their own log in to view the information. Participants received prompts every few days to interact with the program. Number of times that the participant accessed the website and completed the symptom monitoring or learning modules and time spent on the website provided measures of adherence. However, the latter yielded unreliable information since some participants failed to log out of the website between visits (Schmidt et al., 2022).
Measures
Demographic and Injury Characteristics
Caregivers completed a questionnaire regarding their own and their child’s age, race, and ethnicity. Caregivers also provided information about their highest level of education, family income, and medical insurance coverage. Information regarding mechanism of injury, Glasgow Coma Scale (GCS) score, and time since injury was extracted from the medical chart.
Outcomes
Coping
Participants completed the 16-item Coping Strategies Inventory-short form (CSI-SF; Addison et al., 2007) at baseline and each of the four follow-up assessments to assess coping following the mTBI. The CSI-SF was selected because it has been used in prior research on pediatric mTBI (Woodrome et al., 2011) and captures empirically supported dimensions of coping that are relevant to the theoretical framework used to construct the SMART program. The CSI-SF consists of four subscales measuring PFE (e.g., I tackle the problem head on), problem-focused disengagement (PFD; e.g., I try not to think about the problem), emotion-focused engagement (EFE; e.g., I try to let my emotions out), and emotion-focused disengagement (EFD; e.g., I keep my thoughts and feelings to myself). Respondents rated items on a 5-point scale reflecting how often they use that strategy ranging from 1 = never to 5 = almost always. Based on the prior research, we examined PFE as both an outcome and moderator of QoL and symptoms. Cronbach’s α for PFE in the current sample ranged from 0.70 at week 1 to 0.84 at week 4 (average = 0.78). Given the SMART program’s emphasis on return to activities and actively addressing symptoms using relaxation and metacognitive strategies, we hypothesized significant increases in PFE in the SMART versus UC group over time. However, given that CSI scores did not change significantly, we examined baseline PFE scores as a potential moderator of treatment efficacy, based on prior research linking it to better recovery.
Quality of Life
Youth completed the 23-item, generic core scales of the Pediatric Quality of Life Inventory (PedsQL), a widely used measure with established reliability and validity across a range of demographic groups and conditions (Varni et al., 2003). In the current sample, Cronbach’s α was >0.90 at all timepoints. Higher scores reflect better QoL.
Brain Injury Symptoms
Youth completed the 20-item Health and Behavior Inventory (HBI) as index of cognitive and somatic post-concussion symptoms (Ayr et al., 2009). The occurrence of common TBI symptoms is rated on a 4-point Likert-type scale with responses ranging from never to often. The HBI generates two scores reflecting cognitive and somatic symptoms that possess high internal consistency (Cronbach's α in the current sample 0.88–0.96). Time until recovery was defined by the week (1–4) at which cognitive and somatic symptoms returned to pre-injury levels.
Additional Moderators
Resilience
Resilience was measured using the 10-item version of the Connor-Davidson Resilience Scale (CD-RISC; Campbell-Sills & Stein, 2007; Connor & Davidson, 2003) administered during the baseline visit. The CD-RISC measures dimensions believed to be associated with resilience including personal competence, tenacity, tolerance of negative affect, and positive acceptance of change. Respondents rated themselves on each item using a 5-point scale (0 = not true at all, 2 = sometimes true, 4 = true nearly all the time). Scores range from 0 to 40 with higher scores reflecting greater resilience. The CD-RISC has demonstrated satisfactory reliability and validity in the general population (Connor & Davidson, 2003) and in children with mTBI (Durish et al., 2018). In the current sample, Cronbach’s α was 0.89. Raw scores for the CD-RISC were examined in moderation analyses.
Power
A sample size of 100 participants (50 per group) was estimated to provide at least 91% power to detect a group difference in the primary outcome (PedsQL). Due to decreases in patient volumes as a result of the novel SARS-CoV-2 pandemic that began in February 2020, the desired sample size was not achieved.
Statistical Analyses
Differences in demographics and injury characteristics between the SMART and UC groups and completers and dropouts were assessed using t-tests (age, time since injury) and χ2 tests (sex, race, ethnicity, insurance, maternal education, family income, mechanism of injury, GCS score). Limited demographics for those who met eligibility criteria and declined participation were compared to those enrolled. Group differences over time on the CSI and PedsQL were examined using linear mixed-effect models including group × time as fixed effect and subject-specific random effect to account for repeated measures on the same subject. Evaluation of model assumptions via residual plots from the mixed-effect models supported utilization of a normal distribution. To compare the groups on the time until returning to pre-injury symptom levels, we used Kaplan–Meier curves and log-rank tests analysis; when a significant group difference was observed in the Kaplan–Meier curves, post hoc analysis using Fisher’s exact tests was conducted to compare the proportions recovered at each visit.
To examine the moderating role of PFE and resilience, and to account for correlation of outcomes measured on the same subjects over time, we conducted separate linear mixed-model analyses with PFE and resilience on the association between treatment group and QoL ratings over time. We also examined child birth sex as a potential moderator. Demographic and injury characteristics that differed significantly between the SMART and UC groups were adjusted as covariates in all regression models. For models with a significant three-way interaction, post hoc analyses were conducted to evaluate the association of the moderator variable (i.e., PFE or resilience) at low and high ratings (10th and 90th percentiles, respectively). In the presence of dropout and missing data, the linear mixed models allowed us to utilize the partially observed data through likelihood-based methods rather than excluding those with missing data at any of the follow-up assessments. To examine resilience and coping as moderators of the effect of SMART on symptom recovery, Cox proportional hazards models including the interaction of resilience/coping with group were used to model the time until returning to pre-injury levels for adolescent-reported cognitive and somatic symptom scores. For models with a significant interaction of the moderator with group, post hoc analyses were conducted to compare the survival curves between the two groups by fixing the corresponding moderator at low and high levels (10th and 90th percentiles, respectively). We calculated effect sizes (ES) for post hoc contrasts using standardized regression in which we standardized all continuous variables (M = 0, SD = 1) so that the contrasts were akin to standardized mean difference (i.e., Cohen’s d). We did not correct for multiple comparisons. All analyses were conducted in SAS 9.4 (SAS Institute, NC).
Results
Seventy-one youth-parent/guardian dyads provided informed assent/consent and completed baseline assessments. Patients who declined participation were similar in age, sex, and race as to those who consented (see Supplementary Figure 1). Equal numbers of participants were randomized to SMART (n = 35) and UC (n = 36). As reported in Table I, there were racial differences between treatment groups with 31 (86.1%) of participants in the control group being White compared to 23 (65.7%) in the SMART group (χ2 = 4.05, p = .044). Participants in the SMART group also had significantly longer time from injury to ED visit than the UC group (2.37 vs. 0.89 days, t = −2.87, p = .006). There were no other significant differences on other key injury or demographic characteristics, including age, sex, measures of SES, or mechanism of injury. Participants who completed one or more follow-up assessments (n = 66) were included in the linear models. Of those, 59 (83%) completed the final 4-week assessment. Dropout rates did not vary by the treatment group, with 32 (88%) in the UC group completing the final assessment compared to 27 (77.1%) in SMART. The only group by completion difference was found in the SMART group in which all the dropouts (n = 8) were White compared to 15 (56%) completers (χ2 = 5.41, p = .020). No adverse effects, based on youth and parent report, were reported in either group.
. | Total, N = 71 . | Treatment group . | Prob . | |
---|---|---|---|---|
UC, n = 36 . | SMART, n = 35 . | |||
Current age, mean (SD) | 14.2 (1.8) | 14.2 (1.8) | 14.2 (1.9) | .937 |
Grade, mean (SD) | 6.0 (1.9) | 5.9 (2.0) | 6.0 (1.8) | .855 |
Sex male, n (%) | 37 (52.1) | 18 (50) | 19 (54.3) | .718 |
Race,an (%) | .044 | |||
White | 54 (76.1) | 31 (86.1) | 23 (65.7) | |
Black/African American | 17 (23.9) | 5 (29.4) | 12 (34.3) | |
Ethnicity: Hispanic, n (%) | 2 (2.8) | 1 (2.7) | 1 (2.9) | .984 |
Maternal education: > college degree, n (%) | 34 (47.9) | 17 (47.2) | 19 (48.6) | .906 |
Income: above $50,000, n (%) | 46 (64.8) | 23 (63.9) | 23 (65.7) | .726 |
Insurance: private, n (%) | 48 (67.6) | 25 (69.4) | 23 (65.7) | .34 |
Mechanism of injuryb | ||||
Sports/recreational activity | 37 (52.1) | 17 (47.2) | 20 (57.1) | |
Motor vehicle collision occupant | 4 (5.6) | 2 (5.6) | 2 (5.7) | |
Motorized wheeled transport rider | 2 (2.8) | 1 (2.8) | 1 (2.9) | |
Fall to ground, non-elevated | 14 (19.7) | 9 (25.0) | 5 (14.3) | |
Walked or ran into a stationary object | 5 (7) | 1 (2.8) | 4 (11.4) | |
Assault | 5 (7) | 4 (11.1) | 1 (2.9) | |
Object struck head—accidental | 1 (1.4) | 1 (2.8) | 0 (0.0) | |
Other mechanism | 3 (4.2) | 1 (2.8) | 2 (5.7) | |
Use of other treatmentsc by week | ||||
Week 1 | 39 (54.9) | 19 (52.8) | 20 (57.1) | |
Week 2 | 20 (28.2) | 11 (30.5) | 9 (25.7) | |
Week 3 | 6 (14.1) | 6 (16.7) | 4 (11.4) | |
Week 4 | 6 (16.9) | 6 (16.7) | 6 (17.1) | |
Days from injury to ED visit | 1.6 (2.3) | 0.9 (1.6) | 2.4 (2.6) | .006 |
. | Total, N = 71 . | Treatment group . | Prob . | |
---|---|---|---|---|
UC, n = 36 . | SMART, n = 35 . | |||
Current age, mean (SD) | 14.2 (1.8) | 14.2 (1.8) | 14.2 (1.9) | .937 |
Grade, mean (SD) | 6.0 (1.9) | 5.9 (2.0) | 6.0 (1.8) | .855 |
Sex male, n (%) | 37 (52.1) | 18 (50) | 19 (54.3) | .718 |
Race,an (%) | .044 | |||
White | 54 (76.1) | 31 (86.1) | 23 (65.7) | |
Black/African American | 17 (23.9) | 5 (29.4) | 12 (34.3) | |
Ethnicity: Hispanic, n (%) | 2 (2.8) | 1 (2.7) | 1 (2.9) | .984 |
Maternal education: > college degree, n (%) | 34 (47.9) | 17 (47.2) | 19 (48.6) | .906 |
Income: above $50,000, n (%) | 46 (64.8) | 23 (63.9) | 23 (65.7) | .726 |
Insurance: private, n (%) | 48 (67.6) | 25 (69.4) | 23 (65.7) | .34 |
Mechanism of injuryb | ||||
Sports/recreational activity | 37 (52.1) | 17 (47.2) | 20 (57.1) | |
Motor vehicle collision occupant | 4 (5.6) | 2 (5.6) | 2 (5.7) | |
Motorized wheeled transport rider | 2 (2.8) | 1 (2.8) | 1 (2.9) | |
Fall to ground, non-elevated | 14 (19.7) | 9 (25.0) | 5 (14.3) | |
Walked or ran into a stationary object | 5 (7) | 1 (2.8) | 4 (11.4) | |
Assault | 5 (7) | 4 (11.1) | 1 (2.9) | |
Object struck head—accidental | 1 (1.4) | 1 (2.8) | 0 (0.0) | |
Other mechanism | 3 (4.2) | 1 (2.8) | 2 (5.7) | |
Use of other treatmentsc by week | ||||
Week 1 | 39 (54.9) | 19 (52.8) | 20 (57.1) | |
Week 2 | 20 (28.2) | 11 (30.5) | 9 (25.7) | |
Week 3 | 6 (14.1) | 6 (16.7) | 4 (11.4) | |
Week 4 | 6 (16.9) | 6 (16.7) | 6 (17.1) | |
Days from injury to ED visit | 1.6 (2.3) | 0.9 (1.6) | 2.4 (2.6) | .006 |
Note. ED = emergency department; SD = standard deviation; SMART = Self-Management After Recent mild Traumatic brain injury; UC = usual care.
Races other than White and Black/African American were not reported.
No patients reported pedestrian/bicyclist struck or fall from elevation.
Treatments included visits to any of the following general practitioner, concussion clinic, sports medicine, neurologist, physical and rehabilitative medicine, trauma/orthopedics, psychiatrist/psychologist, physical therapist, athletic trainer, and emergency department.
. | Total, N = 71 . | Treatment group . | Prob . | |
---|---|---|---|---|
UC, n = 36 . | SMART, n = 35 . | |||
Current age, mean (SD) | 14.2 (1.8) | 14.2 (1.8) | 14.2 (1.9) | .937 |
Grade, mean (SD) | 6.0 (1.9) | 5.9 (2.0) | 6.0 (1.8) | .855 |
Sex male, n (%) | 37 (52.1) | 18 (50) | 19 (54.3) | .718 |
Race,an (%) | .044 | |||
White | 54 (76.1) | 31 (86.1) | 23 (65.7) | |
Black/African American | 17 (23.9) | 5 (29.4) | 12 (34.3) | |
Ethnicity: Hispanic, n (%) | 2 (2.8) | 1 (2.7) | 1 (2.9) | .984 |
Maternal education: > college degree, n (%) | 34 (47.9) | 17 (47.2) | 19 (48.6) | .906 |
Income: above $50,000, n (%) | 46 (64.8) | 23 (63.9) | 23 (65.7) | .726 |
Insurance: private, n (%) | 48 (67.6) | 25 (69.4) | 23 (65.7) | .34 |
Mechanism of injuryb | ||||
Sports/recreational activity | 37 (52.1) | 17 (47.2) | 20 (57.1) | |
Motor vehicle collision occupant | 4 (5.6) | 2 (5.6) | 2 (5.7) | |
Motorized wheeled transport rider | 2 (2.8) | 1 (2.8) | 1 (2.9) | |
Fall to ground, non-elevated | 14 (19.7) | 9 (25.0) | 5 (14.3) | |
Walked or ran into a stationary object | 5 (7) | 1 (2.8) | 4 (11.4) | |
Assault | 5 (7) | 4 (11.1) | 1 (2.9) | |
Object struck head—accidental | 1 (1.4) | 1 (2.8) | 0 (0.0) | |
Other mechanism | 3 (4.2) | 1 (2.8) | 2 (5.7) | |
Use of other treatmentsc by week | ||||
Week 1 | 39 (54.9) | 19 (52.8) | 20 (57.1) | |
Week 2 | 20 (28.2) | 11 (30.5) | 9 (25.7) | |
Week 3 | 6 (14.1) | 6 (16.7) | 4 (11.4) | |
Week 4 | 6 (16.9) | 6 (16.7) | 6 (17.1) | |
Days from injury to ED visit | 1.6 (2.3) | 0.9 (1.6) | 2.4 (2.6) | .006 |
. | Total, N = 71 . | Treatment group . | Prob . | |
---|---|---|---|---|
UC, n = 36 . | SMART, n = 35 . | |||
Current age, mean (SD) | 14.2 (1.8) | 14.2 (1.8) | 14.2 (1.9) | .937 |
Grade, mean (SD) | 6.0 (1.9) | 5.9 (2.0) | 6.0 (1.8) | .855 |
Sex male, n (%) | 37 (52.1) | 18 (50) | 19 (54.3) | .718 |
Race,an (%) | .044 | |||
White | 54 (76.1) | 31 (86.1) | 23 (65.7) | |
Black/African American | 17 (23.9) | 5 (29.4) | 12 (34.3) | |
Ethnicity: Hispanic, n (%) | 2 (2.8) | 1 (2.7) | 1 (2.9) | .984 |
Maternal education: > college degree, n (%) | 34 (47.9) | 17 (47.2) | 19 (48.6) | .906 |
Income: above $50,000, n (%) | 46 (64.8) | 23 (63.9) | 23 (65.7) | .726 |
Insurance: private, n (%) | 48 (67.6) | 25 (69.4) | 23 (65.7) | .34 |
Mechanism of injuryb | ||||
Sports/recreational activity | 37 (52.1) | 17 (47.2) | 20 (57.1) | |
Motor vehicle collision occupant | 4 (5.6) | 2 (5.6) | 2 (5.7) | |
Motorized wheeled transport rider | 2 (2.8) | 1 (2.8) | 1 (2.9) | |
Fall to ground, non-elevated | 14 (19.7) | 9 (25.0) | 5 (14.3) | |
Walked or ran into a stationary object | 5 (7) | 1 (2.8) | 4 (11.4) | |
Assault | 5 (7) | 4 (11.1) | 1 (2.9) | |
Object struck head—accidental | 1 (1.4) | 1 (2.8) | 0 (0.0) | |
Other mechanism | 3 (4.2) | 1 (2.8) | 2 (5.7) | |
Use of other treatmentsc by week | ||||
Week 1 | 39 (54.9) | 19 (52.8) | 20 (57.1) | |
Week 2 | 20 (28.2) | 11 (30.5) | 9 (25.7) | |
Week 3 | 6 (14.1) | 6 (16.7) | 4 (11.4) | |
Week 4 | 6 (16.9) | 6 (16.7) | 6 (17.1) | |
Days from injury to ED visit | 1.6 (2.3) | 0.9 (1.6) | 2.4 (2.6) | .006 |
Note. ED = emergency department; SD = standard deviation; SMART = Self-Management After Recent mild Traumatic brain injury; UC = usual care.
Races other than White and Black/African American were not reported.
No patients reported pedestrian/bicyclist struck or fall from elevation.
Treatments included visits to any of the following general practitioner, concussion clinic, sports medicine, neurologist, physical and rehabilitative medicine, trauma/orthopedics, psychiatrist/psychologist, physical therapist, athletic trainer, and emergency department.
SMART program utilization varied widely, with the frequency of symptom monitoring ranging from 1 to 34 times (M = 9.66, SD = 9.76). Participants completed 2.58 learning modules on average (SD = 2.52; range 0–8). See Schmidt et al. (2022) for additional details.
Primary and Secondary Outcomes
Both groups had similar baseline psychosocial functioning as assessed by self-reported CSI, PedsQL, and HBI scores. Scores on the CSI subscales remained unchanged over time regardless of group (see Table II). QoL improved significantly, and this did not differ by group. Table II also reports the number and percent of participants who recovered at each time point, as defined as returning to their pre-injury score. By 4 weeks, 85% in the SMART group and 77% in the UC group reported returning to their preinjury levels of physical symptoms. Sixty-seven percent of adolescents had full cognitive recovery at 4 weeks per self-report, regardless of group.
. | . | UC, mean (SD) . | SMART, mean (SD) . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Patient . | . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . |
CSI | Emotion-focused engagement | 11.8 (3.5) | 11.1 (3.4) | 11.6 (3.6) | 12 (4.1) | 11.3 (4.4) | 12 (3.7) | 11.8 (4.3) | 11.1 (3.7) | 11.8 (3.9) | 12.2 (4) |
Problem-focused engagement | 12.8 (3.2) | 13.3 (2.6) | 13 (3.6) | 12.9 (3.8) | 12.5 (4) | 12.9 (4.0) | 13.2 (3.3) | 13.6 (3.7) | 14 (3.8) | 13.8 (3.5) | |
Emotion-focused disengagement | 11.5 (3.8) | 110 (3.3) | 11.1 (3.9) | 10.6 (3.7) | 10.1 (4.1) | 12.1 (3.8) | 12.9 (3.6) | 11.6 (3.2) | 11.7 (3.6) | 11.6 (3.2) | |
Problem-focused disengagement | 11.3 (3.6) | 11.4 (2.5) | 10.1 (3) | 10.3 (3.7) | 9.6 (3.5) | 11.7 (3.6) | 11.9 (3.1) | 11.9 (3.4) | 11.7 (3.2) | 11.4 (3.3) | |
PedsQL | Total | 79.9 (16.2) | 78.3 (14.6) | 83.8 (14) | 85.8 (14.7) | 89 (13.7) | 77.7 (16.3) | 75.7 (22.2) | 80.9 (17.2) | 84.8 (17) | 84.9 (19.6) |
. | . | UC, mean (SD) . | SMART, mean (SD) . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Patient . | . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . |
CSI | Emotion-focused engagement | 11.8 (3.5) | 11.1 (3.4) | 11.6 (3.6) | 12 (4.1) | 11.3 (4.4) | 12 (3.7) | 11.8 (4.3) | 11.1 (3.7) | 11.8 (3.9) | 12.2 (4) |
Problem-focused engagement | 12.8 (3.2) | 13.3 (2.6) | 13 (3.6) | 12.9 (3.8) | 12.5 (4) | 12.9 (4.0) | 13.2 (3.3) | 13.6 (3.7) | 14 (3.8) | 13.8 (3.5) | |
Emotion-focused disengagement | 11.5 (3.8) | 110 (3.3) | 11.1 (3.9) | 10.6 (3.7) | 10.1 (4.1) | 12.1 (3.8) | 12.9 (3.6) | 11.6 (3.2) | 11.7 (3.6) | 11.6 (3.2) | |
Problem-focused disengagement | 11.3 (3.6) | 11.4 (2.5) | 10.1 (3) | 10.3 (3.7) | 9.6 (3.5) | 11.7 (3.6) | 11.9 (3.1) | 11.9 (3.4) | 11.7 (3.2) | 11.4 (3.3) | |
PedsQL | Total | 79.9 (16.2) | 78.3 (14.6) | 83.8 (14) | 85.8 (14.7) | 89 (13.7) | 77.7 (16.3) | 75.7 (22.2) | 80.9 (17.2) | 84.8 (17) | 84.9 (19.6) |
. | Number (percent recovered) . | . | n (%) . | n (%) . | n (%) . | n (%) . | . | n (%) . | n (%) . | n (%) . | n (%) . |
---|---|---|---|---|---|---|---|---|---|---|---|
HBI | Somatic | N/A | 7 (19.44) | 14 (42.00) | 19 (59.06) | 24 (76.12) | N/A | 13 (39.92) | 18 (55.73) | 20 (62.06) | 26 (84.82) |
Cognitive | N/A | 8 (22.22) | 16 (46.47) | 18 (53.16) | 22 (66.54) | N/A | 12 (34.29) | 13 (37.41) | 17 (50.59) | 21 (67.06) | |
. | Number (percent recovered) . | . | n (%) . | n (%) . | n (%) . | n (%) . | . | n (%) . | n (%) . | n (%) . | n (%) . |
---|---|---|---|---|---|---|---|---|---|---|---|
HBI | Somatic | N/A | 7 (19.44) | 14 (42.00) | 19 (59.06) | 24 (76.12) | N/A | 13 (39.92) | 18 (55.73) | 20 (62.06) | 26 (84.82) |
Cognitive | N/A | 8 (22.22) | 16 (46.47) | 18 (53.16) | 22 (66.54) | N/A | 12 (34.29) | 13 (37.41) | 17 (50.59) | 21 (67.06) | |
Note. The groups did not differ significantly at any time point. CSI = Coping Strategies Inventory; HBI = Health Behavior Inventory; N/A = not applicable; PedsQL = Pediatric Quality of Life Inventory; SD = standard deviation; SMART = Self-Management After Recent mild Traumatic brain injury; UC = usual care.
. | . | UC, mean (SD) . | SMART, mean (SD) . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Patient . | . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . |
CSI | Emotion-focused engagement | 11.8 (3.5) | 11.1 (3.4) | 11.6 (3.6) | 12 (4.1) | 11.3 (4.4) | 12 (3.7) | 11.8 (4.3) | 11.1 (3.7) | 11.8 (3.9) | 12.2 (4) |
Problem-focused engagement | 12.8 (3.2) | 13.3 (2.6) | 13 (3.6) | 12.9 (3.8) | 12.5 (4) | 12.9 (4.0) | 13.2 (3.3) | 13.6 (3.7) | 14 (3.8) | 13.8 (3.5) | |
Emotion-focused disengagement | 11.5 (3.8) | 110 (3.3) | 11.1 (3.9) | 10.6 (3.7) | 10.1 (4.1) | 12.1 (3.8) | 12.9 (3.6) | 11.6 (3.2) | 11.7 (3.6) | 11.6 (3.2) | |
Problem-focused disengagement | 11.3 (3.6) | 11.4 (2.5) | 10.1 (3) | 10.3 (3.7) | 9.6 (3.5) | 11.7 (3.6) | 11.9 (3.1) | 11.9 (3.4) | 11.7 (3.2) | 11.4 (3.3) | |
PedsQL | Total | 79.9 (16.2) | 78.3 (14.6) | 83.8 (14) | 85.8 (14.7) | 89 (13.7) | 77.7 (16.3) | 75.7 (22.2) | 80.9 (17.2) | 84.8 (17) | 84.9 (19.6) |
. | . | UC, mean (SD) . | SMART, mean (SD) . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Patient . | . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . | Baseline . | Week 1 . | Week 2 . | Week 3 . | Week 4 . |
CSI | Emotion-focused engagement | 11.8 (3.5) | 11.1 (3.4) | 11.6 (3.6) | 12 (4.1) | 11.3 (4.4) | 12 (3.7) | 11.8 (4.3) | 11.1 (3.7) | 11.8 (3.9) | 12.2 (4) |
Problem-focused engagement | 12.8 (3.2) | 13.3 (2.6) | 13 (3.6) | 12.9 (3.8) | 12.5 (4) | 12.9 (4.0) | 13.2 (3.3) | 13.6 (3.7) | 14 (3.8) | 13.8 (3.5) | |
Emotion-focused disengagement | 11.5 (3.8) | 110 (3.3) | 11.1 (3.9) | 10.6 (3.7) | 10.1 (4.1) | 12.1 (3.8) | 12.9 (3.6) | 11.6 (3.2) | 11.7 (3.6) | 11.6 (3.2) | |
Problem-focused disengagement | 11.3 (3.6) | 11.4 (2.5) | 10.1 (3) | 10.3 (3.7) | 9.6 (3.5) | 11.7 (3.6) | 11.9 (3.1) | 11.9 (3.4) | 11.7 (3.2) | 11.4 (3.3) | |
PedsQL | Total | 79.9 (16.2) | 78.3 (14.6) | 83.8 (14) | 85.8 (14.7) | 89 (13.7) | 77.7 (16.3) | 75.7 (22.2) | 80.9 (17.2) | 84.8 (17) | 84.9 (19.6) |
. | Number (percent recovered) . | . | n (%) . | n (%) . | n (%) . | n (%) . | . | n (%) . | n (%) . | n (%) . | n (%) . |
---|---|---|---|---|---|---|---|---|---|---|---|
HBI | Somatic | N/A | 7 (19.44) | 14 (42.00) | 19 (59.06) | 24 (76.12) | N/A | 13 (39.92) | 18 (55.73) | 20 (62.06) | 26 (84.82) |
Cognitive | N/A | 8 (22.22) | 16 (46.47) | 18 (53.16) | 22 (66.54) | N/A | 12 (34.29) | 13 (37.41) | 17 (50.59) | 21 (67.06) | |
. | Number (percent recovered) . | . | n (%) . | n (%) . | n (%) . | n (%) . | . | n (%) . | n (%) . | n (%) . | n (%) . |
---|---|---|---|---|---|---|---|---|---|---|---|
HBI | Somatic | N/A | 7 (19.44) | 14 (42.00) | 19 (59.06) | 24 (76.12) | N/A | 13 (39.92) | 18 (55.73) | 20 (62.06) | 26 (84.82) |
Cognitive | N/A | 8 (22.22) | 16 (46.47) | 18 (53.16) | 22 (66.54) | N/A | 12 (34.29) | 13 (37.41) | 17 (50.59) | 21 (67.06) | |
Note. The groups did not differ significantly at any time point. CSI = Coping Strategies Inventory; HBI = Health Behavior Inventory; N/A = not applicable; PedsQL = Pediatric Quality of Life Inventory; SD = standard deviation; SMART = Self-Management After Recent mild Traumatic brain injury; UC = usual care.
Correlations among Coping and Resilience Variables
Correlations among the four coping dimensions ranged from −0.01, between PFE and EFD, to 0.63, between EFD and PFD. Correlations between coping dimensions and resilience ranged from 0.00 with PFD to 0.55 with PFE. See Supplementary Table I for all correlations.
PFE as a Moderator of Improvements in QoL
In mixed-model analyses examining self-reported QoL, we found a three-way interaction of PFE, treatment group, and time, F(4, 152) = 3.25, p = .02. As depicted in Figure 1, although the SMART and UC groups did not differ from each other at any time point at either high or low levels of PFE, the trajectory of improvement within the groups differed as a function of their levels of PFE. Specifically, adolescents in the UC group who engaged in high levels of PFE reported a significant decline in QoL at time 1 (ES = 0.60, 95% confidence [CI]: 0.03–1.18, p = .04), whereas those with high levels of PFE in SMART did not report a decline in QoL at any time point. At low levels of PFE, the SMART group reported a significant decrease in QoL from pre-injury to time 1 (ES = 0.68, 95% CI: 0.16–1.21, p = .01), whereas those receiving UC reported significant increases in QoL from time 1 to time 3 (ES = 0.56, 95% CI: 0.03–1.09, p = .04) and time 1 to time 4 (ES = 0.89, 95% CI: 0.39–1.38, p = .01).
PFE as a Moderator of Symptom Recovery
PFE also moderated group differences in recovery on the HBI cognitive subscale (Wald χ2 (1) = 5.12, p = .02). At low levels of PFE, there was a trend for the UC group to report more rapid return to pre-injury symptom levels than the SMART group (hazard ratio [HR] = 2.48, 95% CI: 0.85–7.22, p = .09). Conversely, at high levels of PFE, there was a trend for the SMART group to report more rapid return to baseline/pre-injury levels of symptoms than the UC group (HR = 2.93, 95% CI: 0.99–8.70, p = .05; see Figure 2). Similar findings were found for child-reported somatic symptoms (Wald χ2 (1) = 3.79, p = .05). At low levels of PFE, the groups did not differ on somatic symptom recovery; while at high levels of PFE, there was a trend for more rapid recovery in the SMART group (HR = 2.77, 95% CI: 0.99–7.69, p = .05).
Resilience as a Moderator of Improvements in QoL
In mixed-model analyses examining changes in QoL over time, we found a significant three-way interaction of resilience × group × time, F(4, 107) = 3.71, p = .007. At high levels of resilience, QoL did not differ significantly between the SMART and UC groups at any time point (see Supplementary Figure 3). However, at low levels of resilience, the UC group reported significantly higher QoL than the SMART group at times 1 and 4 (ES = 0.96, 95% CI: 0.08–1.84 and 1.23, 95% CI: 0.34–2.12, respectively, all ps < .05). At high levels of resilience, the UC group reported a decrease in QoL from baseline to time 1 (ES = 0.68, 95% CI: 0.20–1.16, p = .01), whereas the SMART group did not report a decrease.
Resilience as a Moderator of Symptom Recovery
Resilience moderated recovery on HBI cognitive symptoms, after controlling for time since injury (Wald χ2 (1) = 6.06, p = .013). At low levels of resilience, the UC group reported faster return to pre-injury symptom levels than the SMART group (HR = 3.79, 95% CI: 1.03–13.97, p = .04), whereas at high levels of resilience, the SMART group reported faster recovery (HR = 2.65, 95% CI: 1.00–7.04, p = .049). See Supplementary Figure 4. We found a similar pattern of findings for somatic symptom recovery (Wald χ2 (1) = 3.61, p = .057). However, group contrasts were not statistically significant.
In exploratory analyses, sex was examined as potential moderator of QoL and symptom recovery. It did not significantly moderate these outcomes, nor was it significant as a main effect.
Discussion
We report findings from a pilot RCT of an online program designed to promote symptom monitoring and active recovery. Contrary to our primary hypothesis, we found no group differences in coping strategies or overall QoL. As discussed in greater detail, failure to find group differences may be associated with limited completion of the psychoeducational modules within the SMART group, coupled with significant moderation effects. Specifically, we found evidence that PFE and resilience moderated the efficacy of the SMART intervention in terms of both QoL and the secondary outcome of symptom recovery. These findings suggest that one size may not fit all when it comes to self-guided online interventions designed to support self-monitoring and active symptom management and return to activities following mTBI in adolescents. Our findings correspond to expanding literature focusing on heterogeneity of treatment effects and identification of treatment responders (e.g., Pelham et al., 2017). Our findings suggest that those who cope by engaging with the problem are better equipped to benefit from this type of program. In a similar fashion, higher levels of resilience contributed to faster recovery in the SMART group versus UC, whereas the opposite was found at low levels of resilience. Thus, results point to the need to understand factors that contribute to intervention efficacy and to use that information to personalize treatment options and delivery.
We did not find hypothesized improvements in the nature of coping strategies used, and in fact ratings of coping strategies did not change significantly over time. As described by Schmidt et al. (2022), participants accessed between two to three learning modules on average and those most frequently viewed focused on symptom monitoring and guidelines for returning to activities, not coping. Conversely, modules explicitly providing training in coping skills such as cognitive reframing, problem-solving, and self-advocacy were viewed by relatively few participants (range 3–8). These findings suggest that coping skills may need to be more explicitly targeted in initial modules. Alternately, more sophisticated methods of tailoring the learning experience, such as personalized learning, may lead to greater engagement with the materials.
Although overall QoL improved significantly over time, we did not find group differences. It is possible that the broad nature of QoL means that it is influenced by factors other than the mTBI. Moreover, our data collection was paused due to the COVID pandemic and then reinitiated. Thus, stresses associated with the pandemic or other situational factors may have impacted QoL more than some of the scales assessing physical recovery. In addition, as noted previously, the relatively small proportion of participants who viewed modules other than Return to Activities may have limited their opportunities to learn strategies to improve their overall QoL.
Limited prior research suggests that coping focusing on actively addressing the problem is associated with better recovery following TBI. In the current study, the use of PFE moderated changes in self-reported QoL and mTBI symptom recovery over time. As depicted in Figure 1, the moderation effect of PFE on QoL was subtle and not associated with significant group differences over time. Prior research suggests that mTBI contributes to worsening QoL, and we anticipated that SMART would prevent worsening and ultimately improve QoL over the course of follow-up (Fineblit et al., 2016). Consistent with these expectations, SMART participants with high PFE did not endorse declining QoL during the initial week post-injury. Unexpectedly, however, participants in the UC group with low levels of PFE reported increasing QoL between baseline and the week 1 follow-up. It is possible that this is a spurious finding, given that UC participants with high PFE reported declining QoL over the same period. However, it is also possible that the nature of UC and environmental supports differed in these groups, contributing to initial differences in the trajectory of QoL.
Resilience also moderated the effects of group on QoL, with high-resilience SMART participants and low-resilience UC participants reporting stable or increasing QoL across 4 weeks of follow-up and high-resilience UC and low-resilience SMART participants reporting initial declines. Within the UC group, QoL in the low- and high-resilience groups became more similar over time, whereas high-resilience SMART participants continued to report better QoL than their low-resilience counterparts across follow-up. As with coping, it is possible that less resilient adolescents elicit different environmental supports and resources than highly resilient adolescents (e.g., their friends and family rally around them), contributing to improved QoL. Conversely, highly resilient adolescents may be less likely to elicit these supports and consequently struggle more in the absence of an intervention, like SMART, to facilitate proactive return to activities. Further investigation with larger samples and more detailed examination of social supports and other treatments could elucidate potential factors contributing to the unanticipated early improvement in QoL observed in the low-resilience UC group.
Consistent with our exploratory hypotheses, we found that PFE was associated with more rapid recovery in self-reported somatic symptoms, with a similar trend for faster recovery in self-reported cognitive symptoms. Adolescents who approach challenges by actively engaging with them may be uniquely suited to benefit from a program such as SMART that was designed to encourage self-monitoring of symptoms and employment of active strategies for managing them. Although we did not assess this in the current study, youth who cope by engaging with the problem may be more likely to use the strategies and recommendations presented to manage their symptoms. Conversely, for those who do not characteristically cope by trying to solve the problem, there may be a mismatch between the learning content presented in SMART and how they prefer to approach the problem. This mismatch may contribute to a dissonance regarding the utility and value of the program content and further supports the need for additional research on personalized and individualized learning approaches.
Resilience moderated the effects of the SMART intervention on mTBI symptoms in a similar fashion to PFE, which is not surprising given the robust correlation (0.55) between PFE and resilience in this cohort. For those reporting high levels of resilience, cognitive recovery was significantly faster in the SMART group, whereas the opposite was true for those reporting low levels of resilience. Importantly, this difference was not attributable to differences in accessing other treatments among the low-resilience/UC participants, meaning that it is not the case that individuals with lower levels of resilience were benefitting from alternative treatments during our follow-up period. The strategies proposed by SMART, including (a) a stepwise return to activities, (b) using breathing exercises and guided imagery to manage pain, and (c) advocating for accommodations may resonate more with highly resilient individuals, particularly in the absence of therapist or physician involvement. For those with low resilience, assuming greater control of their recovery may be challenging, potentially causing them to feel worse because they are not taking, or unable to take, recommended steps. However, because we were unable to assess behavior change in response to SMART recommendations, we cannot confirm this possibility.
The pattern of findings observed here is consistent with a recent meta-analysis of psychotherapy studies suggesting that characteristic coping styles may influence treatment effectiveness (Beutler et al., 2018). Specifically, those who coped by actively trying to address the problem and develop new behaviors benefited more from psychotherapies that promoted a similar active approach, compared to individuals who preferred to cope by understanding and exploring their feelings or emotions. Thus, our findings suggest a potential need to consider coping style and resilience when recommending treatments following mTBI.
The SMART program was designed as a self-directed mHealth app with participants required to access and implement content with minimal parent involvement and without professional guidance or support. Our findings raise the possibility that this approach may not be effective for some youth. Although program utilization was not correlated with PFE, it was correlated with EFE, and there was a trend for more frequent symptom monitoring among those reporting higher resilience (Schmidt et al., 2022). One aspect of resilience is not being easily discouraged by failure. Given that post-concussive symptoms may vary from day to day as one recovers, resilient youth may be more accepting when they learn that their symptoms have temporarily worsened, whereas non-resilient individuals may be discouraged by that feedback, particularly in the absence of parental or professional feedback and support. Future studies could serve to better match patients to treatment approaches/modalities and involve professionals or families as needed.
Taken together, our findings have potential implications for the acute management of mTBI recovery. A growing literature base, including this study, suggests that assessment of preferred coping strategies and resiliency may allow medical providers to tailor treatment recommendations. Individuals with low levels of PFE or resiliency may require more support to benefit from recommendations for active symptom management or may derive greater benefit from a different approach entirely. Additional research comparing different intervention models and adaptive trial designs could help to further elucidate these questions.
Online or app-based interventions such as SMART offer an approach that can readily be disseminated to other sites with relatively few modifications, particularly given nearly universal internet access among adolescents (Anderson & Jiang, 2018). However, disparities in access among low-income and minoritized populations still exist, with recent evidence from the Current Population Survey suggesting persistent disparities in access among low-income, Hispanic, and Black youth (Dolcini et al., 2021; Ryan, 2018). Thus, app-based interventions may need to be adapted for less-resourced, minoritized communities to promote access. Increased stakeholder engagement with members of minoritized communities using community-engaged research practices could identify strategies to mitigate barriers (see Smith et al., 2012). Co-designed adaptations might include presenting content in a graphic novel/paper format that supplemented online access (Albright & Gavigan, 2014).
The current study was limited by a relatively small and nondiverse sample, recruited from a single site. Although efforts were made to eliminate bias by using electronically administered surveys and concealing group assignment from both the participant and research coordinator until the baseline measures were complete, the coordinator was not blinded to group after randomization and may have approached prompting participants differently based on their group assignment. Although self-report measures are commonly administered soon after mTBI, our primary outcomes, the CSI and PedsQL, had not previously been validated for administration in the ED. Given the study design, we were unable to examine the unique and additive contributions of the symptom monitoring and learning module components of the SMART intervention. In addition, SMART participants, on average, spent relatively little time on the modules, completing fewer than three on average. Consequently, we are unable to ascertain whether completion of the full intervention would have had greater efficacy. We were unable to assess participant behavior and activities and, thus, we cannot know whether differences in recovery patterns within the SMART group were associated with differential incorporation of program recommendations into their day-to-day activities. For example, were more resilient and engaged participants more likely to use the recommended strategies for headache pain? Follow-up was limited to 1 month. We did not evaluate parental coping and resilience and it was not possible to ascertain the influence of these factors on program utilization or treatment response. A small subset of participants was enrolled during the COVID-19 pandemic and it is difficult to know how that affected their participation and recovery.
In summary, our findings suggest that the SMART program may be associated with more rapid recovery following mTBI among youth with higher levels of problem-focused, engaged coping and resilience. The results highlight the potential need to individualize approaches to post-concussion management as well as the need for further investigation of the SMART program itself. Future research could examine whether messaging that is tailored to one’s coping style coupled with therapist involvement could improve the utility of SMART for less resilient or engaged youth. In addition, further research is needed to understand and address the potential effects of internalized stigma on program engagement and adherence (see Bradley et al., 2020).
Supplementary Data
Supplementary data can be found at: https://academic.oup.com/jpepsy.
Funding
This research was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R21HD087844).
Conflicts of interest: None declared.