Article Text

Download PDFPDF

What domains of clinical function should be assessed after sport-related concussion? A systematic review
  1. Nina Feddermann-Demont1,2,3,
  2. Ruben J Echemendia4,5,
  3. Kathryn J Schneider6,7,8,
  4. Gary S Solomon9,
  5. K Alix Hayden10,
  6. Michael Turner11,
  7. Jiří Dvořák2,12,
  8. Dominik Straumann1,2,3,
  9. Alexander A Tarnutzer1,2
  1. 1 Department of Neurology, University Hospital Zurich, Zurich, Switzerland
  2. 2 Swiss Concussion Center, Schulthess Clinic, Zurich, Switzerland
  3. 3 University of Zurich, Zurich, Switzerland
  4. 4 Department of Psychology, University of Missouri, Kansas City, Missouri, USA
  5. 5 Psychological and Neurobehavioral Associates, Inc. State College, Pennsylvania, USA
  6. 6 Sport Injury Prevention Research Centre, Faculty of Kinesiology, University of Calgary, Calgary, Alberta, Canada
  7. 7 Alberta Children's Hospital, Research Institute, Calgary, Alberta, Canada
  8. 8 Hotchkiss Brain Institute, Cummings School of Medicine, University of Calgary, Calgary, Alberta, Canada
  9. 9 Departments of Neurological Surgery, Orthopaedic Surgery & Rehabilitation, and Psychiatry & Behavioral Sciences, Vanderbilt Sports Concussion Center, Vanderbilt University School of Medicine, Nashville, Tennessee, USA
  10. 10 Libraries and Cultural Resources, University of Calgary, Calgary, Alberta, Canada
  11. 11 The International Concussion and Head Injury Research Foundation, London, UK
  12. 12 Department of Neurology, Spine Unit, Schulthess Clinic, Zurich, Switzerland
  1. Correspondence to Dr Nina Feddermann-Demont, Swiss Concussion Center, Schulthess Clinic, Lengghalde 2, 8008 Zurich, Switzerland; nina.feddermann{at}


Background Sport-related concussion (SRC) is a clinical diagnosis made after a sport-related head trauma. Inconsistency exists regarding appropriate methods for assessing SRC, which focus largely on symptom-scores, neurocognitive functioning and postural stability.

Design Systematic literature review.

Data sources MEDLINE, EMBASE, PsycINFO, Cochrane-DSR, Cochrane CRCT, CINAHL, SPORTDiscus (accessed July 9, 2016).

Eligibility criteria for selecting studies Original (prospective) studies reporting on postinjury assessment in a clinical setting and evaluation of diagnostic tools within 2 weeks after an SRC.

Results Forty-six studies covering 3284 athletes were included out of 2170 articles. Only the prospective studies were considered for final analysis (n=33; 2416 athletes). Concussion diagnosis was typically made on the sideline by an (certified) athletic trainer (55.0%), mainly on the basis of results from a symptom-based questionnaire. Clinical domains affected included cognitive, vestibular and headache/migraine. Headache, fatigue, difficulty concentrating and dizziness were the symptoms most frequently reported. Neurocognitive testing was used in 30/33 studies (90.9%), whereas balance was assessed in 9/33 studies (27.3%).

Summary/conclusions The overall quality of the studies was considered low. The absence of an objective, gold standard criterion makes the accurate diagnosis of SRC challenging. Current approaches tend to emphasise cognition, symptom assessment and postural stability with less of a focus on other domains of functioning. We propose that the clinical assessment of SRC should be symptom based and interdisciplinary. Whenever possible, the SRC assessment should incorporate neurological, vestibular, ocular motor, visual, neurocognitive, psychological and cervical aspects.

  • sports
  • head trauma
  • signs and symptoms
  • assessment
  • preseason baseline testing
  • systematic review

Statistics from

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.


Sport-related concussion (SRC) is a clinical diagnosis for which no single diagnostic test or biomarker has been identifed.1 Complicating the diagnostic picture is the fact that the symptoms of SRC vary, occur frequently in non-concussed individuals, are not specific to concussion2 (see table 1) and can change unexpectedly and dynamically.3 A variety of approaches have been used for assessing concussion characteristics with a primary focus on the symptom clusters of cognition, neuropsychiatric symptoms, balance, sleep disturbances and migraine/headache.4 Other common symptoms such as dizziness/vertigo and visual/ocular motor problems have been studied less frequently.5

Table 1

Survey on symptoms (according to 22-item PCSS from SCAT3) and possible aetiology 

The variety of symptoms and disturbances associated with concussion necessitates the use of different assessment tests, batteries and strategies, which may help explain why there is little consistency in approaches to the diagnosis of concussion. One area of consistency that does emerge is the routine assessment of cognition with respect to rehabilitation, return to exercise, routine training and match play decisions.6 7

The determination of preinjury status, mainly through the assessment of neurocognitive functioning, balance and symptoms at baseline, has become popular within the last decade, predominantly in elite sports. The comparison of preinjury and postinjury data was designed to identify changes or abnormalities due to SRC. Yet difficulties exist in the appropriate interpretation of test results due to the psychometric properties of the tests and a variety of potential confounding factors (eg, test conditions, concurrent drug intake, motivation or quality of test instructions).8

The aim of this systematic review was to answer the following key questions:

  1. What domains of clinical function should be assessed post SRC and what is the evidence for the utility of these approaches?

  2. What tools/examination techniques should be used, and when?

  3. When is it appropriate to apply preinjury baseline testing to assist in the interpretation of postinjury test data (eg, cognitive, balance, ocular motor).

Materials and methods

Data sources and searches

To address the above research questions, a systematic review of the literature was performed. A literature search (MEDLINE (OVID), EMBASE (OVID), PsycINFO (OVID), Cochrane Database of Systematic Reviews (OVID), Cochrane Central Register of Controlled Trials (OVID), CINAHL (EBSCO), SPORTDiscus (EBSCO)) was conducted (accessed July 9, 2016) to identify original articles reporting on impaired clinical domains after SRC or sport-related mild traumatic brain injury (mTBI) and their assessment methods. Keywords were first generated by the content expert team (all coauthors), and then adapted by a health sciences librarian (KAH) into a comprehensive search strategy in MEDLINE. The MEDLINE (OVID) search strategy was translated for each database (see online supplementary file 1). Publications were screened with respect to original data.

Supplementary Material

Supplementary file

Only studies with postinjury assessment in a clinical setting within the first 14 days post injury were included to better identify acute symptoms by minimising the overlap with persistent symptoms and the influence of secondary symptoms.9 Research abstracts from meeting proceedings, PhD theses, unpublished studies and non-English language studies were not included in the search. Retrospective or prospective studies with five or more injured participants were eligible for inclusion. This review complies with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.10A formal review protocol was not registered or posted.

Study selection

Title/abstract and full article screening was performed by two of the authors (NFD and AAT) independently. Discrepancies were resolved by discussion and—if required—by a third reviewer. Articles were selected using the following predetermined criteria: English language publications, human population, original research articles, SRC or mTBI as source of injury, data reported on postinjury assessment and evaluation of diagnostic tools within first 14 days post injury, age 13 years or older, and five or more cases (see online supplementary table 2).

Supplementary Material

Supplementary tables

Data extraction and quality assessment

After initial assessment of included studies, it was determined that the diagnostic approaches varied considerably among studies. Therefore, the first question (What clinical domains should be assessed to make the adequate diagnosis of SRC?) was split into two parts. These were as follows: (1) What is the most accurate approach to make a diagnosis of SRC? and (2) What clinical domains should be assessed to identify optimal intervention approaches for rehabilitation and recovery?

Reports on clinical domains and the assessment methods used for concussion diagnosis were considered. Data extraction was performed by NFD and confirmed by AAT. When extracting data from selected studies, the following characteristics were assessed: type of study, participants (including age, gender and sport/level of play), times cited in Google Scholar and Web of Science (Thomson Reuters, New York, New York, USA), concussion definition, source/place/time of the diagnosis, assessment components, clinical domains covered, results and the use of neurocognitive baseline testing. In studies reporting on neurocognitive testing, all tests used were identified and assigned to the category that best described the cognitive domain evaluated.

A standardised risk-of-bias assessment was performed using the Newcastle-Ottawa Scale (NOS).11 The use of the NOS for observational studies has been promoted by the Cochrane Collaboration.12 The NOS requires rating the selection, comparability and exposure/outcome for a total of nine items (see online supplementary file 1). Risk of bias was evaluated by NFD and AAT, independently. Discrepancies were resolved by discussion and—if required—by a third reviewer. Risk of bias was rated as ‘good,’ ‘fair’ or ‘poor’ (see online supplementary file 2, supplementary tables 3 and 4). Level of evidence was completed as per the Oxford Centre for Evidence-Based Medicine 2011 Levels of Evidence.13

Because the quality of retrospective studies is often poor,14 and because a sufficiently large number of prospective studies were identified, it was decided to consider only prospective studies for further evaluation (see online supplementary file 2).

Data synthesis and analysis

Data were extracted and entered into an Excel 2011 spreadsheet. A summary of demographic characteristics of the data was performed (total number of athletes, mean age, distribution of sex, type of sport, level of play and medical background of examiner) to estimate the representativeness of data.

To cover the three predefined research questions, studies were screened for acute postconcussive symptoms reported and diagnostic tests used for concussion diagnosis (question 1) and outcome (question 2). A subanalysis of predominant symptoms was performed for the studies that used a 21-item or 22-item graduated symptom scale (questions 1 and 2). Because neurocognitive functioning was the domain assessed most frequently, a subanalysis used with respect to domains covered, latency of examination and outcome was performed (question 2).

Finally, studies that reported comparisons of postinjury test results to (a) individual preseason baseline tests or (b) normative values and/or control groups were evaluated (question 3).


Database searches resulted in 6152 citations, of which 2281 were duplicate citations, resulting in 3871 unique citations for screening. Preliminary screening of publication type (editorial, conference, abstracts, etc) excluded an additional 1701 citations; 2170 citations for abstract/title screening, 89 citations for full-text screening and 46 full-text articles (figure 1) met inclusion criteria (published between 1995 and 2015 covering 3284 concussed athletes).

Figure 1

Flow chart on the selection process of identified articles. FU, follow-up. 

There were 33 studies classified as prospective and included for further analysis involving 2416 athletes of which 87.3% were male (see table 2).

Table 2

Data extraction—prospective studies*

The mean age of the athletes was 17.4 years, with the majority of athletes participating in collision sports, such as American football (49.3%), rugby (4.9%) and Australian Rules football (5.9%). Most studies took place in American high schools and colleges (n=20, 60.6%). Concussion symptoms in athletes were mainly identified at the sideline (63.7%) by an athletic trainer (54.8%) and in one-third of concussions by the (team) physician (12.0%) or team medical staff (21.7%). Thirty studies covering 2131 athletes (or 88.2%) provided results of neurocognitive testing, while balance data were available from 682 (28.2%) athletes. Ocular motor test results were available from two studies (3.6% of athletes).15 16

For intraindividual comparisons, neurocognitive preseason baseline testing was performed in 81.8% of studies (reflecting 90.7% of all examined athletes). Neurocognitive testing was combined with a balance test in 25.0% of the studies (see table 3).

Table 3

Summary information on included prospective studies

Risk of bias assessment

NOS ratings were ‘good’ (7–9 points) in 5 studies (10.9%, prospective only n=4 studies or 12.1%), ‘fair’ (3–6 points) in 39 studies (84.8%, prospective only n=28 or 84.8%) and 'poor' (0–2 points) in 2 studies (4.3%, prospective only n=1 or 3.0%). Reasons for lower scores and higher risk of bias varied between studies but included factors related to selection, measurement and confounding factors. In many cases, the selection of exposed and non-exposed cases came from different communities15 17–29 or was not specified.30–38 A lack of matching of exposed and non-exposed cases to control for age, sex/gender, education, handedness, preinjury or level of play also commonly occurred.7 15–18 20 21 23–29 39–50 In addition, lack of valid measurement tools and threat of recall bias through self-report or potential for expectation bias through an unblinded assessment7 22 44 were all common sources of systematic error (see online supplementary tables 3 and 4).

Level of evidence of the prospective studies was considered low, with most studies (n=25, 75.8%) classified as level 4 evidence, seven studies (21.2%) as level 3 evidence and only one study33 (3%) as level 2 evidence.

A total of 39 different (standardised) questionnaires and test batteries (on average 2.5 tests/instruments per study) were identified among studies that used a postinjury examination. These measures included 11 symptom scales, assessment of neurocognitive functioning via computerised tests (Computerised Neuropsychological Test; CNP, n=5) and traditional paper and pencil tests (P&P, n=17), balance (n=4), and ocular motor function assessments (n=2) (see table 1).


Post-traumatic symptoms were reported in 28/33 (84.8%) prospective studies. The number of symptoms assessed via standardised scales or questionnaires varied between 745 and 2434 items (see table 4a). Nineteen studies (covering 1618 athletes) reported on symptoms assessed by the Post-Concussion Symptom Scale (PCSS, 21 or 22 items).51 The average postinjury total symptom score using the original form of the PCSS (information provided by 12 studies)4 7 15 23 29 36 41 42 44 48 52 53 ranged between 8.57(<24 hours) and 45.652(24–48 hours). Only three studies32 33 48 provided information on average total number of symptoms with scores ranging between 8.333 (>48 hours) and 2032(24–48 hours) (see table 4a).

Table 4a

Summary information on symptoms assessed by the PCSS, 0–22 items 

Detailed information on specific symptoms endorsed by athletes was available only for two studies.44 53 Although symptoms varied, the most frequently endorsed symptoms (see table 4b) reflected issues with alertness/attention (n=165 athletes, three items pooled), dizziness/balance (n=151 athletes, three items pooled), headache/migraine (n=161 athletes, four items pooled) and consciousness/awareness (n=152 athletes, three items pooled). Turning to individual symptoms, headache (n=70, 71.4%), fatigue or low energy (n=62, 63.3%), concentration problems (n=60, 61.2%), dizziness, drowsiness and feeling slowed down (each n=58, 59.2%) were frequently described, followed by fogginess (53%, 54.1%) and memory problems (n=47, 48.0%). Neck pain was not reported in any of the studies (see table 4b).

Table 4b

Summary information on specific symptoms assessed according to the PCSS SCAT3 (0–22 items)[i]

‘Dizziness’ and ‘fogginess’ were associated with a higher total number of symptoms and prolonged recovery.42 54 55 Houston and colleagues41 for example, demonstrated that 10 days post injury, the average symptom level (3.4±8.4) decreased below the preinjury level (11.0±13.1).

Schmidt and colleagues50 recommended the graded symptom checklist as a core component of any preseason or preparticipation test because it is easily administered, inexpensive, unaffected by group administration and provides an individualised measure of self-reported symptoms.

Neurocognitive testing

Thirty prospective studies included CNP tests for postconcussion assessment. The CNPs used ImPACT (Immediate Post-Concussion Assessment and Cognitive Testing, n=16, NOS rating: good n=3, fair n=13),4 15 16 19 21 23 29 30 32 33 35 36 42 52 53 56 CogSport (Cogstate Concussion test, n=6, NOS rating: fair n=6),34 35 43 46 53 57 ANAM (Automated Neuropsychological Assessment Metrics, n=1, NOS rating: fair),50 Headminder (n=1, NOS rating: fair)30 and CANTAB (Cambridge Neuropsychological Test Automated Battery, n=1, NOS rating: fair),25 whereas three studies used two different batteries in the same study (NOS rating: fair).30 35 53 Moreover, 16 P&P tests were identified and used in nine studies (NOS rating: good n=1, fair n=8),7 22 25 30 34 46 57–59 while five studies reported on the results of the Standardised Assessment of Concussion (NOS rating: fair n=4, good n=1).41 45 47 48 58

The test batteries generally assessed executive function, attention, learning and memory. CNP baseline tests were available in the majority (n=24) of studies (see table 1). Of those studies that compared concussed athletes to controls (n=12), significant group differences (p<0.05) were found in more than half of the studies. Group differences were found in memory and learning (58.3%), executive functioning (54.5%) and attention (50%). When compared with individual baseline data, significant differences (p<0.05) were reported in fewer studies. Among these studies, differences were found in attention (36.4%), memory and learning (32%) and executive functions (20%) (see table 5). Fogginess was associated with reduced memory performance and slower processing speed in one study.42

Table 5

Survey on studies reporting on neurocognitive testing (n=30)

Two studies reported on differences in symptomatic versus asymptomatic athletes after concussion with respect to neurocognitive test results.19 57 Fazio and colleagues19 identified significant (p<0.00) impairments in composite scores of verbal memory, visual memory, reaction time and processing speed in both groups, but the asymptomatic group demonstrated better scores than the symptomatic group. Significant (p<0.01) postconcussion decline of simple, choice and complex reaction times of the symptomatic group compared with the asymptomatic and control groups was reported by Collie and colleague.57

Comparing the baseline approach with the normative method has demonstrated significant advantages of the baseline method with respect to simple reaction time (p=0.043), while mathematical processing was significantly assessed more accurately by the normative comparison method (p=0.001).50


Nine prospective studies presented data on balance by using the (modified) Balance Error Scoring System (BESS; included in the Sport Concussion Assessment Tool (SCAT) as the balance test, n=7, NOS rating: good n=1, fair n=5, poor n=1),33 37 58 the Sensory Organization Test (SOT, n=3, NOS rating: fair n=3)30 37 40 or the Health's Balance Accelerometer Measure (BAM) 60 apart from BESS (NOS rating: poor).20

Studies on the BESS mainly reported the total BESS score.20 40 41 58 One study reported data on the six BESS individual stance conditions20 and found significant differences (p<0.01) for the tandem gait (firm/foam) among athletes with SRC when compared with healthy controls. In the other studies, BESS differences between concussed athletes and controls were either not reported or unclear. Significant postinjury differences (each p<0.01) for the modified BESS (SCAT) were identified in one study.48 BESS balance deficits in concussed college football players were described immediately after the injury with a gradual improvement within several days.16 Houston et al 41 identified an association between BESS score and health-related quality-of-life measures 72 hours post injury.

The BAM was classified as not as effective in identifying abnormal postural control compared with the BESS.20

Broglio and colleagues30 reported impairment in the SOT composite score in 36.5% (23/63) of concussed athletes. Significant differences were identified in the SOT composite score of concussed athletes when compared with baseline (p=0.037) and non-injured control group (p=0.025), detectable up to 14 days post injury.40

An (indirect) ocular motor test (King Devick) was included in one study (NOS rating: fair)16 that evaluated nine concussed high school football players. This test demonstrated a significant increase in reading time (p=0.001) when comparing immediate (<30 min) postinjury results to individual preseason baseline tests and to non-concussed controls. Near point of convergence was compared with performance on a CNP in one study (NOS rating: fair), which revealed that impaired convergence was associated with significant impairment on verbal memory (p=0.02), visual-motor speed (p=0.02) and reaction time (p=0.001) in concussed athletes and was associated with a higher total symptom postinjury score (p=0.02).15


The literature was reviewed to address the following question: Which clinical domains should be assessed post injury? Additional areas of focus included examination of the empirical evidence underlying various assessment approaches (aim 1), the appropriate tools/examination techniques to assess these domains (aim 2) and the contribution of baseline testing (aim 3).

Summary of included participants: representativeness of data

Currently, the literature on the diagnosis of SRC is mainly representative of adolescents and young adult male high school or college athletes who are participating in collision sports, but not for amateur or elite adult athletes. The high frequency of adolescent and young adult athletes might be explained by the large number of athletes participating at these levels, whereas for other levels (elite and amateurs), either access to the players might have been more difficult, funding may have been more limited or medical coverage non-accessible. Although for elite athletes certain limitations identified here might be less relevant (eg, lacking initial assessment by the team physician),32 33 41 future studies focusing on elite (male and female) athletes should be initiated to further address these questions.

Of the 2416 athletes in the included studies who received a diagnosis of SRC, almost half were involved in American football. This might be explained by the high risk of SRC in American football and the popularity of this sport in the USA, or simply that greater research has been devoted to this sport.61 62 The number of athletes involved in other collision or non-collision sports (eg, rugby, ice hockey, or football/soccer) who were diagnosed with SRC was much smaller (<6% per sport), which necessarily limits conclusions that can be made regarding SRC in these sports.

Likewise, only 12.7% of all concussed athletes included in this review were female. At first sight, this observation was unexpected because the incidence of concussion may be higher in women.63 64 However, women typically do not participate in American football, and this may be reflected in the reportedly lower numbers of active participation of female athletes in collision sports.64 65

Among the studies reviewed, the initial diagnosis of SRC was made at the sideline based on the results of a symptom questionnaire. In the majority of athletes (54.8%), this diagnosis was made by an athletic trainer, and less frequently by team physicians or other healthcare providers (37.6%). In light of this variability, the accuracy of sideline diagnosis of concussion should be interpreted with caution. However, it is unclear how many teams had a dedicated team physician and whether athletic trainers actually diagnosed SRC or removed an athlete from play due to suspected concussion. Fuller and colleagues35 reported on 26/81 (32.1%) cases in rugby players, where the initial diagnosis of concussion (made by a physician or other) was not confirmed in the follow-up examination in a clinical setting.

The risk of bias was fair in most (84.8%) of the prospective studies reviewed. Limitations in study designs were noted in several studies. Underpowered studies may have resulted in a type II error, where a difference was not detected when in fact one does exist.16 19 22 25 34 40 46 Selection bias may have resulted in a systematic difference in the test values based on the inappropriate selection of controls, resulting in a potential overestimate in test scores.19 40 A variety of assessment tests were administered by a number of healthcare professionals with different backgrounds. This may have resulted in a misclassification of test results and has the potential to result in an overestimate or underestimate of test scores, thus misinforming the true value of a test. In the studies that required retrospective recall, recall bias may have resulted in an overestimate of symptom reporting, or alternately an under-reporting of symptoms in some cases. This may also have been the case for the diagnosis of concussion, as the gold standard of diagnosis is a clinical evaluation. There is a lack of objective clinical tests in many cases, thus misclassification of test outcomes may have occurred and resulted in an underestimate or overestimate of the true outcome of a test. These threats to internal validity impact the strength of the ultimate conclusions made by the systematic review.

Consequently, future prospective studies should use a standardised definition of concussion. Case–control or cohort studies that aim at diagnostic accuracy and take into account potential distractors (eg, previous history of concussion, the effects of exertion/time since cessation of activity, time of day, fatigue) that may alter the outcome on such tests are urgently needed.

Most studies that reported postinjury assessment in a clinical setting had a strong focus on neurocognitive testing, whereas other assessment approaches (including vestibular, ocular motor or cervical complaints) were rarely addressed. As SRC is a complex injury, and often accompanied by concomitant injuries, an interdisciplinary approach to evaluation and treatment is warranted. Although ideal for good patient care, interdisciplinary teams also bring together a heterogeneous group of healthcare professionals with different areas of expertise and different levels of training that may lead to varying levels of reliability in the diagnosis and treatment of concussion. For this reason, standardised testing methods and the adoption of a single definition of SRC that is accepted across disciplines and applied uniformly within healthcare disciplines are recommended.

The variability in concussion-related symptoms underscores the need for comprehensive interdisciplinary evaluation

This review confirms previous findings that symptoms of concussion are heterogeneous, not specific and can sometimes even be misleading.66 The average total number of symptoms and the average symptoms severity score post injury varied considerably among studies. The highest number of symptoms and total symptom scores were observed between 24 and 48 hours post injury. It was surprising that most studies focused on the total number of symptoms (and the total symptom score) instead of the type of symptoms because symptoms typically guide the diagnostic decision and therapeutic management.66 As few studies provided symptom-specific data, the interpretation of total symptom scores is challenging.

The most prevalent symptoms reported by the athletes in studies using the PCSS were headache (71.4%), fatigue or low energy (63.3%), concentration problems (61.2%), dizziness, drowsiness and feeling slowed down (each 59.2%), fogginess (53%) and memory problems (48.0%). Although the assessment of memory function is typically included in CNPs (to varying degrees) and frequently was used in our study samples, information on the detailed assessment of other potentially relevant symptoms such as headache, fogginess or dizziness was very limited. Only two studies added symptom-specific questionnaires,41 48 although it is well known that specific symptoms such as headache,67 dizziness54 or fogginess42 influence some neurocognitive components and symptoms impact on recovery.19 57 Greater emphasis is needed using symptom-specific diagnostic approaches for the evaluation of these frequently occurring symptoms, particularly in cases of prolonged or atypical recovery.2 45 67–69

Headache is reported to affect most athletes after SRC within the first few hours of injury, with one study reporting headache endorsement in up to 96% of all athletes.45 An accurate diagnosis of headache is critical to differentiate migraine or tension-type headaches from those caused by cervical spine dysfunction or musculoskeletal injury.67 An inaccurate diagnosis may result in inappropriate/missed treatment and therefore increases the risk for a prolonged recovery. Collins and colleagues70 identified that 7 days after concussion, athletes with headaches experienced a large number of other postconcussion symptoms compared with athletes without headache (p=0.001). Similar to headache, causes of dizziness and balance problems are multifaceted, which is reasonable given that maintaining balance requires appropriate integration of three distinct sensory systems (sensorimotor, visual and vestibular).71 The vestibular system has a high degree of plasticity and can compensate for post-traumatic functional disturbances causing dizziness and vertigo. It is fundamental to identify the impaired sensory system to ensure appropriate post-traumatic management.7 30 68 While the vestibulospinal aspects have been assessed by different balance tests (BESS and SOT), the vestibulo-ocular pathways have typically not been included in concussion management, although promising results were reported in different studies, which did not fulfil inclusion criteria due to latency of assessment (>90 days post concussive event),33 publication after the date of the systematic literature search72 or covering non-sports-related mTBI.68 69

The presence of dizziness can influence the total number of symptoms,2 ,73 is associated with a more than six times greater risk for protracted recovery69 72 and might influence neurocognitive performance.42 Therefore, to the extent possible, a standardised vestibular and ocular motor screening examination should be included in existing screening and testing batteries.68 A symptom-based, detailed and interdisciplinary examination by sports medicine clinicians experienced in concussion management should be initiated in athletes with recovery exceeding the typical recovery duration of 14 days. Moreover, it was striking that none of the studies reported on neck pain despite being a frequent concomitant injury after a head trauma, as has been commonly identified following concussion in different studies, which did not meet inclusion criteria for our review.74–76

Neurocognitive testing

Used in most studies (93.8%), neurocognitive testing has been considered the cornerstone in concussion diagnosis/management. However, the variety of tests and test components indicate that a uniform approach to neurocognitive test batteries has not been adopted. Although this complicates research, the observed variability of tests and approaches may be reflective of the complexity of neurocognitive changes that occur post injury. The deficits identified with these measures (eg, memory and learning, attention, and executive function) have been described before,7 77 underscoring the need for neurocognitive testing to include at least these three domains in the assessment of SRC. There were significant (p=<0.05) differences identified in more than half of the athletes in the subacute phase when compared with non-concussed controls or individual baselines in the studies that included statistical analysis. The exact aetiologies of these deficits/complaints remain unclear because they probably represent a complex admixture of factors.

Although neurocognitive testing can currently be regarded as an important component in the guidance of return to play management, especially meaningful after resolution of symptoms,6 the neurocognitive test findings need to be considered in the context of other symptoms and clinical findings to better judge their relative importance. Collins and colleagues,70 for instance, reported on a significant positive correlation of headache and ImPACT memory and reaction-time composite scores (at 7 days post injury) in a sample of 110 concussed high school athletes.

The complexity of selecting the right tests, the relationships among tests and the large number of possible confounding factors (including age, education, sleep habits, drug intake, motivation, language, quality of instructions or frequency of repeated exposure to the test and its relationship to test performance) argue for the use by individuals who are highly skilled in the interpretation of these tests (ie, neuropsychologists) whenever possible.50 However, it is important to underscore that neurocognitive tests and measures should not be used in isolation for the purpose of diagnosis or management of SRC.78

Balance testing

In those studies meeting inclusion criteria that used the BESS, only one study found significant (p>0.05) differences in balance between concussed athletes and (a) controls or (b) in comparison to baseline results.48 However, another study identified significant differences in 2/6 BESS test components (tandem gait on different surfaces) when compared with healthy controls 8 days post injury.20 These results were surprising because significant differences to preseason scores and matched control subjects on day 1 have been described previously.79 One explanation might be that athletes included in the review might have been already symptom-free when performing the BESS test. Studies not included in our review but represented in the review on sideline assessment by Echemendia and colleagues80 indicate the mBESS/BESS appears useful immediately after injury (eg, 24 hours) in differentiating concussed versus non-concussed athletes, but the ability to differentiate decreases significantly after 3–5 days post injury. The SOT identified abnormal findings in at least one system (vestibular, visual or somatosensory) in every third athlete. Keeping in mind the high frequency of balance problems, dizziness and blurred vision following head trauma2 68 81 and the limited amount of normative data,82 dynamic posturography might be a promising approach for balance screening. However, although useful in the research and clinical centres specialising in concussion management, the expense and bulk of systems such as Neurocom place them beyond the reach of most clinician applications. Consequently, greater emphasis should be placed on developing clinically relevant measurement devices that are easily accessible to clinicians (eg, accelerometer-assisted balance tests that can be used via smartphones). In addition, a multifaceted approach to the assessment of balance may also include tests that evaluate dynamic balance and reflect the complexity of tasks required for sport (eg, tandem gait and gait with head motion).

Role of baseline testing

Disagreement exists about the relevance of neurocognitive tests administered at baseline in SRC management due to intraindividual and interindividual differences in cognitive domains assessed across the various tests. This is made more complex in children where cognitive development and maturation occur rapidly and may require much more frequent baseline testing.8 The comparison of postinjury to baseline results has been proven to be useful when performed 2 days post injury by Lau and colleagues.36 However, as individual baseline testing is labour intensive and may exceed the financial resources of many organisations, an alternative approach is to make comparisons between the individual’s postinjury scores and appropriate normative data, where available. To date, studies indicate that the use of normative approaches may be appropriate for a large portion of individuals diagnosed with concussion but may miss some individuals who are not adequately represented in the norming sample.27 43

Conversely, it has been demonstrated that comparing ‘above average’ athletes to normative data on a CNP may result in misclassification.27 Moreover, it is largely accepted in clinical neurocognitive practice that a within-person comparison (ie, using the person as his/her own control) may be an aspirational preference.

The application of the normative comparison method may lead to a more conservative postinjury management.50


The studies reviewed were diverse, making a comparison of individual studies challenging, thus disallowing any formal meta-analysis. Due to the number of exclusion criteria applied, only published English-language articles were included, which leads to the risk of publication and language bias. In addition, only prospective studies were further analysed, which focused on assessment and clinical domains of SRC in the acute phase (up to 14 days post injury). Due to this time period limitation, relevant studies may have been omitted, where the primary focus was sideline assessment, persistent symptoms, SRC modifiers or follow-up (ie, treatment, rehabilitation or return to play). Moreover, results from studies of non-sports-related mild traumatic brain injuries were not examined or evaluated for inclusion in this review. The NOS was deemed to be the most appropriate tool to assess the risk of bias as the majority of studies were case–control or cohort designs. However, in some cases where the study designs were cross sectional or case series, this limits the utility of NOS in some studies. Many of the studies included in this review are vulnerable to measurement bias and selection bias in addition to a lack of control for potential confounding factors, limiting the conclusions that can be drawn. Thus, future research using high-quality designs, including evaluation of multiple systems using standardised, reliable and valid objective measures, will facilitate an improved understanding of the relationship among concussive injury, symptoms and functional alterations following injury.


Symptoms of SRC are heterogeneous and not specific to SRC. The symptoms involve different domains (eg, cognition, dizziness and balance, emotions, headache and vision). Currently evaluated signs of SRC primarily include only neurocognitive and balance dysfunctions.

Consequently, symptoms should be assessed using a standardised and validated symptom scale (eg, original version of PCSS) or other empirically based questionnaires. Identified predominant symptoms may need to be assessed in greater detail by validated symptom-specific scales or questionnaires. A review of the current studies yields an imbalance between domains affected by SRC and domains assessed. Domains that have received a great deal of attention include neurocognitive assessment, balance/postural stability and symptom constellations. Other important but less studied domains within SRC are vestibular, ocular motor, visual, psychological and cervical symptoms. Further research should include empirical studies, using reliable and valid standardised measures, for the objective assessment of these multiple clinical domains.

Baseline testing for the different domains remains optional, but individual baseline data may assist in the interpretation of postinjury test results in some individuals. Comparison of test results to normative data may lead to a more conservative return to play management. The benefit of baseline test results for other minimally evaluated domains to date including vestibular, ocular motor, visual, psychological and cervical functions should be a focus of future study. Early interdisciplinary assessment following a concussion within the first days may facilitate triage for appropriate intervention in a timely fashion.

What is already known?

  • Sport-related concussion (SRC) is a clinical diagnosis; no single objective test or biomarker has been identified to make the diagnosis.

  • Symptoms of SRC are heterogeneous and not specific to concussion.

  • Current approaches to SRC assessment emphasise an individualised and interdisciplinary approach.

  • The comparison of postinjury test data (symptoms, cognitive test scores and balance) to preseason-baseline data and/or data from non-injured controls is common.

What are the new findings?

  • The risk of information bias in SRC diagnosis is possibly increased due to the variety of healthcare professionals assessing SRC.

  • Reporting is often restricted to the total number of symptoms and/or symptom severity, with detailed information on the type of symptoms often unreported.

  • The use of neurocognitive testing in the early time period following concussion is common and more than half of the studies reported significant differences when compared with non-injured controls or individual baseline results.

  • Balance testing has been included in some studies, but vestibular and ocular motor systems have rarely been assessed, although preliminary results appear promising.

  • Few studies used an interdisciplinary assessment approach.


The authors thank Rosa Vissher for the support in study identification, data extraction and analysis. The authors thank Dzmitry Katsiuba for the support in data analysis.


View Abstract


  • Contributors NFD conceived the study, designed the search strategy, selected suitable articles, extracted and analysed the data, drafted the manuscript and approved the final version of the manuscript.

    JD reviewed the initial search strategy, critically reviewed the manuscript and approved the final version. KJ

    S reviewed the initial search strategy and data extraction, determined level of evidence, critically reviewed the manuscript and approved the final version.

    MT reviewed the initial search strategy, critically reviewed the manuscript and approved the final version.

    RJE reviewed the initial search strategy, critically reviewed the manuscript and approved the final version of the manuscript.

    AH designed and reviewed the search strategy, performed study search and approved the final version of the manuscript.

    GSS critically reviewed the manuscript and approved the final version.

    DS reviewed the initial search strategy, critically reviewed the manuscript and approved the final version.

    AAT selected suitable articles, supported data analysis, supported in drafting of the manuscript and approved the final version of the manuscript.

  • Competing interests KJS has received speaking honoraria for presentations at scientific meetings. She is a physiotherapy consultant at Evidence Sport and Spinal Therapy and for many athletic teams. RJE is co-chair of the NHL/NHLPA Concussion Subcommittee, Chair of the MLS Concussion Committee, and a neuropsychological consultant to Princeton University and the US Soccer Federation. He receives financial compensation for each of these activities. He is engaged in the practice of clinical neuropsychology and occasionally serves as an expert in medico-legal cases involving TBI and SRC. He has received speaking honoraria for presentations at scientific meetings. GSS is a full-time employee of the Vanderbilt University Medical Center. He is a consulting neuropsychologist for the NHL Nashville Predators, NFL Tennessee Titans and several collegiate athletic teams, with all fees paid to institution. He is also a member of the ImPACT Scientific Advisory Board, and receives expense reimbursements for attendance at board meetings. He has received speaking honoraria for presentations at scientific meetings.

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