Objective The Concussion in Sport Group guidelines recommend a multifaceted approach to help clinicians make return to sport decisions. The purpose of this study was to identify the most common multifaceted measures used to define clinical recovery from sport-related concussion in young athletes (high school and/or college level) and to summarise existing knowledge of criteria used to make return to sport decisions.
Design Systematic review.
Data sources The PubMed (MEDLINE), SPORTDiscus and Embase electronic databases were searched from 1 January 2000 to 1 March 2017 by three independent reviewers.
Eligibility criteria Inclusion criteria: elementary, high school and college age groups, and a specific definition of clinical recovery that required two or more measures. Exclusion criteria: review articles, articles using the same sample population, case studies, non-English language and those that used one measure only or did not specify the recovery measures used.
Study quality Study quality was assessed using the Downs and Black Criteria.
Results Of 2023 publications, 43 met inclusion criteria. Included articles reported the following measures of recovery: somatic symptom resolution or return to baseline (100%), cognitive recovery or return to baseline (86%), no exacerbation of symptoms on physical exertion (49%), normalisation of balance (30%), normal special physical examination (12%), successful return to school (5%), no exacerbation of symptoms with cognitive exertion (2%) and normalisation of cerebral blood flow (2%). Follow-up to validate the return to sport decision was reported in eight (19%) articles. Most studies were case–control or cohort (level of evidence 4) and had significant risk of bias.
Conclusion All studies of sport-related concussion use symptom reports to define recovery. A minority of studies used multiple measures of outcome or had clearly defined recovery criteria, the most common being a combination of a self-reported symptom checklist and a computerised neurocognitive test. Future studies ideally should define recovery a priori using objective physiological measures in addition to symptom reports.
- sport-related concussion
- mild traumatic brain injury
- student athlete
- return to play
Statistics from Altmetric.com
Concussion incidence is significant in contact sport and recreational activities.1 In 2006, 1.8–3.8 million sport-related traumatic brain injuries were estimated to occur annually in the USA. The majority of these are sport-related concussions.2 Although there is some ambiguity in the definitions of mild traumatic brain injury and concussion, the term concussion is generally used in sport-related injuries.3 Concussion occurs when sudden deceleration and rotational forces applied to the brain4 trigger an acute and subacute pathophysiological metabolic response in the absence of gross brain lesions.5 Concussion results in somatic, cognitive and emotional symptoms, cognitive impairment, abnormal physical examination findings, behavioural issues and sleep disturbance.6 Many patients with sport-related concussion recover within 7–10 days7 8 although a recent study in adolescents, which defined recovery as normalisation of physiological, visual and balance function,9 reported that recovery typically required 3–4 weeks.
The most widely accepted guidelines for return to sport are the Concussion in Sport Group (CISG) guidelines. The most recent are from the Berlin Fifth International Conference on Concussion in Sport that recommend a multifaceted evaluation to include physical examination, neuropsychological testing and a graded return to activity to help determine recovery from concussion. The latest CISG guidelines state that athletes should return to a baseline level of symptoms but do not provide definitions to establish when an athlete is fully recovered physiologically and ready to return to sport. Resolution of symptoms is recognised as a critical part of recovery, but symptom reporting alone is problematic because athletes often under-report symptoms,10 some concussion-related symptoms are reported in populations without concussion11 and symptoms are not specific to concussion.12
While the CISG guidelines recommend that symptom and cognitive recovery must occur before athletes can return to sport, actual clinical practice may differ.13 Buckley et al 14 found that 65% of athletic trainers used a multifaceted assessment to establish recovery from sport-related concussion, while 11% used only one or no assessment tool when deciding on return to sport. The most frequently used assessments were symptom reports (92%), clinical examination (86%), computerised neuropsychological testing (74%), balance testing (65%) and the Standard Assessment of Concussion (54%).
An evidence-based definition of recovery from concussion is important given the risk of more severe consequences should repeat injury occur before resolution of the first concussion15 and increased awareness of possible long-term effects of concussion.16 The test–retest reliability and internal consistency of Immediate Post-concussion Assessment and Cognitive Testing (ImPACT),17 the Buffalo Concussion Treadmill Test (BCTT)18 and the Sport Concussion Assessment Tool (SCAT)19 have been reported on but not for other measures. The international consensus meetings on concussion in sport6 7 has stated that a single criterion may not be sufficient to define concussion recovery; hence, we included only research articles that used at least two measures to define recovery from sport-related concussion. We chose the age groups of elementary, high school and college athletes for this systematic review because of the abundance of research specific to them. There are not many publications on athletes older than college age or in athletes under the age of 12 years.20 The purpose of this systematic review was to summarise the criteria that have been used to define recovery after sport-related concussion. Study quality was assessed to identify areas of potential improvement for future studies.
This review was prospectively registered on 18 January 2016 in the PROSPERO database (registration number: CRD42016032373).21
We included articles published in English that described sport-related concussion and included elementary school, high school and college age athletes. Articles had to report the recovery criteria and use at least two measures. We included those articles that were published since 2000 because of the change of definition of concussion and recommendations for sport concussion management in 2001.22 Cohen’s kappa was used to measure inter-rater reliability between reviewers for article selection.
Exclusion criteria: review articles, case reports and articles that did not clearly define recovery measure(s) used. For articles reporting on the same sample population, we included the study that reported recovery measures. If all reported recovery measures, then only the earliest article was included. We excluded articles that measured the time it took for specific clinical symptoms to resolve or for cognition to return to baseline but did not state that the athlete had recovered or was ready to return to sport or to school since normalisation of symptoms is not the same as clinical recovery.23 We also excluded articles where a physician had documented recovery independent of the study but did not specify the basis for this clinical decision. We excluded articles that stated the CISG guidelines were used but did not include sufficient detail about implementation of the return to sport protocol. Some patient samples were not exclusively sport-related concussion. Here, we identified the mechanism of injury and included the article only if the mechanism of injury for a majority (>50%) of subjects was sport-related concussion or similar to it, and the participants received treatment similar to that used for sport-related concussion.
We searched the PubMed (MEDLINE), SportDiscus and Embase electronic databases in March 2017. Search terms included: ‘Concussion AND recovery AND (athlete OR sport) AND (children OR youth OR teens OR teenagers OR college OR high school); Concussion AND symptoms AND (athlete OR sport) AND (children OR youth OR teens OR teenagers OR college OR high school); and Concussion AND resolution AND (athlete OR sport) AND (children OR youth OR teens OR teenagers OR college OR high school)’. Exact search syntax is provided. PRISMA24 flow chart was made.
Three reviewers independently screened the titles and abstracts of all articles identified in the electronic database search. If it was unclear from the title and/or abstract whether the article should be included, the full text of the article was obtained and independently screened by the three reviewers. Any discrepancies were resolved by consensus.
The following variables were independently extracted from each article by three reviewers: first author, year, study design, sample size, patient age, time to recovery and the definition of recovery. Methods of assessing and conclusions pertaining to postrecovery follow-up were also extracted when reported.
Recovery measures were categorised as:
cognitive performance at rest (computerised or paper–pen neurocognitive tests)
special physical examination
symptom exacerbation during physical exertion
symptom exacerbation during cognitive exertion
ability to maintain academic performance
special tests (eg, cerebral blood flow).
Risk of bias and level of evidence
We assessed risk of bias of included articles using the Downs and Black checklist for methodological quality.25 Level of evidence was determined according to the guidelines by Melnyk and Fineout-Overholt.26 This system uses a seven level grading system that begins with systematic review of randomised controlled trials (level 1) down to expert opinion (level 7).
The literature search yielded a total of 2294 articles (figure 1). The titles and abstracts of 2023 non-duplicate articles were screened, and the full texts of 261 articles were evaluated. Two hundred and seven articles were excluded for not meeting recovery definition criteria, and 11 articles were excluded during data extraction because their recovery criteria were not specific or were incomplete. Forty-three articles were included in this systematic review (table 1). Cohen’s kappa of inter-rater reliability for article selection was 0.61. Three articles had a minority (<50%) of subjects with concussion from non-sport activities. They were included because the authors specifically stated that the mechanism of injury and treatment of the non-sport-related concussion were similar to that of sport-related concussion.
Out of 207 excluded articles, 28 used only one recovery measure (either symptoms or neurocognitive testing), 37 monitored symptom recovery but participants were not defined as clinically recovered in the study, one was a study of sport-related concussion in adults and 141 either did not specify any recovery criteria or did not monitor recovery. The eleven articles excluded during data extraction were: Hang et al 27 and Kelty-Stephen et al,28 which used neurocognitive testing and symptoms to measure recovery but stated that recovery was defined as resolution of symptoms and that the neurocognitive tests were just being validated. Kontos et al 29 used neurocognitive tests, a mood test or professional recommendation to describe recovery but did not mention the basis for the professional recommendation or how many of the participants were cleared exclusively due to the professional recommendation. Moser et al 30 used ImPACT, which includes neurocognitive measures and a symptom checklist, and mentioned that a portion of subjects had fully recovered by the time of their last visit but did not state whether ImPACT was the main tool to determine return to sport or if it was the physician’s decision. Madura et al 31 measured concussion severity rather than recovery. The Meier et al 32 study was excluded, because it did not specify which aspect of the CISG guidelines was used to determine recovery. Studies that used the same sample were excluded. Lau et al,33 ,34 ,35 Darling et al,36 Baker et al,37 and Henry et al 38 ,39 used the same sample so only the earlier studies were included. Kostyun et al 40 examined several parameters (symptoms, cognitive impairment, return to learn and a special physical examination) but defined recovery only by symptoms returning to baseline.
Risk of bias assessment and level of evidence
Thirty-eight out of 43 studies were case–control or cohort studies (level of evidence 4).26 Study quality according to the Downs and Black criteria25 is presented in table 2. Some of the Downs and Black questions did not apply to most of the studies since they were not randomised trials. Except for Maerlender et al,41 only questions 1–3, 5–7, 10–12, 16–18, 20–22 and 25 were relevant to the majority of the studies. Most studies were of low quality (case–control or cohort, level of evidence 4) and had significant risk of bias (Downs and Black score <14). Studies in general had well defined objectives (Q1), main outcomes (Q2) and patient characteristics (Q3). Principal confounders (Q5) were not documented in some of the studies. The main findings (Q6), the random variability (Q7), probability values (Q10) and source population (Q11) were clearly documented in almost all studies. Most studies did not mention the proportion of the potential participants who agreed to participate in the study (participant representation, Q12). Most studies had appropriate internal validity (Q16-18 and Q20-22) except that there was very little adjustment for confounding variables (Q25).
All 43 studies reported symptom recovery, 37 studies (86%) used neurocognitive testing, 21 (49%) used a provocative exercise test, 13 (30%) used normalisation of balance, 5 (12%) used a special physical examination, 2 (5%) used successful return to school, 1 (2%) used absence of symptoms during cognitive exertion and 1 (2%) used normalisation of cerebral blood flow.
All of the included studies in table 3 used symptom recovery according to the following checklists: 31 used either the Post-Concussion Symptom Scale (which is part of the SCAT or the symptom checklist from ImPACT, one used CogSport for Kids, two used the Graded Symptom Checklist, one used the Post-Concussion Symptom Scale-Revised, one used CNS Vital Signs, two used the Post-Concussion Symptom Inventory, one used the Subjective Symptom Rating Scale, one used the Rivermead Post-concussion Symptom Questionnaire and three listed patient-reported symptoms or did not specify the symptom instrument used. Seven studies33 37 42–46 used a cut-off of a minimal symptom score (less than 7 out of a maximum of 132), 14 studies38 47–59 used the cut-off of ‘symptom-free’ (symptom score=0) and 10 studies41 60–68 used ‘return to baseline symptoms’ as their recovery measure. The other studies did not specify what qualified as a normal level of symptoms or did not use a 22-symptom Likert scale checklist.
The most common computerised cognitive test was ImPACT (27 studies), one used CogSport, one used Axon Computerized Cognitive Assessment Tool, one used CNS Vital Signs, five used a paper–pencil test and two did not specify. One study36 initially used the Automated Neurophysiological Assessment Metrics (ANAM) but changed to ImPACT. Seventeen studies used a non-specific definition of provocative exercise to test for exacerbation of symptoms, that is, ‘no exacerbation of symptoms on exertion’. Four studies systematically assessed exercise tolerance: three used the BCTT and one clearly described the return to sport protocol from the 2012 CISG guidelines. For balance measures, 11 used the Balance Error Scoring System (BESS, which is also part of the SCAT), one used the Sensory Organization Test and one did not specify how balance was assessed.
Combined recovery criteria
Table 4 is the contingency table for the recovery criteria (see data extraction) employed in each study. Eleven out of 43 studies used a combination of somatic symptom scales and cognitive performance to assess recovery representing the most common multimodal recovery battery employed. The remaining 32 studies used some combination of the eight recovery measures.
Eight articles out of 43 (19%) followed up with subjects after recovery. The details of postrecovery follow-up and results are in table 5. We note that it is not standard clinical practice to perform postrecovery surveillance because return to sport without symptoms is considered to be a successful outcome.
We systematically reviewed the literature for the most common measures used to make the return to sport decision after sport-related concussion. Given that the definition of concussion varies across the literature, it was not surprising that investigators varied in what criteria they used for return to sport. There are many studies that describe recovery from concussion but most do not indicate what measures were used to objectively make the return to sport decision. This is important for possible replication. It has been suggested that multiple measures should be used to make the return to sport decision from concussion, but our search revealed that most studies used only one outcome measure (usually symptom resolution). The studies that used more than one recovery measure, those that qualified for inclusion in this review, used different combinations of measures.
Level of evidence and study quality
Most of the studies were cohort or case–control studies. For study quality, the major source of bias was lack of documentation of principal confounders. History of previous concussions, for example, may be associated with longer duration of recovery and should always be documented.69 70 This could be due to the studies not documenting confounding variables or that appropriately matched controls were not employed.
Symptom resolution was the most common measure used to make the return to sport decision—all included studies used some version of a symptom assessment. The challenge for clinicians is that symptom reports are non-specific and may not coincide with brain recovery since physiological abnormalities (eg, cerebral blood flow (CBF) and diffusion tensor imaging) persisting beyond reported symptom resolution have been reported in multiple studies of sport-related concussion.71 Symptom recovery was not, however, defined always as an ‘asymptomatic state’. Healthy adolescents have been reported to have symptom severity scores of up to 6 (out of a maximum of 132) when given concussion symptom checklists72; hence, several studies used a cut-off score of less than 7 to define symptom recovery. Some studies used ‘return to baseline’ and some used the terms ‘asymptomatic’ or ‘symptom score of 0’ to define symptom recovery. We found only two articles73 74 that did not use symptoms as a recovery criterion, but since they used only one measure to define recovery (ie, neurocognitive testing), they were not included in the final sample. Symptoms are usually assessed by symptom checklists. Concussion symptoms, however, are not specific to concussion. Leddy et al,12 for example, found no difference in the symptom patterns reported by those with concussion when compared with those who had cervical and vestibular issues.
The second most common measure used for the return to play decision was return to baseline using neurocognitive testing, usually with a computer test like ImPACT. ImPACT assesses symptoms and aspects of cognition including visual memory, verbal memory, visual motor speed and reaction time that can be compared with individual preinjury or age-normative values. The other computerised neurocognitive tests used were ANAM,75 Axon Computerised Cognitive Assessment Tool, CogSport for Kids76 and CNS Vital Signs,77 which are similar to ImPACT in that they measure different aspects of cognition and have a symptom checklist. Computerised neurocognitive tests can be administered in a 25–30 min sitting and are used widely throughout the world.78 However, there are limitations of these tests with respect to retest reliability. For example, the intraclass correlation coefficient for ImPACT has been reported to range between 0.15 and 0.39.79 Pen and paper tests (Children’s Color Trails, Rey Auditory Verbal Learning Test, Rey Complex Figure Test, Stroop Color and Word Test Children’s version, Symbol Digit Modalities Test, Trail Making Test-B and Digit Symbol Substitution Test) have the limitation that they need a neuropsychologist to interpret the results and are not reliable when used in succession due to the learning effect.43
Physical exertion testing
The CISG guidelines’ graduated return to sport strategy is a clinical guideline that can be adapted to specific sports. Exacerbation of symptoms at any level is a reason to return to the previous step until the athlete can exercise without symptoms at the level of exertion required for that sport. The principle of return of normal exercise tolerance has been used in 21 studies to establish physiological recovery from concussion. Three of the studies systematically evaluated exercise tolerance after concussion using the BCTT, all from the institution where the test was developed. This test is considered to be a clinical measure of autoregulation of CBF during exercise.42 It has been used to establish physiological recovery from concussion in adolescents after sport-related concussion.36 The BCTT is the only functional test that has been shown to safely80 and reliably18 diagnose and establish recovery from exercise intolerance36 after sport-related concussion and is used throughout the world in the assessment of athletes after head injury.
Balance and special physical examination tests
We included those measures that are considered to be more concussion specific, for example, the vestibular-ocular examination.81 Balance is an important criterion for the return to sport decision since persisting and untreated balance deficits could lead to future injuries on the playing field.33 The most common balance measure used was the BESS test, which is part of the SCAT. Vestibular components of the physician physical examination that measure balance, such as tandem gait, might have been used, but the studies that utilised a physical examination did not provide a description of the elements performed. There is a clear need for a sensitive and a specific physical examination that includes the elements most related to the injury that produces concussion, specifically the cervical, oculomotor and vestibular systems. A concussion-relevant physical examination could help clinicians with diagnosis and to establish recovery from concussion.82
Physiological measures fMRI with N-Back testing,60 resting CBF83 and EEG84 have been used to try to establish concussion recovery using physiological parameters that do not involve motivation or effort. There is discordance between normalisation of these tests and symptom resolution. Physiological assessment of concussion has the potential to establish objective ‘physiological biomarkers’ of recovery. Physiological tests, however, require validation in larger and more diverse samples because some participants had abnormal fMRI, CBF and brain electrophysiological measures despite reporting symptom resolution. The clinical significance of these measures and whether they represent ongoing neuronal or cerebrovascular damage, recovery or adaptation remains unknown.
Studies assessing return to sport follow-up and their implications for concussion assessment
Validation of the return to sport decision made by clinicians can be established if athletes who were returned to sport after concussion did not have return of symptoms while playing their sport. Eight articles in this systematic review evaluated the return to sport decision and they were diverse in terms of the length of time to follow-up. Among high school athletes, 39% who returned to sport (after apparent recovery based on symptom resolution and normalisation of cognitive performance, balance and exercise tolerance on the BCTT) reported new or increased problems on return to school36 (although it was not clear whether return to school occurred within or beyond the typical timeframe for recovery). Over a quarter (28%) of athletes diagnosed as being recovered from concussion demonstrated cognitive impairment on ImPACT after moderate physical exertion,46suggesting that physical exertion precipitated cognitive problems not identified at rest. However, physical exertion appears to adversely affect ImPACT scores,85 suggesting that it might not be appropriate to administer the test immediately after training or games. Multimodal testing on a weekly basis for up to 1 month following sport-related concussion (including general health questionnaires, symptoms, neurocognitive and balance testing and exertion testing) may be the best approach for determining recovery from concussion in young athletes.43 61 67 86 However, more research on multimodal measures of recovery after concussion is needed.
A limitation of this study is the inability to also perform a meta-analysis with the data. Table 4 shows the heterogeneity in the return to sport criteria used across studies. thus preventing a meta-analysis of the 43 studies. Furthermore, while some of the studies employed similar return to sport criteria (eg, row 1 of table 4), they did not provide individual patient level data but merely summary statistics with little uniformity in the provided summary statistics. These reasons make a meaningful meta-analysis impossible. Average return to sport times reported for the same set of criteria used by different studies are not truly comparable given the different study inclusion criteria and variable or undefined treatment.
Our systematic review is at risk of publication bias since we only included published, peer-reviewed articles, and we could not search all the grey literature. There is also a risk for language bias since we only included English-language articles. In addition, some researchers may have used additional measures for the return to sport decision that were not mentioned in their studies. Recent research has identified the frequency and importance of oculomotor dysfunction as an objective indicator of concussion.87 The King Devick Test for visual tracking has been shown to be sensitive for diagnosing concussion.88–90 Oculomotor testing, however, has been used primarily to diagnose concussion and not as much to establish concussion recovery. Return to baseline on neurocognitive testing can be determined using a Reliable Change Index (RCI). ImPACT uses RCI, but studies using other neurocognitive tests did not report using an RCI.
Three articles included concussions that were not exclusively sport related. Corwin et al 91 and Crowe et al 43 had 23% and 40% non-sport-related concussion subjects, respectively. They clearly mentioned that the mechanisms of injury were similar to sport-related concussion and that the concussions were treated like sport-related concussion. Brown et al 92 included a ‘majority of sport-related concussion’ and said that their sample was treated like sport-related concussion.
A more objective definition of sport-related concussion is needed so that we may better establish valid return to sport criteria. Future studies must define their return to sport criteria a priori, use validated measures whenever possible, and until a universally accepted definition of concussion is reached, use a multimodal approach to decide readiness to return to sport. This could encompass return to a baseline or normal level of symptoms, return to baseline or to normal cognitive performance, a normal physical examination (based on physical examination elements pertinent to concussion) and, in certain cases, demonstrating normal exercise tolerance on a graded physical exertion test.
There has been much written and studied about recovery from concussion, especially in the last 10–15 years. Only a minority of studies used multiple measures of recovery and had clearly defined return to sport criteria. The most common combination was self-reported symptom checklists and computerised neurocognitive tests. Second was the combination of self-reported symptoms, a computerised neurocognitive test and a physical exertion test. We conclude that there are disparate measures of recovery being used in sport-related concussion research, that the research is of limited quality and subject to bias and that it has led to conclusions with limited applicability. We recommend that a consensus be reached in the discipline regarding a set of reliable measures to make the return to sport decision. This would promote consistent study design in sport-related concussion research as well as uniform and safe applications of return to sport strategies.
What are the new findings?
Since the year 2000, 43 papers that defined recovery from sport-related concussion using two or more measures were identified.
All of the included articles used symptom recovery, 86% used cognitive recovery, 49% used response to physical exertion, 30% used balance testing, 12% used a special physical examination, 5% used return to full academic activities, 2% used a cognitive exertion test and 2% used normalisation of cerebral blood flow.
How might it impact on clinical practice in the future?
Researchers and physicians are encouraged to use standardised multiple criteria to establish recovery from sport-related concussion. Examples of these include normalisation of symptoms, a concussion-relevant physical examination, cognitive performance and exercise tolerance.
Contributors MNH, JJL, JGB, SP, MK and BSW contributed to the conception and design of the research, collection of data and writing, editing and approval of the manuscript. JCM contributed to analysis and editing and approval of the manuscript
Funding Research reported in this publication was supported by the National Institute of Neurological Disorders and Stroke of the National Institutes of Health under award number 1R01NS094444. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Funding was also provided by the Ralph and Mary Wilson Foundation and the Robert Rich Family Foundation.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Additional data can be requested from Dr Mohammad Haider at firstname.lastname@example.org.
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.