Differential Effects of Acute and Multiple Concussions on Gait Initiation Performance

Abstract

Objective:

To identify the effect of acute and multiple concussions on gait initiation performance.

Design:

Cohort study.

Setting:

University research center.

Participants:

A population-based sample of participants (N = 45) divided into 3 groups: No Prior Concussion, ≥3 Prior Concussions, and Acute Concussion. The Acute Concussion participants were assessed within 24 hours of their concussion. Participants were matched based on (1) sport, (2) position, and (3) anthropometric measures.

Interventions:

Participants were tested on a single occasion and performed 5 trials of gait initiation on 4 force plates. The No Prior Concussion and ≤3 Prior Concussions groups were tested out of their primary athletic season.

Main Outcome Measures:

The dependent variables were the posterior and lateral displacement and velocity of the center of pressure (COP) during the anticipatory postural adjustment phase and initial step kinematics (step length and step velocity). Comparisons between groups used a 1-way analysis of variance with Tukey post hoc when significant effects were identified and effect sizes were calculated.

Results:

There were significant effects for group for all 6 outcome measures with large effect sizes. Post hoc tests identified differences between Acute Concussion and No Prior Concussion groups for all measures. The ≥3 Prior Concussions group and No Prior Concussion group were different for COP posterior displacement (4.91±1.09 and 4.91±1.09cm, respectively, P = .032, d = 0.91) and velocity (0.18±0.06 and 0.27±0.08m/s, respectively, P = .002, d = 1.27).

Conclusion:

There was continuum of performance identified whereby the Acute Concussion participants performed worse and the No Prior Concussion participants performed the best. The ≥3 Prior Concussions generally fell between these 2 groups, but only statistical significance on COP posterior displacement and velocity. These results suggest there may be subtle neurophysiological deficits present in collegiate student-athletes with ≥3 Prior Concussions group, and further investigation over the lifespan is warranted.

Sports-related concussion is a heterogeneous injury that presents with diverse symptoms across multiple neurologic domains including cognitive, visual, autonomic, and postural control^1–7; thus, a multifaceted assessment is recommended for both acute evaluation and ongoing assessment of recovery.⁸ Impaired postural control is a well-recognized consequence of concussion and the Balance Error Scoring System (BESS) is the most commonly used assessment technique.^1,7,9,10 However, the BESS assesses only upright quiet stance and is inherently limited by poor test psychometrics, including improved performance secondary to repeated administration, high minimal detectable change score, low sensitivity, as well as influences of test environment and fatigue.^7,11–17 The typical recovery time for collegiate student-athletes on the BESS is 3–5 days and overall concussion recovery usually occurs within 2 weeks^18,19; however, instrumented gait assessments have routinely identified persistent deficits well beyond clinical recovery.^20,21 These results suggest neurophysiological recovery may persist beyond clinically defined recovery.²²

The specific association between multiple concussions and later life neuropathology remains inconsistent and a cause-and-effect relation has not been established.¹ However, multiple studies have suggested that a lifetime history of multiple concussions is associated with elevated rates of later life pathology.^23–31 For example, Guskiewicz et al identified up to a threefold increased risk of mild cognitive impairment and depression in football players with a lifetime history of ≥3 concussions.^23,24 In the military setting, a history of multiple concussions was also associated with poorer mental health outcomes.^26,27 Although neurocognitive and mental health deficits have received most of the studies,^23–28 impairments in postural control have received less attention. Impairments during quiet stance in individuals with prior concussions have been noted in some, but not all, youth and collegiate athletes.^29–31 Concussion history has also been associated with a more conservative gait strategy in collegiate athletes.^32,33 However, the effect of concussion history on the ability to plan and initiate a movement, a critical component of appropriate motor control, has not been elucidated.

Although both quiet stance and gait challenge balance, transitional movements further stress the postural control systems that require decoupling the center of pressure (COP) and center of mass (COM).³⁴ Gait initiation (GI) requires developing propulsive forces to transition from a quiet stance to locomotion as well as effective postural control to maintain upright balance during the transition.³⁵ The GI supraspinal control is regulated by the supplementary or premotor area, and this inherently destabilizing task has been a sensitive indicator of impaired postural control across diverse neurologic and aging populations.^36–40 Specifically, the posterior and lateral displacements of the COP during the anticipatory postural adjustment (APA) phase is consistently impaired; which suggests a decreased ability to organize and plan the movement.⁴¹ Specific to concussions, reduced posterior and lateral COP displacements have been identified acutely postconcussions; however, the effect of multiple concussions is unknown.⁴²

The effect of multiple concussions on postural control has received limited attention in the literature.⁴³ Postural control impairments have been identified during the GI task across multiple neurologic patient populations; therefore, the purpose of this study was to identify the effect of multiple concussion history on GI. We hypothesized that athletes with ≤3 concussions would demonstrate postural control performance closer to acutely concussed individuals and poorer than healthy athlete controls.

Methods

Participants

We recruited 45 active NCAA Division I participants based on 3 group characteristics: (1) Acute Concussion, (2) ≥3 Prior Concussions, and (3) No Prior Concussion (table 1). The Acute Concussion group participants experienced sports-related concussions that were identified by an athletic trainer and diagnosed by a licensed team physician based on clinical diagnostic criteria consistent with the Concussion in Sport Consensus statement(s) (current at the time of assessment) and were tested within 24 hours of their concussion.^44,45 The ≥3 Prior Concussions group self-reported at least 3 prior diagnosed concussions, but none within the prior 6 months, denied concussion related deficits, and had been medically cleared for full sports participation. The No Prior Concussion group denied history of diagnosed, unreported, or probable prior concussion. All participants denied a history of chronic ankle instability or other prior injury, disease (eg, neurologic, visual, vestibular), or medications that would adversely affect their postural control. The No Prior Concussion participants were matched based on sex-specific sport (eg, men’s soccer matched to men’s soccer), position as feasible, and height (within 5cm). Both the ≥3 Prior Concussions and No Prior Concussion group participants were active members of an intercollegiate athletic team, and the Acute Concussion participants were removed from participation at the time of testing, but all returned to participate within the same season as their concussion. All participants provided written and verbal informed consent prior to participating as approved by the university’s institutional review board.

Table 1.

Participant demographic and anthropometrics

Group	Sex	Age (y)	Height (m)	Weight (kg)	Concussion History* (Range)
Acute Concussion	8F/7M	20.4±1.6	1.74±0.15	70.6±18.9	Mean: 0.93±0.88
					Median: 1 (0–3)
≥3 Prior Concussions	8F/7M	20.1±1.3	1.74±0.11	80.2±23.1	Mean: 3.60±0.91
					Median: 3 (3–6)
No Prior Concussions	8F/7M	20.0±1.0	1.74±0.09	75.4±12.7	N/A

NOTE. There were no group differences for age (P = .674), height (P = .998), or weight (P = .334).

^*As expected, based on inclusion criteria, there was a significant effect for concussion history (F_2,44 = 97.586, P<.001) and post hoc identified differences between all 3 groups.

Instrumentation

The GI trials were performed on 4 embedded force plates (Model OR-6)^a mounted flush with the laboratory floor along a 10-m walkway. Kinetic data were sampled at 1000 Hz, and the COP was calculated using standard biomechanical formulas.⁴⁶

Concussion history was ascertained by participant self-report on a previously used (reported reliability intraclass coefficient:0.92) health history questionnaire and interview with research staff.⁴⁷ The No Prior Concussion group also denied history of head impacts associated with loss of consciousness, memory loss, seeing stars, or knocked silly.

Procedures

All participants were tested on a single occasion with the Acute Concussion group assessed within 24 hours of their concussion and the ≥3 Prior Concussions and No Prior Concussion groups assessed outside their traditional sport season.

All participants completed 3 components of the Sport Concussion Assessment Tool-3rd Edition (current protocol at time of testing) including the BESS, standard assessment of concussion (SAC), and a 22-item graded (0–6) graded symptom checklist (GSC) of common concussion-related symptoms.⁴⁴ These tests have been extensively described in the literature and are commonly used by athletic trainers and physicians in concussion management.^{1,8–10,48,49} The assessments were completed by trained members of the research staff.

All participants completed 5 trials of GI during the testing session. The participants began each trail with their feet in a self-selected and comfortable stance on 2 adjacent force plates and their position was marked to maintain consistency across trials. In response to a consistent verbal cue (go) from a research staff member, participants initiated gait by stepping onto a third force plate and continued walking down an unobstructed walkway for ~10m. Participants were instructed to initiate gait with their self-reported dominant limb. All participants were allowed to practice the task prior to data collection and no participant took >2 practice trails.

Data analysis

This study was a cross-sectional analysis with each participant tested on a single occasion. The Sport Concussion Assessment Tool-3rd Edition assessments were performed and scored using standard protocols consistent with the existing recommendations with the exception that the full BESS (3 stances on 2 surfaces) was performed not the modified BESS (3 stances on 1 surface).

The GI task was subdivided into 3 phrases (APA, transitional, locomotor) based on 2 landmarks in both the anterior/posterior (A/P) and medial/lateral (M/L) directions^36,37,42 (fig 1). Movement initiation during GI was determined by the first change (2 standard deviations from the mean of the first 0.5s) in ground reaction force and confirmed by COP displacement.^36,37,42 The APA phase, the first region, began with movement initiation and ended at landmark 1, the most posterior and lateral position of the COP toward the initial swing limb. The transition phase, the second region of COP displacement, represents the transition of the COP to the initial stance limb from the initial swing limb ending at landmark 2. Thereafter, the COP moved forward under the initial stance limb during the locomotor phase that continues until the stance limb until toe-off. The primary GI outcome measures were the displacement and mean velocity of the COP during the APA phase which are reliable measures (intraclass coefficient: >0.70) in neurologically impaired populations and are highly consistent in young athletic populations.^42,50

Fig 1.

Exemplar COP displacements. After movement initiation, the COP translates posterior and lateral toward the initial swing limb that represents the APA phase.

Step length and step velocity were also calculated as the step characteristic outcomes. Because kinematics was not used in this study, the initial step length was calculated from the kinetic data as the distance between the initial position of the COP at movement initiation and its position at the foot-off time of the trailing stance foot in the sagittal plane.⁵¹ Consistent with prior work, the initial step time was calculated as the time from movement initiation to the initial heel strike on the subsequent force plate, and this was used as the denominator in the mean initial step velocity calculation.^42,51

Statistical analysis

Descriptive characteristics were calculated (mean and standard deviation) for participant characteristics. The APA phase COP displacement and velocity were analyzed along with the initial step length and velocity as the primary outcome measures. A 1-way analysis of variance was performed on each of the 6 outcome measures (APA A/P displacement and velocity, APA M/L displacement and velocity, initial step length, velocity). Partial eta squared effect sizes, classified as small (0.01), medium (0.06), and large (>0.14), were calculated for significant effects. Significant outcomes were followed up with Tukey post hoc tests, and Cohen d effect sizes were classified as small (0.20), medium(0.50), and large (>0.80). The α level was set at .05 for all outcome measures.

Results

All participants were able to complete all trials without complications and no trials had to be repeated for instrumentation or participant error.

Sport concussion assessment tool

There was a significant effect of group for the BESS (F_2,44 = 4.194, P = .022, = 0.166), SAC (F_2,44 = 3.544, P = .038, = 0.144), and GSC (F_2,44 = 24.628, P<.001, = .540). For the BESS, post hoc tests identified a significantly increased number of errors between Acute Concussion (18.5±10.5 errors) and No Prior Concussion (11.5±4.7 errors, P = .026, d = 0.86) groups only. For the GSC, post hoc tests identified significantly more symptoms for the Acute Concussion (32.5±22.9) than both ≥3 Prior Concussions (3.6±4.4, P <.001, d = 1.76) and No Prior Concussion (1.7±1.5, P<.001, d = 1.90) groups. There were no significant post hoc outcomes for the SAC (table 2).

Table 2.

Concussion clinical measures outcome mean ± standard deviation

Group	BESS (95% CI)	SAC (95% CI)	GSC (95% CI)
Acute Concussion	18.5±10.5 (12.7–24.3)	26.6±2.3 (25.3–27.9)	32.5±22.9 (19.8–45.2)
≥3 Prior Concussions	12.7±4.5 (10.2–15.1)	28.1±3.6 (27.6–28.6)	3.6±4.4 (1.3–5.9)
No Prior Concussion	11.5±4.7 (8.9–14.1)	28.1±1.7 (27.1–29.0)	1.7±1.5 (0.9–5.6)

NOTE. There was a significant effect for BESS (F_2,44 = 4.194, P = .022), SAC (F_2,443.544, P = .038), and GSC (F_2,44=24.628, P<.001). Post hoc testing identified a group difference in BESS between Concussion and No Prior Concussion (P = .026, d = 0.86) and for GSC between Concussion and both >3 Prior Concussions (P<.001, d = 1.75) and No Prior Concussion (P<.001, d = 1.89) groups. There were no significant group differences on the SAC.

COP displacement

Postpriori analysis with a 1-way analysis of variance indicated there were no differences between groups for initial stance width (F_2,44 = 0.055, P = .947). Exploratory post hoc identified no differences between Acute (27.1±12.6cm), ≥3 Prior Concussions (28.0±6.2cm), and No Prior Concussion (28.3±5.1cm) groups.

There was a significant effect of group for the COP displacements in both the A/P (F_2,44 = 42.845, P<.001, = 0.671) and M/L (F_2,44 = 20.204, P<.001, = 0.490) directions. In the A/P direction, post hoc tests identified significantly less displacement in the Acute Concussion group (2.25±1.02cm) than both ≥3 Concussions (4.98±1.18cm, P<.001) and No Prior Concussion (6.00±1.29cm, P<.001) groups. The ≥3 Prior Concussions group had significantly reduced displacement compared to the No Prior Concussion groups (P =.032).

In the M/L direction, post hoc tests identified significantly less displacement in the Acute Concussion group (2.60±1.18cm) than both ≥3 Prior Concussions (5.73±1.23cm, P<.001) and No Prior Concussion (6.16±2.34cm, P<.001) groups, but no differences between the ≥3 Prior Concussions and No Prior Concussion groups (P = .770) (fig 2).

Fig 2.

COP displacements. Data scatterplot and the horizontal line represents the group mean. * represents a significant difference from the Acute Concussion group and ^† represents a significant difference from the ≤3 Prior Concussions group. (A) There was a significant difference between groups (F_2,44 = 42.845, P<.001) and post hoc test identified significant differences between the Acute Concussion (2.25±1.02cm) and both the ≥3 Prior Concussions (4.98±1.18cm, P<.001, d = 2.51) and the No Prior Concussion groups (6.00±1.29cm, P<.001, d = 3.22). There was also a significant difference between ≥3 Prior Concussions and No Prior Concussion groups (P = .032, d = 0.91). (B) There was a significant difference between groups (F_2,44 = 20.204, P<.001) and post hoc test identified significant differences between the Acute Concussion(2.60±1.18cm) and both the ≥3 Prior Concussions (5.73±1.23cm, P<.001, d = 2.60) and the No Prior Concussion groups (6.16±2.34cm, P<.001, d = 1.92). There was no difference between ≥3 Prior Concussions and No Prior Concussion groups (P = .770). A/P, Anterior/Posterior; COP, center of pressure, M/L, Medial/Lateral.

COP velocity

There was a significant effect of group for the COP velocity in both the A/P (F_2,44 = 19.074, P<.001, = 0.476) and M/L (F_2,44 = 11.431, P<.001, = 0.352) directions. In the A/P direction, post hoc tests identified significantly lower velocity in the Acute Concussion group (0.12±0.06m/s) than in both ≥3 Prior Concussions (0.18±0.06m/s, P = .037) and No Prior Concussion (0.27±0.08m/s, P<.001) groups. The ≥3 Prior Concussions group had significantly lower velocity than the No Prior Concussion group (P = .002).

In the M/L direction, post hoc tests identified significantly lower velocity in the Acute Concussion group (0.13±0.05m/s) than in both ≥3 Prior Concussions (0.21±0.06m/s, P = .024) and No Prior Concussion groups (0.27±0.08, P<.001), but no differences between the ≥3 Prior Concussions and No Prior Concussion groups (P = .119) (fig 3).

Fig 3.

COP velocity. Data scatterplot and the horizontal line represents the group mean. * represents a significant difference from the Acute Concussion group and ^† represents a significant difference from the ≥3 Prior Concussions group. (A) There was a significant difference between groups (F_2,44 = 19.074, P<.001) and post hoc test identified significant differences between the Acute Concussion (0.12±0.06m/s) and both the ≥3 Prior Concussions (0.18±0.06m/s, P = .037, d = 1.04) and the No Prior Concussion groups (0.27±0.08m/s, P<.001, d = 1.24). There was also a significant difference between ≥3 Prior Concussions and No Prior Concussion groups (P = .002, d = 1.27). (B) There was a significant difference between groups (F_2,44 = 11.431, P<.001) and post hoc test identified significant differences between the Acute Concussion (0.13±0.05m/s) and both the ≥3 prior Concussion (0.21±m/s, P = .24, d = 1.36) and the No Prior Concussion group (0.27±0.08m/s, P<.001, d = 1.59). There was no difference between ≥3 Prior Concussions and No Prior Concussion groups (P = .119).

Initial step characteristics

There was a significant effect of group for initial step length (F_2,44 = 5.155, P =.010, = 0.197) and initial step velocity (F_2,44 = 6.622, P = .003, = 0.240). For the initial step length, post hoc tests identified a significantly shorter step length in the Acute Concussion group (0.57±0.09m) than the No Prior Concussion group (0.66±0.09m, P = .008). There were no significant differences between Acute Concussion and ≥3 Prior Concussions groups (P = .104) or the ≥3 Prior Concussions and No Prior Concussion groups (P = .539).

For the initial step velocity, post hoc tests identified a significantly slower step velocity in the Acute Concussion group (0.56±0.10m/s) than the No Prior Concussion group (0.69±0.12m/s, P = .002). There were no significant differences between Acute Concussion and ≥3 Prior Concussions groups (P = .090) or the ≥3 Prior Concussions and No Prior Concussion groups (P = .321) (fig 4).

Fig 4.

Initial step characteristics. Data scatterplot and the horizontal line represents the group mean. * represents a significant difference from the Acute Concussion group. (A) There was a significant difference between groups (F_2,44 = 5.155, P = .010) and post hoc test identified significant differences between the Acute Concussion (0.57±0.09m) and the No Prior Concussion groups (0.66±0.09m, P =.008, d = 1.00). There were no differences between Acute and the ≥3 Prior Concussions (0.63±0.04m, P = .104) or between ≥3 Prior Concussions and No Prior Concussion groups (P = .539). (B) There was a significant difference between groups (F_2,44 = 6.622, P = .003) and post hoc test identified significant differences between the Acute Concussion (0.56±0.10m/s) and the No Prior Concussion groups (0.69±0.12m/s, P = .002, d = 1.00). There were no differences between Acute Concussion and the ≥3 Prior Concussions (0.64±0.08m/s, P = .090) or between ≥3 Prior Concussions and No Prior Concussion groups (P = .321).

Discussion

The primary purpose of this study was to investigate the effects of ≥3 Prior Concussions group on dynamic postural control. The main findings were that a continuum of performance was identified whereby those with acute concussions demonstrated the most substantial impairments, those with ≥3 prior concussions demonstrated moderate impairments, and the healthy student-athletes performance was consistent with healthy adults.^37,42,52 These findings suggest that beyond cognitive and psychological deficits, multiple concussions adversely affect the postural control systems, and further investigation of these effects across the lifespan should be considered to identify potential long term complications.

A reduction in the posterior displacement and velocity of the COP during the APA phase of GI is commonly observed in neurologic populations.^36–40 Herein, the Acute Concussion group demonstrated substantial reductions, with large effect sizes, as compared to the healthy participants which was expected and consistent with prior literature.⁴² The novel finding of this study was the continuum of performance whereby the athletes with ≥3 or more concussions displayed impairments in the APA COP posterior displacement and velocity with large effect sizes (d = 0.91 and d = 1.27, respectively). The posterior COP displacement during the APA is the primary mechanism for generating forward momentum for the initial step.⁵³ In older adults and patients with neurologic disorders, this reduced displacement is frequently the result of an impaired motor program, specifically inappropriate prolonged activation of the soleus and gastrocnemius muscles that counter the posterior displacement generated by the tibialis anterior. The ≥3 Prior Concussion group also had reduced posterior COP velocity. Although COM was not measured in this study, this reduction could suggest a postural control strategy potentially limiting the separation of the COP and the COM which typically occurs during GI in healthy adults. A conservative gait strategy (reduced step velocity and length, wider base of support, increased time in double support) was observed in young adults with any concussion history.³² However, the results of this study did not identify differences in stepping characteristics between the ≥3 Prior Concussions group and either the Acute Concussion or No Prior Concussion group, which were consistent with a previous finding.⁴² The ≥3 Prior Concussions group was in the middle of the spectrum, but differences were only noted between the Acute Concussion group and No Prior Concussion groups. The healthy student-athletes’ COP displacement, velocity, and initial step characteristics were largely consistent with prior findings of young adults.^37,42,52 This suggests that COP kinematics, particularly in the posterior direction, is more sensitive than stepping characteristics in identifying potential subtle neurologic deficits.

The clinical importance, particularly when the participant is still young, of these subtle deficits is unknown. The specific neurophysiological mechanisms for these deficits were not able to be elucidated from this design; but, impairments in transitional gait tasks after concussion have been identified and suggested to be an acute strategy to maintain stability.^42,54,55 However, the more important societal and public health question that remains is are these deficits simply minor and transient strategy alterations or precursors to later life neuropathological outcomes. The ≥3 Prior Concussions group’s COP posterior displacements (4.91cm) and velocity (0.18m/s) is well above prior reports of older adults and various neurologic populations which is broadly encouraging.^36–40 Conversely, and consistent with prior results, the Acute Concussion group performance was similar or worse than these populations (eg, Parkinson disease, stroke, elderly adults).^36–40 However, these results suggest that, even with repeated concussions, performance does improve over time but may not achieve age-matched characteristics, although the cross-sectional design of this study suggests caution should be taken with this interpretation. Although very premature to suggest causation, impaired postural control should be further investigated across the lifespan to identify potential later life neurologic deficits. Because repetitive head impacts appear to be associated with later life neuropathologies in some, but not all, studies,^56,57 it is important to note that participants were matched by sport in this study suggesting the group differences are more likely from the concussion history as opposed to repetitive head impacts. There is speculation that earlier age of first exposure to collision sports, and thus total head impact burden, is associated with elevated risk of neurocognitive pathologies,^58,59 but the participants herein were still collegiate student-athletes and recent findings suggest no associations in this population.^60,61

As expected, there were significant group differences in the 3 clinical measures whereby the Acute Concussion group performed worse on all outcomes with large effect sizes. However, there were no differences between the ≥3 Prior Concussions and No Prior Concussion groups. This is not surprising because the clinical tests were developed to clinically identify deficits in acutely concussed individuals and lingering deficits are identified with more sophisticated instrumented measures.²² The participants in all 3 had test performance consistent with prior findings for collegiate student-athletes providing ecological validity that these were typical concussions.^5,18 Given the disconnect between the instrumented and clinical findings, continued work is needed to identify feasible assessment techniques to more sensitively identified neurophysiological deficits.²²

Study limitations

This study assessed current collegiate student-athletes, and it is currently unknown how these findings may be magnified as each age. Martini failed to identify gait-related deficits in individuals with an average of 2–3 prior concussions over the lifespan, but did not include a group specifically with ≥3 prior concussions; thus, this remains a limitation in the literature.^28,33 Most of ≥3 Prior Concussions group participants had either 3 or 4 concussions, and thus the potential dose response beyond 3 is unknown. Although the participant’s concussion history was rigorously assessed, underreporting and lack of concussion awareness are well-known challenges and the inability to exactly document concussion history is a limitation of this study.⁶² Future studies should also incorporate motion capture, electromyography, and dual task challenges to further detail biomechanical outcome measures because results here were limited to force plate measures. Finally, this study only investigated current NCAA Division I student-athletes and thus extrapolation to other populations is not warranted.

Conclusions

The results of this study suggest that individuals with ≥3 prior concussions have impaired APAs during GI as compared to healthy individuals, but fewer impairments than individuals with acute postconcussion. Thus, a continuum of performance was identified in the posterior COP displacement and velocity. The clinical meaningfulness of these subtle deficits are unknown; however, these findings are consistent with emerging evidence of prolonged neurophysiological deficits that persist beyond apparent clinical recovery.^22,63,64 Future studies should investigate transitional gait paradigms in older adults with multiple prior concussions to elucidate the long-term effects on the postural control systems.

List of abbreviations:

A/P

anterior/posterior

APA

anticipatory postural adjustment

BESS

Balance Error Scoring System

COM

center of mass

COP

center of pressure

GI

gait initiation

GSC

graded symptom checklist

M/L

medial/lateral

SAC

standard assessment of concussion