Article Text

Download PDFPDF
Is early activity resumption after paediatric concussion safe and does it reduce symptom burden at 2 weeks post injury? The Pediatric Concussion Assessment of Rest and Exertion (PedCARE) multicentre randomised clinical trial
  1. Andrée-Anne Ledoux1,2,
  2. Nick Barrowman3,4,
  3. Vid Bijelić3,
  4. Michael M Borghese5,
  5. Adrienne Davis6,
  6. Sarah Reid1,7,
  7. Gurinder Sangha8,
  8. Keith Owen Yeates9,10,
  9. Mark S Tremblay1,4,
  10. Candice McGahern1,
  11. Kevin Belanger1,
  12. Joel D Barnes1,
  13. Ken J Farion1,7,
  14. Carol A DeMatteo11,
  15. Nick Reed12,
  16. Roger Zemek1,7
  17. PERC PedCARE Concussion team
    1. 1 Children's Hospital of Eastern Ontario Research Institute, Ottawa, Ontario, Canada
    2. 2 Cellular and Molecular Medicine- Neuroscience, University of Ottawa, Ottawa, Ontario, Canada
    3. 3 Clinical Research Unit, Children's Hospital of Eastern Ontario Research Institute, Ottawa, Ontario, Canada
    4. 4 Department of Pediatrics, University of Ottawa, Ottawa, Ontario, Canada
    5. 5 Environmental Health Science and Research Bureau, Health Canada, Ottawa, Ontario, Canada
    6. 6 Division of Pediatric Emergency Medicine, Department of Pediatrics, Hospital for Sick Children and University of Toronto, Toronto, Ontario, Canada
    7. 7 Division of Emergency Medicine, Department of Pediatrics, University of Ottawa, Ottawa, Ontario, Canada
    8. 8 Division of Emergency Medicine, Department of Pediatrics, Western University, London, Ontario, Canada
    9. 9 Department of Psychology, University of Calgary, Calgary, Alberta, Canada
    10. 10 Alberta Children’s Hospital Research Institute and Hotchkiss Brain Institute, University of Calgary, Alberta, Canada
    11. 11 School of Rehabilitation Science, McMaster University, Hamilton, Ontario, Canada
    12. 12 Department of Occupational Science and Occupational Therapy, University of Toronto, Toronto, Ontario, Canada
    1. Correspondence to Dr Andrée-Anne Ledoux, Emergency Reseach Department, Children's Hospital of Eastern Ontario, Ottawa, ON K1H 8L1, Canada; aledoux{at}cheo.on.ca

    Abstract

    Objective Investigate whether resuming physical activity (PA) at 72 hours post concussion is safe and reduces symptoms at 2 weeks, compared with resting until asymptomatic.

    Methods Real-life conditions, multicentre, single-blinded randomised clinical trial, conducted in three Canadian paediatric emergency departments (ED). Children/youth aged 10–<18 years with acute concussion were recruited between March 2017 and December 2019, and randomly assigned to a 4-week stepwise return-to-PA protocol at 72 hours post concussion even if symptomatic (experimental group (EG)) or to a return-to-PA once asymptomatic protocol (control group (CG)). The primary outcome was self-reported symptoms at 2 weeks using the Health and Behaviour Inventory. Adherence was measured using accelerometers worn 24 hours/day for 14 days post injury. Adverse events (AE) (worsening of symptoms requiring unscheduled ED or primary care visit) were monitored. Multivariable intention-to-treat (ITT) and per-protocol analyses adjusting for prognostically important covariates were examined. Missing data were imputed for the ITT analysis.

    Results 456 randomised participants (EG: N=227; mean (SD) age=13.3 (2.1) years; 44.5% women; CG: N=229; mean (SD) age=13.3 (2.2) years; 43.7% women) were analysed. No AE were identified. ITT analysis showed no strong evidence of a group difference at 2 weeks (adjusted mean difference=−1.3 (95% CI:−3.6 to 1.1)). In adherent participants, initiating PA 72 hours post injury significantly reduced symptoms 2 weeks post injury, compared with rest (adjusted mean difference=−4.3 (95% CI:−8.4 to –0.2)).

    Conclusion Symptoms at 2 weeks did not differ significantly between children/youth randomised to initiate PA 72 hours post injury versus resting until asymptomatic; however, many were non-adherent to the intervention. Among adherent participants, early PA was associated with reduced symptoms at 2 weeks. Resumption of PA is safe and may be associated with milder symptoms at 2 weeks.

    Level of evidence 1b.

    Trial registration number NCT02893969.

    Registry name Pediatric Concussion Assessment of Rest and Exertion (PedCARE).

    • brain concussion
    • randomized controlled trial
    • physical activity
    • pediatrics
    • rest

    Data availability statement

    Data are available upon reasonable request. The data that support the findings of this study are available on reasonable request from the corresponding author in accordance to ethical and legal regulations regarding the sharing of information as to prevent compromising the privacy of research participants.

    Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Current concussion consensus guidelines1–3 recommend 24–48 hours rest, followed by symptom-limited activity. Once symptoms have resolved, patients may gradually resume physical activity (PA). Limited trial evidence exists, especially in younger paediatric athletes and in the general paediatric population, on whether early resumption of PA, even if symptomatic, is safe and prognostically better.

    Although rest is important in the initial 24–48 hours post concussion, protracted rest can lead to fatigue, reactive depression and physiological deconditioning.4–6 Conversely, therapeutic PA has been associated with milder symptoms,6–11 faster recovery12–15 and reduced risk of persistent post-concussion symptoms (PPCS).13 15 However, most studies investigating PA as treatment have small sample sizes,6 8 10 16 17 retrospective designs12 18 or initiated PA 1 week post injury14 or later.13 17 19 Many studies included only youth with sport-related concussion,14–17 19 who recover faster compared with the general paediatric population,20 or were conducted in controlled environments,14 16 which may not be pragmatic or generalisable. A pragmatic randomised clinical trial (RCT) is necessary to determine the efficacy and safety of resuming PA 72 hours post concussion in the general paediatric population.

    Our primary objective was to investigate if reintroduction of non-contact PA 72 hours post concussion (even if symptomatic) is safe and reduces symptoms at 2 weeks in children/youth aged 10–<18 years, compared with resting until symptom resolution.21 We hypothesised that introducing early non-contact PA in a standardised, incremental fashion 72 hours post injury would result in a lower symptom burden, compared with incremental introduction of PA once asymptomatic, and not cause adverse events (AE). Sex, age, symptoms on ED presentation and the Predicting and Preventing Postconcussive Problems in Paediatrics (5P) score (validated risk score for PPCS at 4 weeks)22 are predictors of PPCS and are associated with recovery duration.22–24 Subgroup analyses evaluated whether age, sex, initial symptom severity or 5P score moderated the association between the protocol followed and symptom burden at 2 weeks.

    Methods

    This study was a real-life condition, multicentre randomised clinical trial25 (1:1). Participants were recruited from March 2017 to December 2019 in three Paediatric Emergency Research Canada network tertiary paediatric emergency departments (ED). Patients and public contributed to the study design and conduct.26 The protocol was previously published25 and registered (ClinicalTrials.gov).

    Study population

    Patients aged 10–<18 years with acute head injury within 48-hours before ED presentation and diagnosed with concussion according to the Zurich/Berlin consensus definition were invited to participate.1 21 An adapted version of the Centers for Disease Control and Prevention (CDC) tiered framework was used to confirm concussion.27 Those with one ‘highest level of certainty’ symptom (eg, retrograde/anterograde amnesia; loss of consciousness) or two symptoms immediately or within 1 hour of injury with ‘higher level of certainty’ (eg, nausea/vomiting; headache) were eligible. Exclusion criteria were: Glasgow Coma Scale score ≤13; abnormality on brain CT or MRI; neurosurgical intervention, intubation or intensive care unit admission; multisystem injury requiring hospitalisation; severe pre-existing neurological developmental delay with communication difficulties; intoxication; absence of trauma history; previously enrolled; language barrier; or inability to complete follow-up. Participants and parents/guardians provided written informed consent and/or assent as appropriate.

    Intervention

    We compared incremental resumption of non-contact PA beginning 72 hours post injury versus incremental return to PA once asymptomatic.

    Participants randomly assigned to the experimental group (EG) were instructed to initiate progressive return-to-PA 72 hours post ED visit even if symptomatic (table 1).25 If their symptoms became intolerable (eg, making it difficult to perform the activity) during or within 30 min following PA, they were instructed to return to the last well-tolerated stage the next day and reattempt progression after 24 hours. Participants randomly assigned to the control group (CG) were instructed to rest-until-asymptomatic before initiating the progressive return-to-PA protocol (table 2). Experimental and control protocols were identical except: (1) CG participants must be asymptomatic for PA initiation, and (2) if CG participants developed symptoms during PA, they were instructed to stop, rest-until-asymptomatic and resume PA at the previous tolerated level. To ascertain the maximum exertion required for each step, both groups used the Pictorial Children’s Effort Rating Table (PCERT)28 diagram, including the prescribed level of PA intensity for each stage and a list of concussion symptoms.

    Table 1

    Intervention protocol25

    Table 2

    Control group protocol25

    The protocol25 is summarised below. Research assistants (RA) completed assessments in the ED at recruitment. Participants answered questions on demographics, concussion and migraine history and injury characteristics. Balance was assessed with the Balance Error Scoring System (BESS).29 Pre-injury and post-injury symptoms were measured using the Health and Behaviour Inventory (HBI)30 and retrospective HBI. The HBI is a 20-item self-report questionnaire using a 4-point rating scale of symptom frequency (0=not at all/never to 3=a lot/often) yielding a total score from 0 to 60, with subscores for cognitive and somatic symptoms. The scale has good construct validity, reliability, is validated for children/adolescents,30 and is recommended as an National Institutes of Health (NIH) core common data element for concussion.31 32 After baseline questionnaires were completed and a physician confirmed the concussion diagnosis, RAs randomised participants via Research Electronic Data Capture (REDCap).

    The CHEO Clinical Research Unit provided data management and retained randomisation codes. The randomisation sequence was created using R V.3.1.1 software33 and was stratified by site and sex, with a 1:1 allocation using random blocks of 4 and 6 in size. Only the RA who obtained consent and gave intervention instructions to participants was unblinded. Researchers, research staff and biostatisticians were blinded to the allocated intervention protocol. Unblinding was permissible in the event of a serious AE.

    On ED discharge, participants received standardised instructions on return-to-learn protocols, concussion education and an Actical accelerometer (Phillips Respironics), which estimated PA34 as a measure of protocol adherence. The Actical, worn on a belt around the waist at the right midaxillary line for 14 consecutive days (24 hours/day) post ED visit, collected data in 1-min epochs (excluding aquatic activities). Participants were followed for 4 weeks (online supplemental table 1). They were contacted daily by phone or email link to a REDCap survey for the first 14 days to assess PA, symptoms and bed/wake times. Participants completed follow-up surveys, including the HBI and AE at 1 week, 2 weeks and 4 weeks post injury.25

    Supplemental material

    Outcomes

    Because the greatest symptom decrease in paediatric concussion occurs between 0–2 weeks,23 the primary outcome measure was total self-reported HBI score at 2 weeks post injury.

    AEs were defined as worsening of symptoms to the extent that an unscheduled ED or primary care visit was required. Possible AEs were identified through a question asked at 1 week and 2 weeks post injury. Possible AEs were followed-up by the research coordinator and site investigator to determine if the unscheduled visit was related to the concussion/intervention.

    Statistical analysis

    Frequencies and descriptive statistics were used to summarise participant baseline characteristics for the overall sample and by group.

    Intention-to-treat analysis

    The primary outcome—HBI total score—was compared between EG and CG using a linear multivariable model. The model was adjusted for site and prognostically important ‘5P’ clinical prediction score22 covariates: sex, age, prior concussion and symptom duration, migraine history, balance error scoring (tandem stance), answering questions slowly, headache, sensitivity to noise and fatigue. Restricted cubic splines were applied to age (four knots) and tandem stance score (three knots) to allow for non-linearity.35 Homogeneity of variance and normality of distribution of residuals were assessed and no significant departure from these assumptions that would affect the model estimates was found. Predictive mean matching multiple imputation was used for missing baseline and outcome data.35 We used multiple imputation to create and analyse 200 multiple imputed data sets. Incomplete variables were imputed under fully conditional specifications, using the default settings of the aregImpute function of the Hmisc 4.4–1 package in R (Harrell).36 Individual components of the HBI as well as variables used in the primary analysis were imputed. In addition to variables used in the primary analysis. The imputation model also included baseline and historical HBI as well as HBI at 1 and 4 weeks. The model parameters were estimated in each imputed data set separately, and combined using Rubin’s rules.37

    Subgroup analyses

    Secondary analyses were performed to examine four interactions: group-x-sex; group-x-age; group-x-5P PPCS risk22; and group-x-ED HBI score. For group-x-sex and group-x-age, the primary analysis was repeated with the addition of coefficients representing the interactions. For group-x-5P, a linear multivariable analysis with the following covariates was generated: site, 5P score, group, group-x-5P score. The group-x-ED HBI interaction was assessed with a linear multivariable analysis with the following variables: site, group, ED HBI score, group-x-ED HBI score. Product terms were used to represent interactions and four knots restricted splines were used for group-x-age; group-x-5P PPCS risk; and group-x-ED HBI.

    Actical data processing

    Actical data processing procedures, consistent with Colley et al,38 were conducted using SAS V.9.4 (SAS Institute) (online supplemental methods 1). A valid measurement day was defined as ≥10 hours of accelerometer data.38 Time spent in each movement intensity was classified according to established intensity cut-offs: vigorous: ≥6500 counts/min; moderate: <6500 to ≥1500 counts/min; light: <1500 to ≥100 counts/min39; and sedentary: <100 counts/min.40

    Supplemental material

    Per-protocol analysis

    A sensitivity analysis was generated with participants who adhered to the intervention. The same analysis was performed as the main analysis, but missing outcome data were handled with case-deletion instead of imputation.

    Protocol adherence was measured using the Actical PA data and the daily symptomatic self-report questionnaires for days 4 and 7. To be adherent, CG participants had to be asymptomatic for 24-hours before starting the protocol and to be asymptomatic on days 3 and 6 to do PA on days 4 or 7, or abstain from PA if they were symptomatic on those days; EG participants had to rest for 72 hours and be progressing through the stages unless symptoms were intolerable during/after PA. To assess adherence on days 4 and 7, participants required one valid Actical day between days 1 and 3 and between days 5 and 6, and to complete at least one daily symptom-log between days 1 and 3 and days 5 and 6. Missing data were imputed by carrying-over previous days’ Actical and logbook data. For example, for participants missing day 3 logbooks, data from day 1 and 2 were carried over to day 3; those missing all days 1–3 were considered non-adherent. Depending on participants’ group instructions and symptoms the preceding days, their current intervention stage and level of PA compliance were determined.

    To assess adherence with movement behaviour, stages 1 and 2 were combined, and stages 3 to 5 were combined (tables 1 and 2). Table 3 specifies PA for each stage. These levels were defined a priori based on the prescribed intensity of activity (per PCERT intensity scale) for each intervention stage.

    Table 3

    Physical activity levels for per-protocol analysis

    Reliable change

    We computed two reliable change z-scores to determine if per-protocol (PP) results were clinically meaningful. First, we examined whether post-treatment symptom changes from pre-injury symptoms were reliably higher than expected when compared with an orthopaedic injured normative group. Z-scores ≥1.65 indicate a significantly greater-than-expected increase in symptoms. Second, we explored whether post-treatment symptom changes relative to pre-treatment symptoms were reliably lower or higher when compared with a normative concussion group (online supplemental methods 2).41 Z-scores ≤−1.65 indicated an unusually large decline in symptoms compared with a typical recovery trajectory; z scores ≥1.65, a smaller-than-expected decline. We did not include confounders to the analyses because z-scores took pre-injury/pre-treatment symptoms into account.

    Supplemental material

    For all analyses, main effect results and interaction terms, two-sided p values <0.05 were considered statistically significant.

    Based on unpublished data on children with concussions drawn from two source studies,42 43 1 month after the injury, the mean HBI was 15.85 and the SD was 9.83. We anticipated a 20% decrease in HBI total score at 2 weeks post injury (HBI mean 12.68) among the EG, and HBI mean 15.85 among the CG. We assumed the SD would be 9.83 in both groups. Fixing the probability of type I error at 5%, a sample size of 152 per group would give 80% power to detect this difference in mean HBI total scores between groups. Accounting for a 15% loss in follow-up, a sample size of 175 per group was required.

    Results

    Characteristics

    Of 799 eligible children/youth, 456 participants were randomised (EG n=227, 44.5% women, mean (SD) age=13.3 (2.1); CG n=229, 43.7% women, mean (SD) age=13.3 (2.2)); 343 declined consent (figure 1). Of those randomised, 353 (EG n=180/227 (79.2%); CG n=173/229 (75.5%)) had complete follow-up at 2 weeks post injury. Median (IQR) hours from injury to ED arrival were EG=3.3 (1.6-22.0) and CG=3.0 (1.7-17.6). The most common injury mechanism was sport-related (EG=114 (50.4%); CG=124 (54.4%)). Demographics were similar between groups, except previous concussions (EG=65/227 (28.6%); CG=85/229 (37.1%)) and answers questions slowly (EG=110/227 (48.5%); CG=94/229 (41.0%)) (table 4).

    Table 4

    Personal characteristics

    Figure 1

    Consolidated Standards of Reporting Trials flow diagram. ED, emergency department; GP, general practitioner; HBI, Health and Behaviour Inventory; ITT, intention-to-treat; MD, doctor of medicine; RA, research assistant.

    Withdrawn, lost to follow-up

    Twenty-four participants (EG: 12 and CG: 12) withdrew immediately after randomisation; 17 initiated the protocol (EG: 9 and CG: 8), but withdrew (figure 1). A higher proportion of the CG (36, 17%) than the EG (26, 12%) were lost to follow-up.

    Physical activity

    Of the 456 participants, 308 had minimum 4 days of valid accelerometer data.44 Figure 2 displays median, 90th and 10th percentile of daily movement behaviours between groups for light PA, moderate-to-vigorous PA and sedentary behaviour. PA levels were similar for both groups.

    Figure 2

    Minutes spent in each movement behaviour between rest-until-asymptomatic (control group, dark grey) and early return-to-physical activity (experimental group, light grey). Box shows 25th and 75th percentiles. White dot shows median. Whiskers extend to the 10th and 90th percentiles.

    Intention-to-treat analysis (n=456)

    The EG had a mean HBI total symptom load at 2 weeks of 14.9 (SD=11.9), which was lower than the mean of 16.4 (SD=11.9) in the CG; however, the difference was not statistically significant and the CI did not provide strong evidence for an effect of the intervention (mean difference=−1.3 (95% CI: −3.6 to 1.1], p=0.30); online supplemental table 2) contains multivariable model results.

    Supplemental material

    Sixteen participants (EG=7 and CG=9) had unscheduled visits unrelated to the intervention or concussion; none resulted in deviation from their randomised protocol. No AEs were recorded.

    Subgroup analysis

    No evidence of an interactions emerged for group-x-sex (p=0.36), group-x-age (p=0.36), groups-x-presentation 5P PPCS score (p=0.79) or group-x-HBI total score on ED presentation (p=0.86). However, due to large CIs, firm conclusions about interactions cannot be drawn (online supplemental table 3).

    Supplemental material

    Per-protocol analysis

    A total of 259/456 (58%) participants had valid Actical data; EG=77/113 (68%) and CG=89/146 (60%) were adherent to protocol. Demographics were similar between both adherent groups and between adherent and non-adherent groups (online supplemental table 4), except for migraine history (adherent EG=0 (0.0%); adherent CG=8 (9.0%)). Among the non-adherent, EG=34 and CG=15 (asymptomatic) failed to comply with resumption of 12 min of moderate PA at day 4 (stage 1 of intervention). Of the 166 adherent participants, n=143 (EG=67 and CG=76) had complete primary outcome data. PP analyses indicated the EG reported significantly fewer symptoms than the CG (mean difference=−4.3 (95% CI: −8.4 to −0.2); p=0.038) at 2 weeks post injury. Reliable change analyses found evidence that that the EG were less likely than the CG to have increased symptoms (EG=6%, CG=15.8%; group difference=−9.8 (95% CI: −20.3 to 0.8); p=0.11) or delayed recovery (EG=4.5%, CG=12% (group difference=−7.5 (95% CI: −17.2 to 2.1); p=0.14) (table 5).

    Supplemental material

    Table 5

    Reliable change

    Discussion

    In this study, children/youth with acute concussion instructed to return-to-PA 72 hours post injury (even if symptomatic) did not differ significantly in symptom levels at 2 weeks, compared with those instructed to rest-until-asymptomatic. No AE were identified. About half of those with valid Actical data adhered to the protocol. In those who adhered, early PA resumption was associated with lower self-reported symptoms at 2 weeks. For children/youth who were more symptomatic or more at risk of developing PPCS,22 PA did not increase their symptoms at 2 weeks.

    Our results contribute to evidence that early PA is not harmful, does not increase post-concussion symptoms and does not precipitate AE. A large cohort study demonstrated that resumption of PA within 7 days post injury reduced the risk of PPCS at 4 weeks.13 Similarly, a clinical trial in adolescents with sport-related concussion showed that controlled resumption of PA within 7 days post concussion promoted recovery and potentially reduced PPCS.14 Several papers reported associations between resumption of PA and shorter recovery duration.6 11–14 17

    Despite distinct standardised written and pictorial instructions, both groups demonstrated similar PA, regardless of group randomisation and symptoms. Many participants randomised to rest-until-asymptomatic (CG) engaged in PA despite endorsing symptoms; this may reflect recent concussion recommendations permitting PA.1–3 It is also possible that athletes or those who felt better were motivated to return to PA. Conversely, many participants randomised to EG did not engage in at least 12 min of moderate PA on day 4. This was not attributed to intolerable symptoms. Reasons for non-compliance are unknown; some participants may not have been psychologically ready for PA, or it may reflect overall inactivity in children/adolescents.45 Low compliance indicates that concussion patients may self-regulate PA regardless of instructions, increasing the importance of RCTs using real-life conditions. Because measured movement was similar between groups, we conducted a robust PP analysis. In protocol adherent participants, we observed a significant group difference in symptom burden at 2 weeks favouring early PA. Early PA resumption may protect against persistent symptoms and delayed recovery. However, while we can establish associations based on PP analysis, we cannot draw causal inferences. Methods to promote adherence to prescribed PA, such as weekly contact with participants or supervision, must be considered in future research/clinical guideline implementation.

    This trial has several strengths. It is the first multicentre, real-life condition RCT to demonstrate that PA can be resumed safely 72 hours post concussion. The intervention did not require participants to visit a subspecialty clinic, but rather, empowered self-management at home, which is more generalisable and feasible. However, the effectiveness of this early intervention does not preclude the possibility of even greater gains if children/youth, especially those at risk of PPCS, are also seen in a subspecialty clinic. We followed participants for 14 consecutive days, and measured compliance through device-based monitoring; participants were blinded to device data. The study was not limited to sport-related concussion, suggesting PA can be resumed safely by the general paediatric population (10–18 years) with varying injury mechanisms. Finally, our cohort was recruited from EDs, thereby limiting the interval between injury and study enrolment. By recruiting children within 48-hours of injury (most enrolled ≤3 hours), we reduced heterogeneity in participants’ clinical status when starting the protocol.

    The study has limitations. First, no objective biomarker is available to diagnose concussions. However, our selection criteria were more restrictive (higher/highest probability of a concussion27 than previous studies, and thus, likely more accurate for diagnosis. Also, ED physicians confirmed the diagnosis before participants were randomised to the protocol. Second, because of withdrawal or loss to follow-up, 23% of participants did not complete the primary outcome; we mitigated this limitation through multiple imputation. We determined that completing an intention-to-treat (ITT) with imputation, and a PP without imputation, as pre-specified by our protocol, was the best option for our data, demonstrating a clear picture about returning to PA after a concussion. However, the ITT and PP estimates are subject to measurement bias and selection bias, respectively, and additional G-estimation or instrumental variable analysis would be required to obtain unbiased estimates. Further, the validity of the ITT analysis results depends on the correct specification of the multiple imputation model.

    Third, we cannot generalise these results to younger children; nonetheless, Grool et al 13 demonstrated that early PA is beneficial for children aged 5–9 years. Fourth, ED recruitment might create a sampling bias toward more severe concussion cases. Fifth, recruitment occurred as the new Berlin consensus statement endorsing early PA was released; this may explain why some CG participants were less compliant with their protocol.1 Further, in recent years, post-concussion PA has received favourable media attention, and guidelines outlining the benefits of early return to PA have emerged. These influences may have led CG participants to initiate PA while symptomatic (some as soon as 24 hours post injury). To address this concern, we conducted a robust PP analysis. Finally, a large proportion of participants did not complete the daily logbook or wear the Actical. Future research should examine the type, intensity, duration and total volume of PA associated with clinical outcomes; associations between PA and different symptom clusters (cognitive and somatic); and compliance factors.

    Conclusion

    PA is a promising concussion treatment. Post-concussion symptoms at 2 weeks did not differ significantly between children/youth with acute concussion randomised to initiate PA at 72 hours post injury versus those instructed to rest-until-asymptomatic. However, those with valid accelerometer data had low protocol adherence. Among participants who adhered, early PA resumption was associated with reduced symptoms at 2 weeks. Early resumption of non-contact PA, even if symptomatic, is safe and may be associated with milder symptoms at 2 weeks.

    What are the findings?

    • Our primary analysis found that among children/youth with acute concussion, early resumption of physical activity (PA) at 72 hours post injury did not reduce symptoms at 2 weeks, when compared with rest until asymptomatic. Self-adherence to the prescribed protocols was poor. Children/youth who did adhere to the protocol and undertook early resumption of PA had reduced concussion symptoms at 2 weeks.

    How might it impact on clinical practice in the future?

    • Early resumption of progressive, non-contact PA at 72 hours post injury in children/youth with concussion, even if symptomatic, could be clinically recommended, as it is safe and may be associated with lower symptom burden at 2 weeks in those adherent to treatment. Methods to promote adherence to prescribed PA, such as supervision, must be considered in future research and clinical guidelines.

    Data availability statement

    Data are available upon reasonable request. The data that support the findings of this study are available on reasonable request from the corresponding author in accordance to ethical and legal regulations regarding the sharing of information as to prevent compromising the privacy of research participants.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Ethics approval was obtained from CHEO Research Ethical Board (16/80X).

    Acknowledgments

    The authors would like to thank and acknowledge the research coordinators and research assistants across the three sites responsible for patient recruitment, enrolment and follow-up. Student volunteers at all three sites provided assistance in patient screening at the emergency department (ED). The authors appreciate the collaboration and assistance of all the treating physicians of the EDs across the sites.

    References

    Footnotes

    • Collaborators PERC PedCARE Concussion team: Candice McGahern, BA; Tyrus Crawford, BA; Lauren Dawson, MSc, RPN; Katie Healey, BA; Sarah Reid, MD; Ken J Farion, MD; Andrée-Anne Ledoux, PhD; Roger Zemek, MD; Gurinder Sangha, MD; Adrienne Davis, MD; Nicolas Reed, PhD; Carol DeMatteo, MSc; Keith O Yeates, PhD; Mark Tremblay, PhD; Michael Borghese, PhD.

    • Contributors RZ and A-AL had full access to all of the data (except were blinded to the groups) in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Design and conduct of the study: A-AL, MST, CAD, KY, KF, SR, NR and RZ. Collection, management, analysis and interpretation of the data: A-AL, NB, VB, MMB, AD, SR, KF, GS, MST, KB, JB, KY and RZ. Preparation, review or approval of the manuscript: A-AL, NB, VB, MMB, AD, SR, KF, GS, MST, KY, KB, JB, CAD, NR and RZ. Acting guarantor: A-AL and RZ. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

    • Funding This study was supported by OSSU IMPACT Award.

    • Disclaimer The funders acknowledged herein had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript; or decision to submit the manuscript for publication.

    • Competing interests KY reported being one of the authors of the Health and Behaviour Inventory (HBI); the HBI is freely available and he receives no financial benefit for its use. There are no other conflicts of interest relevant to this article to disclose. RZ is scientific director and minority shareholder in 360 Concussion Care, an interdisciplinary concussion clinic. RZ and KY are investigators in a multicentre study funded by the National Football League (NFL) Scientific Advisory Board; they do not receive any financial benefit. There are no other conflicts of interest relevant to this article to disclose.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.