The Role of Active Rehabilitation in Concussion Management: A Systematic Review and Meta-analysis : Medicine & Science in Sports & Exercise

Journal Logo


The Role of Active Rehabilitation in Concussion Management: A Systematic Review and Meta-analysis


Author Information
Medicine & Science in Sports & Exercise 53(9):p 1835-1845, September 2021. | DOI: 10.1249/MSS.0000000000002663


Concussions are the most common type of brain injury, representing 70% to 90% of all traumatic brain injuries (TBI) (1). The annual incidence of concussions in Canadians is estimated at 600 per 100,000 inhabitants (1,2), creating a major health concern. Currently, the most agreed-upon definition of a concussive event is any biomechanical injury that causes transient neurological dysfunction (3). A concussion is often considered a subset of mild TBI (mTBI); however, these terms are frequently used interchangeably (4).

The clinical presentation of concussions varies between individuals and between individual injuries (5,6). Nevertheless, the most common postconcussive symptoms (PCS) can be classified into four categories; 1) physical (e.g., headache, dizziness, and fatigue), 2) cognitive (e.g., memory and concentration), 3) sleep (e.g., difficulty falling asleep), or 4) emotional (e.g., irritability, anger, and depression) (5,7). Concussions have historically been considered a transient injury with PCS often resolving spontaneously within 2 wk of injury (7). However, up to 10% to 33% of individuals suffer from persistent postconcussion symptoms (PPCS) beyond 1 to 3 months after injury (8).

Traditionally, consensus-based recommendations have emphasized strict physical and cognitive rest after a concussion until symptom resolution (9). Upon symptom resolution and clearance by a qualified health care professional, individuals then began a stepwise return to play protocol that introduces physical activity as tolerated by the patient (10). Rest recommendations were primarily motivated by animal research that provided evidence of a vulnerability period early after a concussion in which the brain is more susceptible to repeat injury (11,12). In addition, animal models suggest that concussive events result in a neurometabolic response involving ionic fluxes, impaired neurotransmission, and increased energy consumption for neural repair and return to homeostasis (13). It was therefore theorized that placing high-energy demands on the system through excessive physical activity during this period could compromise restorative events and consequently impair recovery (14).

However, recent evidence from humans with concussion has shown that strict cognitive and physical rest has no beneficial effect on symptom severity when compared with no rest (15). In fact, research has shown that prolonged rest can slow recovery and precipitate secondary symptoms of fatigue, reactive depression, anxiety, and physical deconditioning (10,13,14,16,17). As such, the efficacy of strict rest has been challenged in recent years (12,14,18), and a gradual increase in low-level activities has been encouraged after 24–48 h of rest.

Physical activity has wide-ranging benefits, including improved physical well-being, mental well-being, and cognitive function (9,14,18), and has been established as an effective intervention for a wide variety of health-related conditions (9). Although the association between the pathophysiology of concussions and clinical symptoms remains to be elucidated, it is speculated that exercise can improve recovery of concussion symptoms through a multifaceted physiological mechanism (14). The metabolic and pathophysiological changes after a concussion result in, among other things, altered function of the autonomic nervous system, altered control of cerebral blood flow, bioenergetic challenges, and impaired neurotransmission (13,19). Some of the proposed mechanisms for the benefits of exercise include improved oxygen extraction and brain metabolism and the promotion of neuroplasticity (8,14). In addition, a recent study by Leddy et al. (20) suggested that aerobic exercise normalized local cerebral blood flow regulation and consequently improved clinical PCS, whereas a placebo stretching group did not.

As a result of this research, the management of concussions has changed significantly over recent years. In 2017, the fifth international consensus statement on concussion in sport updated their postconcussion physical activity recommendations from waiting until symptom resolution to now encouraging light physical activity after a 24- to 48-h rest period (21). With the most recent international consensus statement recommending a more active approach to concussion treatment, closer examination is needed to determine the interaction between physical activity after concussion and concussion recovery. Since 2017, a handful of systematic reviews (22,23) and meta-analyses (24–26) focusing on PCS as the primary outcome measure has been published. However, they have been limited in the number of studies included (n = 7–14) (22,24–26), specific to particular age groups (22,24,25), specific to a sport-related concussion (SRC) population (23–25), or specifically focused on individuals with persistent symptoms (22), and they have not analytically considered potential moderators of the outcomes of interest (22–24).

Therefore, the primary objective of this article is to provide a systematic review and meta-analysis of the efficacy of active rehabilitation in the treatment of concussions and the influence of potential moderators. Subjective reports of PCS are the most common metric used to define recovery, which is important for assessing and documenting PCS after a concussive event (3), and are a metric that is common to most research in this area. This review will therefore focus specifically on the effect of physical activity on PCS ratings and evaluate the influence of common moderators of this effect, such as the symptom scale used, time of injury to recruitment, and mechanism of injury (MOI).


Literature search

The review was conducted in reference to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (27). A comprehensive literature search was performed on the role of exercise in concussions utilizing the following electronic databases: MEDLINE (OVID), Scopus, PubMed, and Web of Science. The search was performed on articles up to and including articles published until May 7, 2020. To identify these articles, the following search terms were used: (concussion OR brain concussion OR post-concussion syndrome OR mild traumatic brain injury (mTBI) OR traumatic brain injury (TBI)) AND (exercise OR exercise therapy OR physical activity).

To be considered for inclusion in this review, articles had to 1) be original research (including randomized clinical trials, quasi-experimental designs, case studies and series, prospective and retrospective studies, cohort studies, and pilot studies); 2) include human participants who had sustained a concussion or mTBI, with PCS at the time of assessment; 3) implement a treatment that involved physical activity; and 4) evaluate and report the effect of such treatment on PCS. For the purposes of this review, physical activity was defined as any planned aerobic exercise (e.g., walking or cycling) or activities of daily living that elevate energy expenditure above resting levels (28). Physical activity did not include any anaerobic activity, such as weightlifting or strength training.

Only articles published in English were included. Review articles and articles published in abstract form only were excluded. In addition, articles were excluded that did not feature aerobic exercise as a form of treatment for concussion or did not discuss PCS as an outcome measure. There were no limits on the age of participants, MOI, or date of publication. Studies involving treatment both in the acute and chronic phases after concussion were included. Study titles and abstracts were assessed against the inclusion/exclusion criteria. The full text of all potentially relevant studies was then reviewed to determine the final study selection based on criteria stated previously. If any of the aforementioned criteria was not met in full-text articles, then they were excluded. References of included articles and systematic reviews were also used to search for additional articles. Included studies were imported into Zotero software, and duplicate articles were removed. In studies with duplicate data (companion publications), only the original study or the study with the most detailed or recent data was included.

The meta-analysis was conducted in reference to the Meta-Analysis of Observational Studies in Epidemiology protocol (29). Data extracted from included studies consisted of information on study design, participants (sample size, age, sex, athletic level, duration of symptoms, sampling methods), details of the intervention(s), and key findings on outcome measures associated with PCS.

Effect size calculation

Hedges’ g was used to compute the standardized effect size for each study as the difference in the presymptom and postsymptom scores, divided by the pooled SD, using Comprehensive Meta-Analysis, Version 3.0 (Biostat, Engelwood, NJ):


A random-effects model was used, which assumes that the participants across different studies are sampled from different populations (30). The overall effect size was calculated through weighted values, with the weight of effect sizes determined by sample size and variability (31). Therefore, studies with larger sample sizes and lower variance were given more weight (32). Effect sizes were coded as “positive” when studies reported an improvement in symptom scores, which was represented by a decrease in postsymptom scores compared with presymptom scores. Effect sizes were also calculated for each moderator variable.

Moderator coding

The moderators that were coded for this meta-analysis are variables that were commonly reported in the included studies and that the authors found to be important in relation to potential differences in results across studies. The following moderators were used in the analysis: physical activity type (subthreshold aerobic, multimodal, or other), symptom scale (postconcussion symptom scale (PCSS), postconcussion symptom inventory (PCSI), or other), time from injury to recruitment (≤2 wk postinjury or >2 wk postinjury), and MOI (SRC, mixed sport- and non–sport-related, or not identified).

Heterogeneity and bias

When design and/or methodological differences produce different results across studies, the standardized effect will be heterogeneous. Heterogeneity was tested by calculating 95% confidence interval around the mean standardized effect and by the calculation of the Q statistic through Comprehensive Meta-Analysis (31). A significant Q statistic indicates that the standardized effects are heterogenous. Publication bias was assessed through a funnel plot of Hedges’ g effect size versus SE (Fig. 1). A symmetrical distribution across the funnel plot indicates low publication bias (33).

Funnel plot of SE by standard difference in means. Open circles represent each effect size calculated from the included studies. The vertical line represents the Hedges’ g mean, whereas the two angled lines represent 95% confidence interval (CI). Open circles that are falling outside the 95% CI lines represent studies that do not follow the expected result, which translates to a weak or low publication bias.

Risk of bias (ROB) within studies was determined using the ROB visualization (ROBVIS) tool. Possible bias in the 23 included studies was determined through the overall outcome from five domains (34). The five domains included 1) bias arising from the randomization process, 2) bias due to deviations from intended intervention, 3) bias due to missing outcome data, 4) bias in measurement of the outcome, and 5) bias in selection of the reported result. Using ROBVIS, a traffic light plot (Fig. 2) was generated to show every judgment of ROB for each study in each domain and the overall judgment (34). A weighted bar plot was also generated (Fig. 3), which shows the proportional ROB for each domain and overall (34).

Traffic light plot of all studies in each domain and overall judgment. Dobney et al. (2018) had more than one physical activity group, but the judgments in each domain were the same for each group. Dobney et al. (2018) was the only study to produce an overall judgment of high risk because of the study being a case series.
Weighted bar plot of the amount of judgments in each domain and overall ROB. Bias due to randomization had the highest ROB, followed by bias due to missing data. The overall ROB of this review was >50% low risk.


A total of 1070 articles were identified via the electronic database search; however, 265 were duplicates and therefore removed. Overall, 805 articles were screened for eligibility, with the full text of 56 articles retrieved for detailed evaluation. Of the 56 reviewed articles, 23 records met the inclusion criteria (see Methods for criteria and Fig. 4 for PRISMA flow diagram).

PRISMA flow diagram as follows: records after duplicates removed 805, records screened 805, and records excluded 749. The inclusion/exclusion criteria of each step are outlined in the Methods section.

Participant Characteristics

The 23 articles included a total of 2547 concussed participants (age 8–53 yr; 1248 (49%) females, 1283 (50%) males), with one study not reporting their participants’ sex (35). In total, 14 of the 23 articles included children and adolescents (≤18 yr) (1,2,5,7–9,35–42), and 8 studies included adult cohorts (>18 yr) (10,17,19,20,43–46). The study by Maerlender et al. (47) did not report the age of their participants (see Table, Supplemental Digital Content 1, which summarizes the characteristics of the included articles,

Effect Sizes

A total of 29 standardized effect sizes were calculated from the 23 studies and are presented in Figure 5. Where a single study had more than one physical activity group, an effect size was calculated for each group. A positive effect size indicated a decrease in symptom severity, that is, an improvement in symptoms. None of the studies reported any adverse events in symptomatic participants after subthreshold exacerbation aerobic exercise. In the few studies where there was an exacerbation of symptoms during procedures, the symptoms resolved and did not have a negative effect on overall recovery (38,47).

Effect sizes for all studies. Twenty-nine effect sizes were calculated from 23 studies. Dobney et al. (2018) and Howell et al. (2016) had more than one physical activity group and therefore produced multiple effect sizes. All effect sizes were positive, indicating that every study showed improved symptom scores. The overall effect size of physical activity on concussion recovery was g = 1.03 (P < 0.001).

The overall effect size of physical activity on concussion recovery was g = 1.03 (Fig. 5), indicating a large, positive effect of physical activity. All 29 standardized effects were positive, indicating that every study showed improved concussion symptom scores with a physical activity intervention.

Moderators of Active Rehabilitation on Concussion Management

Physical activity type

Of the 18 studies that reported treatment duration, the most common length was 6 wk (5,8,36,38); however, durations ranged from 1 to 12 wk. The majority of studies implemented a unimodal subthreshold aerobic exercise treatment (12/23 articles; 52%). Subthreshold aerobic activity generally targeted 80% of the maximum heart rate achieved during a graded-symptom threshold test (7,17,19,20,36,37,40–42). The remaining studies used a self-selected pace (44) or the Borg rating of perceived exertion scale (8,47) to prescribe a light–moderate pace. Exercise duration ranged from 15–20 min per session or until symptom exacerbation. Ten of the 23 included studies implemented a multimodal treatment design (1,2,5,9,35,38,39,43,45,46). The content of each multimodal treatment varied but commonly consisted of components such as 1) subthreshold aerobic exercise, 2) coordination training, 3) visualization and imagery, 4) training education, and 5) home-based exercise programs. The remaining study by Howell et al. (10) did not prescribe aerobic exercise, but rather assessed self-reported physical activity levels utilizing a physical activity scale adapted from the return to play protocol in the consensus statement on concussion in sport (48). The data from Howell et al. were from the groups that self-reported light-moderate or sport-specific physical activity.

Effect sizes were positive for all types of physical activity. Subthreshold aerobic activity provided the largest effect size (g = 1.71), whereas the multimodal category had a moderate effect size (g = 0.70), and the study by Howell et al. (10) classified as “other” had a large effect size (g = 1.19; Fig. 6).

Effect sizes for each moderator category. All moderator categories were positive and moderate or strong. STA, subthreshold aerobic; M, multimodal activity; O, other; M, mixed sport- and non–sport-related concussion; NI, not identified; ≤2, recruited within 2 wk after injury; >2, recruited greater than 2 wk after injury.

Symptom scale

The symptom reporting scales varied across studies (see Table, Supplemental Digital Content 1, which summarizes the characteristics of the included articles, and included PCS scales from the Sports Concussion Assessment Tools 2/3, PCSI, the Graded Symptom Checklist, Rivermead PCS Questionnaire, and the PCS scale found in the Immediate Post-Concussion Assessment and Cognitive Testing. For all scales, a lower score indicated an improvement in PCS.

For the meta-analysis, the symptom scales were categorically coded as PCSS, PCSI, or other. The majority of studies (n = 18) used PCSS from the Sports Concussion Assessment Tools 2/3 or Immediate Post-Concussion Assessment and Cognitive Testing, three used the PCSI, and the two studies that were coded in the “other” category used the Graded Symptom Checklist and Rivermead PCS Questionnaire. The standardized effect was positive for all categories of symptom scales, with g = 0.98 for PCSS, g = 1.21 for PCSI, and g = 1.42 for “other” (Fig. 6).

Time of injury to recruitment

Seven of the 23 studies recruited participants within the acute phase, which was defined as ≤2 wk after injury for the purposes of this review (19,37,39–42,44). The remaining articles assessed individuals who were said to be slow to recover after concussion and have PPCS, as classified by the criteria from the World Health Organization (8,17,20,43,45), or a minimum symptom score (36,43). The remaining articles did not provide a specific definition for “persistent symptoms” but included participants with a broad range of symptom durations (i.e., >2 wk to ≤6 months after injury). Studies in which physical activity interventions began >2 wk after obtaining a concussion had a large effect size (g = 0.92). The effect size was also large for those where participants began the physical activity intervention within the first 2 wk after injury (g = 1.46; Fig. 6).

Mechanism of injury

The majority of studies included participants who suffered from an SRC (n = 20 studies; 1591 (62%) participants)); however, other articles included participants with non-SRC MOI (n = 11 studies, 245 participants), such as motor vehicle accidents, falls, or assault (1,2,5,7–9,17,35,36,45,46). Three studies did not report participants’ MOI (n = 717 participants) (20,39,47).

For the purposes of the meta-analysis, MOI was divided into three categories, based on the participant population described in each study: SRC (n = 9), mixed (n = 11), or not identified (n = 3). The effect sizes were positive for all categories of MOI (SRC: g = 1.28; mixed: g = 1.33; not identified: g = 0.59), indicating improvements in symptoms with physical activity, regardless of MOI (Fig. 6).

Risk of Bias

The Q statistic was significant (P < 0.001), indicating heterogeneity of results across studies, and the distribution of the funnel plot (Fig. 1) suggests that the observed effects were not a result of substantial publication bias. The overall ROB of this review was >50% of studies that were determined to be low risk, and just >25% were determined high risk (Fig. 3). The largest concern within studies was bias due to randomization (Figs. 2, 3). Case series were included in this review (7,38,39) and produced the highest ROB (Fig. 2).


Physical activity as a treatment for concussions is an emerging topic within the literature. The purpose of this study was to evaluate the efficacy of active rehabilitation in the treatment of concussions and its effect on outcomes related to PCS ratings. The 23 studies included in this systematic review and meta-analysis provided evidence to support the benefits of physical activity interventions in the treatment of concussion symptoms. The majority of studies showed clear benefits of physical activity in decreasing PCS scores, and the overall effect across studies was large and positive.

Safety and benefits of physical activity

No adverse events were reported in any of the 23 studies, supporting the safety of physical activity interventions in concussed individuals. In the original reports, the majority of studies (20/23 articles; 87%) reported that physical activity after concussion was associated with a significant decrease in PCS across their various study intervals. Our results indicated that across all 23 studies, the effect size for improvements in symptoms was positive, with an overall strong effect across studies of g = 1.03. Collectively, these studies therefore provide strong support for the effectiveness of physical activity interventions in reducing PCS.

Studies that organized PCS into different domains similarly found that PCS clusters decreased with subthreshold aerobic exercise (5,9,20,41). A prospective cohort study by Dobney et al. (9) exemplified that each symptom cluster (physical, cognitive, emotional, and sleep) was improved compared with preintervention (P < 0.05). Similarly, Thastum et al. (46) found that there was a significantly larger reduction in somatic and emotional symptoms with the individually tailored intervention group compared with previous standards of care. In this case, previous standards of care pertained to recommendation of regularly exercising above the symptom threshold as long as it did not lead to prolonged exacerbation of symptoms and psychoeducation about the biopsychosocial understanding of PPCS (46).

Comparisons to control groups were not included in the meta-analysis; as for those studies that did identify a control group, the characteristics of and prescription for this group greatly varied across studies. The majority of studies that included a control condition within their experimental design showed that the magnitude of symptom reduction across the recovery timeline was greater in the treatment group compared with their respective control (8,20,36–38,40–42,44,46). However, Gauvin-Lepage et al. (5) compared aerobic exercise with previous standards of care and found that both participants in active rehabilitation and standard of care had similar decreases in PCS scores over time.

Potential mechanisms behind the influence of physical activity

Currently, the biopsychosocial mechanism underlying the decrease in PCS with physical exercise is not fully understood, although multifaceted speculations have been made. Physiologically, physical exercise may improve cerebral autoregulation (10,20,38) and promote neuron growth and repair through increased levels of brain-derived neurotrophic factor (41,42). Exercise may also improve PCS through psychosocial factors such as reducing the fear of physical activity (38,49) and illness perception (46). Previous work has revealed that participants advised to exercise may adopt a more proactive approach to recovery and be less likely to attend to or report every symptom potentially related to concussion (41). Encouraging patients to reengage in physical activity may also promote re-integration to social environments (38), which is thought to be important to symptom recovery (14,16).

Moderators on the effects of physical activity on concussion recovery

We found a moderate to large effect of physical activity on the reduction in PCS scores, regardless of the symptom scale used, the type of physical activity intervention, the time since injury, or the MOI. A wide range of symptom scales were used, resulting in different absolute values across studies (see Table, Supplemental Digital Content 1, which summarizes the characteristics of the included articles, However, most include an overlap in the types of symptoms assessed, such as headache, dizziness, fatigue, irritability, insomnia, concentration problems, and memory difficulty (5,9,39). It is likely that the overlap in symptoms assessed across metrics contributed to the lack of a moderating effect on the observed reduction in symptoms with physical activity; the effect size was large for all symptom scale classifications used in this meta-analysis.

The contents and duration of each treatment varied depending on the study. Of the 17 studies that reported treatment duration, the most common length was 6 wk; however, durations ranged from 1 to 12 wk. The type of physical activity intervention was categorized into subthreshold aerobic, multimodal, or other. In all cases, the effect was positive; however, we did note a large effect (g = 1.71) for subthreshold aerobic activity and a moderate effect for multimodal (g = 0.70). The classification of “other” included only a single study, in which participants reported on their activity rather than following a prescribed intervention (10). Studies comparing aerobic exercise to placebo-like stretching showed improvement in PCS in both treatment groups, which may indicate that even minimal activity may be beneficial (8), as supported by our findings of positive effects regardless of intervention type. However, it has also been suggested that stretching is not beneficial, as one study showed that outcomes are largely indistinguishable between rest and placebo-like stretching groups (42). Further work is necessary to characterize the best parameters (frequency, intensity, duration, and type) for physical activity interventions for concussed individuals.

The studies within this meta-analysis assessed the effect of active rehabilitation in both the acute and chronic phases after concussion, up to 6 months after injury. Although the definition of PPCS varied across studies (8,17,20,47), we found a large, positive effect of physical activity on symptom score reduction, regardless of time since injury. This is consistent with the findings of Dobney et al. (38), who reported that participation in active rehabilitation results in an improvement in PPCS irrespective of when the intervention began. However, the authors note that participants initiating active rehabilitation within 2–3 wk after injury had significantly less severe PPCS at follow-up than participants initiating physical activity ≤2 or ≥4 wk after injury. However, it should be cautioned that Dobney et al. (38) is a case series with a high ROB. Further work is therefore necessary to determine the optimal timing and to distinguish best practices for those with acute and sustained concussive symptoms. These results are promising in support of the beneficial effect of physical activity at any postconcussive stage.

Previous work has suggested that athletes may respond faster to aerobic interventions than nonathletes (8,17); therefore, bias may be introduced because of variable sample compositions. The studies included here reported variable proportions of athletes and nonathletes within study groups. Although data were not presented in a way that allowed us to include athletic status as a moderator, we did examine the moderating effect of the MOI, as a sports-related or non–sports-related injury. We found a similar, large effect of the physical activity interventions on symptom recovery, regardless of the MOI. It is possible that the non-SRC MOI, such as motor vehicle accidents, falls, or assault (1,2,5,8,9,17,35,36,40,44,49), included athletes or highly active individuals. Future work should focus on characteristics of individuals who will benefit most from physical activity interventions after concussion.

Additional factors

Other demographic determinants that may influence the effect of physical activity on concussion recovery were discussed but not specifically addressed within the studies. They included age, sex, history of migraines, and premorbid comorbidities such as depression, anxiety, and attention deficit/hyperactivity disorder. However, the most commonly cited factors were biological sex and a history of previous concussions. Sex was not examined as a moderator in this meta-analysis, as data from males and females were rarely reported independently. However, there is a trend toward female patients reporting higher PCS scores throughout study durations (5,9,42). Dobney et al. (9) reported that female sex was a significant predictor of increased PCS at follow-up. Similarly, Howell et al. (10) found that both total PCS scores at initial clinic visit and female sex were both independently associated with longer duration of symptoms. Some researchers have hypothesized that this difference in symptom reporting is due to hormonal differences between sexes (5,50), or other factors including stress, depression, anxiety, or awareness of their well-being (5,51). As a result, a higher or lower percentage of females in a study may confound results related to recovery symptoms. However, Leddy et al. (40) had 46% females in their aerobic exercise group and demonstrated that females responded to early aerobic exercise treatment just as well as males. Even in studies with a higher percentage of females than males, both females and males experienced a decrease in PCS after participation in an active rehabilitation program after concussion (9).


It is important to note that the studies in this systematic review collectively had several limitations. The included studies all used different tests and measures to evaluate intervention responses, which makes it difficult to truly synthesize results across studies. Other limitations include a lack of randomization of participants to study groups (1,2,5,7,9,17,20,38,41–45), variable sample compositions (e.g., athletes vs nonathletes), lack of control group (1,2,7,9,10,17,38,39,43,45), and the retrospective nature of many of the studies (1,7,43). In addition, many of the studies did not blind participants to treatment, which could potentially lead to intervention bias (8,40,41,46,47). Individuals who were given an active intervention may be more likely to report improvement over individuals prescribed rest (47). Moreover, many of the included studies utilized a multimodal active rehabilitation treatment. Because of the multiple components within this design, it is difficult to determine which components offer benefit over others (43).

Another potential limitation is that many of the studies did not directly monitor participants’ adherence to their respective interventions (5,41,43,46,47). Compliance is a critical component when assessing the effectiveness of exercise-based interventions. As a result, participants prescribed rest or placebo-like stretching could have been exercising without the knowledge of the researchers, which could have reduced the effect of exercise on recovery that was observed (45,47). In addition, the specific duration or intensity of exercise performed by participants was not recorded in some studies (1,10), which limits the ability to assign causal effects pertaining to different components of exercise.

An inherent limitation with using PCS as a measure for concussion recovery is that many PCS are subtle and can vary significantly both within and across patients (49). In addition, PCS are not always specific to concussions (2) and do not always emanate from the brain (52). For example, PCS commonly overlap with symptoms associated with attention-deficit/hyperactivity disorder and migraine headaches (49). Because of the self-reporting nature of PCS scales, it is possible that participants could ignore, hide, or exaggerate symptoms (49). As a result, utilizing self-reported symptoms alone to identify concussion recovery is challenging. A number of studies utilized secondary measures to assess the effect of exercise on concussion recovery, such as neuropsychological performance, measures of health quality of life, balance outcomes, and cognitive performance (2,5,36,40–42,44–46). Readers are encouraged to refer to the original articles as a full report of all of the tests and secondary measures that are outside the scope of this review.

A weak publication bias is represented by asymmetries across the funnel plot (33). However, as evident in Figure 1, a large number of effect sizes are clustered toward the point of the triangle. This clustering may be due to the expected decrease in symptom scores, regardless of physical activity interventions. As symptom scores are expected to decrease rather than increase over time, a common positive effect size for all studies is enhanced. Relying only on the funnel plot to interpret publication bias is therefore cautioned. In these data, the lack of publication bias was also supported by the Q statistic significance (P < 0.001). Few of the included studies were randomized controlled trials, resulting in “some concerns” or “high” randomization bias in many studies (1,2,5,7,9,17,20,38,39,41–45). However, the overall positive effects across all studies demonstrated here support future work in this area, specifically enhancing the number of randomized controlled trials.

Because of a lack of control groups in some studies and inconsistencies in the definition of the control group across others, only the physical activity intervention group was included in the current meta-analysis. As more studies become available, additional meta-analyses that consider the change in symptom scores in exercising groups relative to controls should be conducted.

The low number of studies involved in some of the moderator values is important to note when looking at which conditions or methods provided the best outcomes. Symptom scales were categorized into three categories. Only three studies used PCSI (g = 1.21), and two studies made up the “other” category, which had the largest effect size (g = 1.42). Similar findings can be observed for the “other” category in the physical activity type moderator, where only one study by Howell et al. (10) was included and had a large effect size (g = 1.19).

Clinical implications

Following the current guidelines, clinicians should allow an initial physical and cognitive rest period within the first 24–48 h after injury (21). Evidence from this review suggests that subthreshold aerobic activity is beneficial for PCS recovery regardless of the time since injury (≤2 wk or >2 wk after injury). However, much of this work involved supervised physical activity, further work is necessary to ensure safety in unsupervised situations. Clinicians should also note the positive, large effect sizes of all the categories regardless of the MOI or symptom scale used. These findings suggest that the potential benefits of physical activity are not limited to a particular subpopulation of concussion patients or by the symptom assessment tool. In addition, our findings indicate that unimodal subthreshold aerobic activity elicited the largest effect size, suggesting that unimodal subthreshold aerobic activity may be the best mode of physical activity to prescribe for PCS recovery.


There is increasing evidence that sustained rest after concussion is not beneficial and may be potentially harmful to recovery (10,13,14,17). As a result, individualized active rehabilitation approaches have begun challenging the old paradigm of prolonged rest as the best concussion treatment. This systematic review and meta-analysis demonstrates that current evidence supports the notion that physical activity is beneficial in decreasing PCS in both the acute and chronic phases after concussion. Subthreshold aerobic activity should therefore be considered by clinicians in treating concussion patients. However, despite this growing body of evidence, additional research is needed to determine the optimal intensity, duration, and time to initiation of aerobic exercise after concussion.

There was no funding body involved in this review. The authors do not have any conflicts of interest to declare. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the study do not constitute endorsement by the American College of Sports Medicine.


1. Gauvin-Lepage J, Friedman D, Grilli L, Gagnon I. Effect of sex on recovery from persistent postconcussion symptoms in children and adolescents participating in an active rehabilitation intervention. J Head Trauma Rehabil. 2019;34(2):96–102.
2. Imhoff S, Fait P, Carrier-Toutant F, Boulard G. Efficiency of an active rehabilitation intervention in a slow-to-recover paediatric population following mild traumatic brain injury: a pilot study. J Sports Med. 2016;2016:1–11.
3. MacFarlane MP, Glenn TC. Neurochemical cascade of concussion. Brain Inj. 2015;29(2):139–53.
4. Domínguez D, Raparla M. Neurometabolic aspects of sports-related concussion. Semin Speech Lang. 2014;35(3):159–65.
5. Gauvin-Lepage JR, Friedman D, Grilli L, et al. Effectiveness of an exercise-based active rehabilitation intervention for youth who are slow to recover after concussion. Clin J Sport Med. 2020;30(5):423–32.
6. Lovell MR, Iverson GL, Collins MW, et al. Measurement of symptoms following sports-related concussion: reliability and normative data for the post-concussion scale. Appl Neuropsychol. 2006;13(3):166–74.
7. Chrisman SPD, Whitlock KB, Somers E, et al. Pilot study of the Sub-Symptom Threshold Exercise Program (SSTEP) for persistent concussion symptoms in youth. NeuroRehabilitation. 2017;40(4):493–9.
8. Kurowski BG, Hugentobler J, Quatman-Yates C, et al. Aerobic exercise for adolescents with prolonged symptoms after mild traumatic brain injury: an exploratory randomized clinical trial. J Head Trauma Rehabil. 2017;32(2):79–89.
9. Dobney DM, Grilli L, Kocilowicz H, et al. Evaluation of an active rehabilitation program for concussion management in children and adolescents. Brain Inj. 2017;31(13–14):1753–9.
10. Howell DR, Mannix RC, Quinn B, Taylor JA, Tan CO, Meehan WP. Physical activity level and symptom duration are not associated after concussion. Am J Sports Med. 2016;44(4):1040–6.
11. Anderson V, Manikas V, Babl F, Hearps S, Dooley J. Impact of moderate exercise on post-concussive symptoms and cognitive function after concussion in children and adolescents compared to healthy controls. Int J Sports Med. 2018;39(9):696–703.
12. Leddy J, Hinds A, Sirica D, Willer B. The role of controlled exercise in concussion management. PM R. 2016;8(3S):S91–100.
13. Giza CC, Hovda DA. The New neurometabolic cascade of concussion. Neurosurgery. 2014;75:S24–33.
14. Silverberg ND, Iverson GL. Is rest after concussion “the best medicine?”: recommendations for activity resumption following concussion in athletes, civilians, and military service members. J Head Trauma Rehabil. 2013;28(4):250–9.
15. de Kruijk JR, Leffers P, Meerhoff S, Rutten J, Twijnstra A. Effectiveness of bed rest after mild traumatic brain injury: a randomised trial of no versus six days of bed rest. J Neurol Neurosurg Psychiatry. 2002;73(2):167–72.
16. Thomas DG, Apps JN, Hoffmann RG, McCrea M, Hammeke T. Benefits of strict rest after acute concussion: a randomized controlled trial. Pediatrics. 2015;135(2):213–23.
17. Leddy JJ, Kozlowski K, Donnelly JP, Pendergast DR, Epstein LH, Willer B. A preliminary study of subsymptom threshold exercise training for refractory post-concussion syndrome. Clin J Sport Med. 2010;20(1):21–7.
18. Fiuza-Luces C, Santos-Lozano A, Joyner M, et al. Exercise benefits in cardiovascular disease: beyond attenuation of traditional risk factors. Nat Rev Cardiol. 2018;15(12):731–43.
19. Leddy JJ, Hinds AL, Miecznikowski J, et al. Safety and prognostic utility of provocative exercise testing in acutely concussed adolescents: a randomized trial. Clin J Sport Med. 2018;28(1):13–20.
20. Leddy JJ, Cox JL, Baker JG, et al. Exercise treatment for postconcussion syndrome: a pilot study of changes in functional magnetic resonance imaging activation, physiology, and symptoms. J Head Trauma Rehabil. 2013;28(4):241–9.
21. McCrory P, Meeuwisse W, Dvorak J, et al. Consensus statement on concussion in sport—the 5th International Conference on Concussion in Sport held in Berlin, October 2016. Br J Sports Med. 2017;51(11):838–47.
22. McIntyre M, Kempenaar A, Amiri M, Alavinia SM, Kumbhare D. The role of subsymptom threshold aerobic exercise for persistent concussion symptoms in patients with postconcussion syndrome: a systematic review. Am J Phys Med Rehabil. 2020;99(3):257–64.
23. Schneider KJ, Leddy JJ, Guskiewicz KM, et al. Rest and treatment/rehabilitation following sport-related concussion: a systematic review. Br J Sports Med. 2017;51(12):930–4.
24. Langevin P, Frémont P, Fait P, Dubé M-O, Bertrand-Charette M, Roy J-S. Aerobic exercise for sport-related concussion: a systematic review and meta-analysis. Med Sci Sports Exerc. 2020;52(12):2491–9.
25. Powell C, McCaulley B, Scott Brosky Z, Stephenson T, Hassen-Miller A. The effect of aerobic exercise on adolescent athletes post-concussion: a systematic review and meta-analysis. Int J Sports Phys Ther. 2020;15(5):650–8.
26. Lal A, Kolakowsky-Hayner SA, Ghajar J, Balamane M. The effect of physical exercise after a concussion: a systematic review and meta-analysis. Am J Sports Med. 2018;46(3):743–52.
27. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339(jul21 1):b2535.
28. Buchner DM, Bishop J, Brown DR, et al. 2008 Physical Activity Guidelines for Americans. Available from: Accessed October 19, 2020.
29. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA. 2000;283(15):2008–12.
30. Hedges LV, Vevea JL. Fixed- and random-effects models in meta-analysis. Psychol Methods. 1998;3(4):486–504.
31. Hedges LV, Pigott TD. The power of statistical tests in meta-analysis. Psychol Methods. 2001;6(3):203–17.
32. Turner HM III, Bernard RM. Calculating and synthesizing effect sizes. Contemp Issues Commun Sci Disord. 2006;33(Spring):42–55.
33. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315(7109):629–34.
34. McGuinness LA, Higgins JPT. Risk-of-bias VISualization (robvis): An R package and Shiny web app for visualizing risk-of-bias assessments. Res Synth Methods [Internet]. Available from: doi: 10.1002/jrsm.1411. Accessed January 6, 2021.
35. Chan C, Iverson GL, Purtzki J, et al. Safety of active rehabilitation for persistent symptoms after pediatric sport-related concussion: a randomized controlled trial. Arch Phys Med Rehabil. 2018;99(2):242–9.
36. Bailey C, Meyer J, Briskin S, et al. Multidisciplinary concussion management: a model for outpatient concussion management in the acute and post-acute settings. J Head Trauma Rehabil. 2019;34(6):375–84.
37. Micay R, Richards D, Hutchison MG. Feasibility of a postacute structured aerobic exercise intervention following sport concussion in symptomatic adolescents: a randomised controlled study. BMJ Open Sport Exerc Med. 2018;4(1):e000404.
38. Dobney DM, Grilli L, Kocilowicz H, et al. Is there an optimal time to initiate an active rehabilitation protocol for concussion management in children? A case series. J Head Trauma Rehabil. 2018;33(3):E11–7.
39. Gagnon I, Galli C, Friedman D, Grilli L, Iverson GL. Active rehabilitation for children who are slow to recover following sport-related concussion. Brain Inj. 2009;23(12):956–64.
40. Leddy JJ, Haider MN, Ellis MJ, et al. Early subthreshold aerobic exercise for sport-related concussion: a randomized clinical trial. JAMA Pediatr. 2019;173(4):319–25.
41. Leddy JJ, Haider MN, Hinds AL, Darling S, Willer BS. A preliminary study of the effect of early aerobic exercise treatment for sport-related concussion in males. Clin J Sport Med. 2019;29(5):353–60.
42. Willer BS, Haider MN, Bezherano I, et al. Comparison of rest to aerobic exercise and placebo-like treatment of acute sport-related concussion in male and female adolescents. Arch Phys Med Rehabil. 2019;100(12):2267–75.
43. Grabowski P, Wilson J, Walker A, Enz D, Wang S. Multimodal impairment-based physical therapy for the treatment of patients with post-concussion syndrome: a retrospective analysis on safety and feasibility. Phys Ther Sport. 2017;23:22–30.
44. Howell DR, Brilliant AN, Oldham JR, Berkstresser B, Wang F, Meehan WP. Exercise in the first week following concussion among collegiate athletes: preliminary findings. J Sci Med Sport. 2020;23(2):112–7.
45. McGeown JP, Zerpa C, Lees S, Niccoli S, Sanzo P. Implementing a structured exercise program for persistent concussion symptoms: a pilot study on the effects on salivary brain-derived neurotrophic factor, cognition, static balance, and symptom scores. Brain Inj. 2018;32(12):1556–65.
46. Thastum MM, Rask CU, Næss-Schmidt ET, et al. Novel interdisciplinary intervention, GAIN, vs. enhanced usual care to reduce high levels of post-concussion symptoms in adolescents and young adults 2-6 months post-injury: a randomised trial. EClinicalMedicine. 2019;17:100214.
47. Maerlender A, Rieman W, Lichtenstein J, Condiracci C. Programmed physical exertion in recovery from sports-related concussion: a randomized pilot study. Dev Neuropsychol. 2015;40(5):273–8.
48. McCrory P, Meeuwisse W, Aubry M, et al. Consensus statement on concussion in sport—the 4th International Conference on Concussion in Sport held in Zurich, November 2012. J Sci Med Sport. 2013;16(3):178–89.
49. Chrisman SPD, Whitlock KB, Mendoza JA, et al. Pilot randomized controlled trial of an exercise program requiring minimal in-person visits for youth with persistent sport-related concussion. Front Neurol. 2019;10:623.
50. Brown DA, Elsass JA, Miller AJ, Reed LE, Reneker JC. Differences in symptom reporting between males and females at baseline and after a sports-related concussion: a systematic review and meta-analysis. Sports Med. 2015;45(7):1027–40.
51. Frommer LJ, Gurka KK, Cross KM, Ingersoll CD, Comstock RD, Saliba SA. Sex differences in concussion symptoms of high school athletes. J Athl Train. 2011;46(1):76–84.
52. Quatman-Yates C, Cupp A, Gunsch C, Haley T, Vaculik S, Kujawa D. Physical rehabilitation interventions for post-mTBI symptoms lasting greater than 2 weeks: systematic review. Phys Ther. 2016;96(11):1753–63.


Supplemental Digital Content

Copyright © 2021 by the American College of Sports Medicine