Evaluating Recovery After Two and Three Repeated Concussions Using Growth Curves

Abstract

The results of prior research concerning the effects of repeated concussions have been mixed. The aim of this study was to evaluate how concussion outcomes and presentation changed within patients who were evaluated at a concussion specialty clinic multiple times with a concussion. Subjects included 202 patients (54% male) aged 10–21 years (M = 13.17) who presented to a specialty concussion clinic for two and three concussions (77% sport-related) and were followed through formal clearance. First, growth curve models were estimated to determine recovery time and initial symptom burden across the multiple injuries. Second, covariates were added to these models to evaluate which demographic, risk factor, or injury variables predicted any change that did occur in evaluation or outcome variables. Models indicated that each subsequent concussion linearly resulted in significantly fewer days to recovery (-4.62 days, p < 0.047) across three concussions, and significantly lower (and linear) symptom scores on the post-concussion symptom scale (PCSS) (-2.16, p = 0.05). More severe presentation (i.e., days to recovery; higher symptom score) was significantly associated (-.62, p = 0.005) with greater improvement in recovery time (-.62, p = 0.005) and symptom burden (-.56, p < 0.001) at subsequent injuries. No covariates were significantly associated with improvement (or lack thereof) at subsequent injuries. This study adds to evidence suggesting multiple injuries is not associated with protracted recovery at subsequent injuries, in the context of treatment and full clearance for each injury at a multi-disciplinary clinic.

Introduction

Approximately 2.5 million concussions occur each year among physically active high school adolescents, and just under one million such adolescents report multiple concussions in the last year.¹ Concussions are characterized by symptoms across cognitive, physical, and behavioral domains.² For a significant majority of those with concussion, such effects are transient, resolving within weeks.³ Concerns surrounding cumulative consequences of repetitive concussions, however, have captured the public's attention recently.⁴

The conclusions of prior research concerning repeated concussions have been equivocal. To date, research has relied predominantly on athletes self-reporting concussion history, either during baseline neurocognitive testing^5

–9 or after sustaining a concussion.^9

–12 Of the studies focusing on how concussion history affects baseline assessments, outcomes have been mixed. While some studies found higher symptom report⁵ or lower neurocognitive performance^6,7 in those with a history of concussion, other researchers found no differences⁸ between those with and without self-reported history.

Studies that have evaluated athletes after self-reporting repeated concussion have been similarly inconsistent. One study found that concussed athletes with a self-reported history of three or more concussions performed worse on neurocognitive testing at eight days post-injury.⁹ Similarly, Guskiewicz and colleagues¹⁰ found that college athletes with self-reported concussions were more likely to sustain another concussion and more likely to have slowed recovery.

In contrast, another study¹¹ prospectively followed more than 2000 collegiate athletes and compared presentations of athletes after sustaining a second concussion with athletes presenting after sustaining their first concussion, finding that neuropsychological performance and symptom burden were not significantly different across the two groups.

Finally, a more recent study utilized a within-subjects design of adolescents who had sustained two concussions, finding no differences in symptom burden, neurocognitive performance, or days to clearance from the first to the second concussion.¹²

The current study aims to help clarify this question by applying a novel statistical approach (i.e., growth curve modeling) to evaluate outcomes in patients who presented to a specialty clinic for concussion management of multiple injuries. The broad aim of this study was to evaluate how concussion outcomes and presentation changed within patients who were evaluated at a concussion specialty clinic multiple times with a concussion.

Within this broad aim, we had two objectives. First, (1) we aimed to evaluate change in both recovery and initial symptom burden across the injuries, including whether the change was negative/worse with subsequent injuries and cumulative in nature (i.e., worse with more injuries). Our second (2) objective was to evaluate which variables predicted any change that did occur in evaluation or outcome variables. In this regard, we sought to evaluate whether severity at first injury (across all variables), demographic variables (e.g., gender), obtained medical history, or initial presentation characteristics (e.g., time to evaluation, symptom burden) predicted the change.

Based on research noted above,¹² our hypotheses were that initial presentation severity and time to recovery would stay relatively unchanged or worsen across injuries. We did not generate specific hypotheses with regard to which variables might predict the change, given the exploratory nature of these analyses and given that this has not yet been investigated.

Growth curve modeling

In contrast to previous studies, the current study used growth curve modeling to test the change across time points, which offers a number of advantages over more traditional methods (e.g., general linear models) when there are at least three time points. Specifically, by estimating a curve to the change in data over time, growth curves allow researchers to directly assess the amount and shape (e.g., linear; quadratic) of change across assessments.

Further, growth curve modeling can control for and assess the impact (if any) of intercepts (i.e., value at first measurement/concussion), time-dependent factors (e.g., age at each injury; time to initial evaluation at each injury); and time-independent factors (e.g., gender; age at first injury; previous diagnoses). Put plainly for the purposes of this study, growth curves can directly test whether outcomes worsened (or improved) across multiple injuries, as well as whether such change was impacted by time-varying or time-invariant factors.

From a terminology standpoint, “intercept” refers to the estimated values at the first measurement, and “slope” refers to the amount of average linear change across each time point. “Shape” is used instead of slope when the change is not estimated as linear and refers to the total amount of change from the first to the last measurements (i.e., change between first and last concussion). Covariate contributions are expressed as regression coefficients and express how much the intercept or slope varies conditioned on the covariate (e.g., how much does age at first injury impact how much worse [or better] are outcomes from subsequent injuries).

Methods

Participants, procedures, and materials

A retrospective chart review gathered records from 202 adolescents (46% female) between the ages of 10–21 (at initial injury) who presented to a concussion specialty clinic within 30 days of injury for multiple injuries from years 2011 to 2021. Participants and parents of minor participants consented to their electronic health records being retrospectively accessed.

Inclusion criteria included patients who had been seen in the same specialty clinic for two or three concussions. Records were excluded for history of previous concussion, moderate or severe traumatic brain injury, previously diagnosed vestibular or neurological disorder, presentation to clinic after 31 days post-injury, or sustained a reinjury over the course of their treatment for any given injury.

Participants all presented to a specialty clinic after sustaining a head injury that was diagnosed as a concussion by a licensed healthcare professional (e.g., neuropsychologist, physician, athletic trainer) trained in concussion, using concussion in sport group (CISG) criteria.¹³ Notably, the specialty clinic is multi-disciplinary with neuropsychological care, therapies (e.g., vestibular; oculomotor), and psychiatric care available on-site. This study was approved under an exempt medical records review protocol by the University Human Subjects Institutional Review Board.

Immediate Post-concussion Assessment and Cognitive Testing (ImPACT)

Participants completed ImPACT with the embedded Post Concussion Symptom Scale (PCSS) at initial visit for each concussion and at follow-up visits until clearance. ImPACT is a computerized neurocognitive test that comprises six subtests yielding four composite scores—verbal memory, visual memory, motor processing speed, and reaction time. ImPACT has been found to have high test-retest reliability among high school and college athletes with a one-year retest intraclass correlation coefficient (ICC) range of 0.62 – 0.85 and two-year retest ICC range of 0.43 – 0.74, respectively.¹⁴ The test takes approximately 25–30 min to complete.

PCSS

Participants completed the PCSS, embedded into ImPACT, at initial visit for each concussion and at follow-up visits until clearance. The PCSS is a brief, computerized symptom inventory. Participants are asked to rate their experience of 22 commonly self-reported concussion symptoms on a Likert scale from 0 (none) to 6 (severe). The PCSS has been shown to have a high internal consistency among concussed athletes with Cronbach of 0.8 – 0.9.¹⁵ The PCSS takes approximately 5 min to complete.

Definition of recovery

Recovery was defined as number of days between date of injury and date of clearance back to normal activities by the treating clinician. Clearance criteria included (1) symptom-free at rest (or return to estimated pre-injury baseline symptoms), (2) cognitive performance within baseline expectations, and (3) symptom free after exertion. All patients met these criteria after each injury before unrestricted return-to-play.

Analytic procedure

Recovery time (days) for each injury, days from injury to initial evaluation for each injury (“days to evaluation”), PCSS total symptom score at initial evaluation (“PCSS total”), and demographic/medical history (i.e., gender, age at first injury, self-reported history of concussion prior to first concussion treated in this specialty clinic, history of motion sickness, history of oculomotor diagnoses, history of migraine, history of anxiety, and history of depression) were included in the analyses.

A series of growth curve models were estimated on recovery time (days), PCSS total score, and days to initial evaluation. The modeling procedure occurred in the following sequence: (1) Initially, linear random-intercept random-slope (RIRS) models were estimated for each outcome to (a) test whether the measured change was linear and, if so, (b) estimate the parameters of change (i.e., estimates of intercept, slope, and random effects of these). (2) If this initial model did not fit, sources of misfit were identified and the model was modified, either by addressing local misfit or changing the specified curve (e.g., fixed effects; unspecified shape; see Results for specific modifications).

(3) Once adequate-fitting models were established for each outcome (see fit criteria below), the following time-invariant covariates were included in the models for days to recovery and PCSS total score: History of migraine, anxiety, depression, oculomotor diagnoses, motion sickness, concussions sustained prior to their first specialty clinic evaluation, gender, and age at first injury. (4) Finally, to more specifically evaluate the effect of age of injury on recovery time, age was included as a time-varying covariate in the recovery time model. To control for the potential changes in concussion care or outcomes associated with the wide date range of collected data, year of first injury was included as a covariate for slope of days to recovery and intercept of PCSS scores.

To accommodate any non-normality in ratings, robust maximum likelihood estimation (MLR) was used for the analyses using Mplus statistical software version 8.7.¹⁶ Full information maximum likelihood was used to accommodate missing data in the estimation procedure. Three procedures were used to evaluate model fit: (1) comparative fix index (CFI), (2) Tucker Lewis Index (TLI), and (3) the root mean square error of approximation (RMSEA), and we followed fit criteria outlined by Brown (CFI/TLI >0.05; RMSEA <0.08).¹⁷

Results

Demographics

Demographic statistics are displayed in Table 1. Average ages at first, second, and third injuries were 13.17, 14.95, and 14.49, respectively. Notably, 202 participants sustained two concussions, of whom 68 sustained a third concussion. On average, the second injury occurred 638 days after the first injury, and the third concussion occurred 482 days after the second concussion. Between 75–80% of concussions were sport-related at each injury.

Table 1.

Participant Demographic Variables (N = 202)

Variable
	Injury 1	Injury 2		Injury 3
Age (years), M (SD)	13.17 (1.99)	14.95 (2.13)		15.49 (2.11)
Days between injuries M (SD)	638 (514)		481 (321)
Sport-related mechanism n (%)	158 (78.2%)	154 (76.6%)		55 (80.9%)
Days to initial evaluation M (SD)	6.45 (5.04)	5.66 (5.03)		5.41 (4.55)
Days to clearance M (SD)	39 (36)	35 (39)		29 (24)
PCSS Total M (SD)	29.86 (22.40)	26.93 (23.15)		26.36 (20.07)
	Total Sample
Sex (N female, % female)	93, 46%
Concussion history (n, % yes)	196, 97%
Ocular history (n, % yes)	21, 10.5%
Motion sickness history (n, % yes)	58, 29%
Depression history (n, % yes)	16, 8%
Anxiety history (n, % yes)	41, 20.5%
Migraine history (n, % yes)	67, 33.2%
ADHD or LD history (n, % yes)	33, 16.4%

SD, standard deviation; ADHD, attention-deficit/hyperactivity disorder; LD; M, mean; PCSS, Post Concussion Symptom Scale.

Growth curves

Estimation and fit

Fit for the growth curve models are displayed in Table 2. For days to recovery and PCSS total symptom score, fit was adequate for the random-intercept random slope model. For days to initial evaluation, unspecified change models (“level by shape”) were used because of non-linear growth whereby there was less change between injuries two and three than between injuries one and two. Fit was adequate for this unspecified change model, although the shape was fixed because of misfit (TLI/CFI and RMSEA above Brown's recommendations).

Table 2.

Goodness-of-Fit Statistics for All Models

Model	χ2(df), p-value	CFI	TLI	RMSEA (90% CI)
Days to recovery^*	.06(1), 0.81	1.000	1.000	.000 (.000 - .115)
PCSS total^*	.36(1), 0.55	1.000	1.000	.000 (.000 - .157)
Days to initial evaluation^**	.28(4), 0.99	1.000	1.000	.000 (.000 - .152)
Days to recovery with time-invariant covariates^*	10.32(10), 0.41	.991	.973	.013 (.000 - .079)
PCSS total score with time-invariant covariates^*	11.35(10), 0.33	.985	.956	.026 (.000 - .084)
Days to recovery with time-varying age covariate^*	.39 (1), 0.53	1.000	1.000	.000 (.000 - .158)

CFI, comparative fix index; TLI, Tucker Lewis Index; RMSEA, root mean square error of approximation; CI, confidence interval; PC, post-concussion symptom scale.

Random intercepts random slope linear model.

Level by shape model with fixed shape.

Table 3 denotes the specific models estimated for each variable. Put more practically, all estimates of “model fit” were within adequate ranges, which means the parameter estimates (discussed below) can be interpreted.

Table 3.

Parameter Estimates of Slopes, Intercepts, and Intercept-Slope Correlations

	Intercept (SD)	Average Rate of Change^† (SD)	Correlation
Days to recovery	39.07^* (24.48^*)	-4.62^* (14.02)	-.62^*
PCSS total symptom score	29.63^* (18.48^*)	-2.16^* (9.90^*)	-.56^*
		Total Change^‡ (SD)
Days to initial evaluation	6.45^* (2.12)	-1.06 (–)	–

SD, Standard deviation calculated from the unstandardized variance estimate of the model; PCSS, post-concussion symptom scale.

indicates significant at p < 0.05. —parameter not estimated because of nature of model (e.g., fixed estimate).

†

“slope” parameter—represents average linear change between each subsequent injury.

‡

“shape” parameter—represents total amount of change from injury 1 to injury 3.

Slopes, intercepts, random effects, and correlations

Table 3 shows average rate of change¹ (i.e., “slope”¹ ), total change¹ (i.e., “shape”), intercepts, and standard deviations of these estimated by the models, as well as standardized correlations of intercepts and slopes for those models with random effects. Slopes indicated that recovery time and PCSS scores linearly improved across injuries, with each subsequent concussion resulting in fewer days to recovery (-4.62, p < 0.047) and lower PCSS score at initial visit (-2.16, p = 0.05). Shape from the unspecified change (“level by shape”) model of days to initial evaluation indicated these days decreased across the injuries by about one day (-1.06), although this effect was not significant (p = 0.097) and not linear, with most of the change occurring between the first and second injuries (0.74).

For models with random effects, variability of slopes and intercepts across the sample are estimated as variances, but standard deviations (SD) were then calculated and reported from these variances for ease of interpretation. The only significant slope variance was for the PCSS total score slope (SD = 9.90, p = 0.03). Variances were significant (p < 0.05) for all intercepts except days to initial evaluation (p = 0.13).

Correlations of intercepts and slopes (only estimated for random-intercept random-slope models) are displayed in Figure 1; intercepts were negatively related to slope for recovery time (-.62, p = 0.005) and PCSS total (-.56, p < .0001), such that those with greater (i.e., worse) recovery time and symptom burden at initial visit for their first concussion experienced faster decrease (i.e., improvement) in recovery time and symptom burden at subsequent concussions.

FIG. 1.

Slope and intercept are negatively correlated for both recovery days and PCSS total, indicating that longer recovery time and higher symptom burden are associated with more improvement (less severe outcomes) in these variables at subsequent injuries.

Covariates

Model results for time-invariant covariates can be viewed in Table 4. Models including time-invariant covariates indicated that gender (.73 days, p = 0.002) and history of anxiety (.50 days, p = 0.036) were significantly correlated to the intercept for recovery time, such that females, those with a history of anxiety, and those who were older at initial injury had longer recoveries at the initial injury. Of note, age at initial injury was only significant as an unstandardized estimate (p = 0.039), but not standardized (p = 0.076). In addition, gender (.75, p < 0.001) and history of anxiety (.63, p = 0.004) were significantly correlated to the PCSS total score intercept, such that females and those with anxiety had higher PCSS scores at the initial injury.

Table 4.

Standardized/Unstandardized Regression Coefficients of Time-Invariant Covariates With Intercepts and Slopes

	Intercept		Slope
Variable	Recovery Days	PCSS Total	Recovery Days	PCSS Total
Gender	.73^/16.87^	.75^/14.04^	.48/6.76	-.28/-2.44
ADHD/LD	-.07/-1.59	-.16/-2.96	.36/5.18	.39/3.46
Age at First Injury	-.12/-2.81^*	-.03/-.52	.09/1.27	.01/.08
Previous Concussion	.26/6.04	.06/1.20	-.22/-3.20	-.19/-1.66
Motion Sick History	.33/7.56	-.06/-1.10	-.68/-9.63	.12/1.03
Ocular History	.49/11.24	-.15/-2.75	-.44/-6.26	.43/3.83
Migraine History	.29/6.81	.17/3.24	.01/.20	-.29/-2.54
Anxiety History	.50^/11.56^	.63^/11.84^	.17/2.44	.14/1.22
Depression History	.46/10.71	.31/5.72	-.30/-4.25	-.72/-6.32

Note. ^* indicates significant at p < .05.

No covariates had significant correlations with slopes from recovery time or PCSS scores. Including age at injury as a time-varying covariate of recovery days indicated no significant relationship at any of the three injuries (ps = 0.73 - 0.30). Our year of first injury covariate, added to control for any influence of our wide date range of collected data, was not significantly associated with slope of recovery days (β = .1, p = 0.217) or PCSS severity at first injury (β = -.005, p = 0.930), suggesting these results were not significantly conditioned on when a participant presented for the first injury.

Discussion

Prior research examining repeat concussions in adolescents has been inconclusive regarding recovery outcomes (e.g., longer vs. shorter recovery with subsequent injuries) and methodologically limited (e.g., self-report injuries without prospective measurement of recovery). The current study aimed to expand on the limited literature by applying a novel statistical approach to evaluate outcomes in patients who presented to a specialty clinic for concussion management of multiple injuries.

The primary finding of this study is that recovery time did not worsen across concussions and actually improved—adolescents were cleared for full activity more than four days faster at each subsequent concussion. Second, age at any injury was not significantly related to slope, suggesting the impact of age does not account for the improved recovery time across injuries.

Third, this study found that higher symptom burden and recovery days at the first injury were correlated with more improvement in these variables at subsequent injuries (i.e., more severe patients at initial injury experienced more improvement in severity at subsequent injuries). Fourth, this study found that while anxiety, younger age, and gender predicted outcomes at the first injury, no included categorical/invariant covariates significantly predicted the slope, suggesting even patients with known risk factors (e.g., anxiety, female sex, migraine history, etc.) had improved outcomes across subsequent concussions. Overall, the results of this novel study suggest that patients who present to a concussion specialty clinic have improved recovery trajectories after having repeated concussions.

We cannot glean from this analysis the exact mechanism for reduced recovery time for subsequent concussions. There are some plausible explanations, however, and it is important to interpret these findings within the context of the setting (e.g., a concussion specialty clinic with a multi-disciplinary team). Patients in this study had access to concussion-trained neuropsychologists, and, if required, in-house referrals for vestibular/ocular therapy and psychiatry. Recovery could have been improved given access to this high level of care.

An interesting and unexpected finding of our study was such that more severe first injuries were associated with greater improvement in recovery time at subsequent injuries. One possible hypothesis for this observation could be increased education and awareness after the first injury. Concussion is associated with increased mood and emotional symptoms,² and given extensive media attention on long-term effects of concussions, fear regarding the injury and whether recovery will be complete can contribute to these concerns. Increasing the patient's education and confidence that active treatments can help reduce symptoms and improve recovery time could have been associated with shorter recovery times for subsequent injuries.

Another hypothesis explaining this finding is that education that occurs at the initial concussion leads to effective behavioral responses at subsequent concussions. Specifically, patients often presented to clinic earlier at each subsequent concussion, thereby initiating the treatment program sooner. Given previous research suggesting outcomes improve with quicker presentation to specialty care,^18
–20 it may be that this behavior is contributing to improved outcomes at subsequent concussions. Second, after their subsequent concussions, patients may be implementing effective behavioral strategies (e.g., maintaining engagement with non-risk activity) learned after their first concussion, thereby improving their outcomes.

Last, patients may be reducing iatrogenic behaviors (e.g., cocooning) after their subsequent concussions based on the education provided at the initial concussion. Such a hypothesis of more effective behavioral responses supports the importance of effective and active care for concussed individuals, especially after their initial concussion. It is also possible that clinicians require more visits before clearance for an individual's first injury, thus prolonging number of days until recovery.

In the context of previous research suggesting cumulative effects of repeated concussion, it is likely that concussion care has improved, resulting in fewer repeat concussive injuries before an initial injury is resolved,²¹ and thereby negating cumulative effects that could result from repeat injuries in quick succession. Indeed, most of the research finding this effect was published before 2012,^5
–7 a time in which not even all US states had laws mandating no return-to-play the same day or need for medical clearance after a concussion.

Further, previous research suggesting lingering effects primarily used self-report, which could have led to bias of those misattributing dysfunction and nonspecific symptoms to previous concussions. Indeed, the “good-ole-days” bias has been documented in concussion patients by which they interpret their pre-injury functioning as better than the average person.²²

The finding that younger age, female sex, and prior anxiety diagnoses were related to poorer outcomes after first concussion is consistent with previous research.²³ It is interesting no known risk factors for higher symptom burden and longer recovery time were related to rate of improvement, suggesting the improvements in recovery time and symptom burden at subsequent concussions were unaffected by these risk factors.

Such findings suggest that although risk factors do have an impact on recovery time, they do not prevent the improvements in recovery time at subsequent injury. In other words, the findings suggest risk factors do not interact with concussion history to prolong recovery; thus, these findings suggests risk factors do not contribute to cumulative effect of concussion. It is also possible that a combination of active treatment and increased education and confidence in recovery reduce the impact of these pre-injury risk factors. Such a finding may be impactful in a clinical setting with setting appropriate recovery expectations for patients with risk factors.

Limitations

The current study has limitations worth considering. It is possible that patients who presented to the specialty clinic for their first injury had repeated concussions but did not return to the clinic. Similarly, patients with knowledge and experience may readily detect concussions, even when mild, prompting evaluation for injuries that may otherwise go undetected or untreated. As such, loss to follow-up and/or bias to present with mild injury that would otherwise have spontaneous recovery could be a concern.

Also, this study is only generalizable to patients who have access to specialty care; results may have been different in patients who sought other sources of healthcare (e.g., primary care physician, emergency department, etc.). The data from this study is also not generalizable to patients with more severe brain injuries, patients who may have had repeated occupational concussions (such as military personnel or police), or patients who have sustained four or more concussions.

Further, the results of this study found some large but non-significant effect sizes, suggesting the future investigations should look to replicate these findings, ideally with larger samples sizes. It is also important to note that although this study suggests there is not short-term cumulative effects (i.e., <8 years) of concussion, it does not have any bearing on longer term effects.

Conclusion

The current study is the first to apply growth curve modeling to evaluate outcomes in concussion patients who were prospectively evaluated at a specialty clinic after two or three injuries. Such a methodological approach overcame prior limitations of self-report biases, examined multiple demographic, medical, and time-to-treatment variables of interest affecting recovery outcomes, and provided updated data on repeat concussion outcomes in the context treatment advances in concussion management across the past decade.

In contrast to some older studies suggesting cumulative negative effects of repeat concussions, this study found recovery time improved at subsequent concussions relative to the first, and higher symptom burden and recovery days at the first injury was associated with improved recovery at subsequent injuries. Even patients with pre-injury risk factors—namely, younger patients, females, and those with anxiety in this study—demonstrated the same pattern of recovery after subsequent injuries. These findings collectively suggest that acute risk of cumulative burden or worsening recoveries may be lower than previously thought for up to three concussions when all injuries were managed by a multi-disciplinary concussion specialty clinic.

This study does not provider comparative data to other settings/clinics, and these data may not generalize to other settings. Future research will be needed to replicate this finding and evaluate the potential effect of care setting (i.e., specialty clinic compared with emergency department, primary care physician, or sports medicine physician), treatment strategies, interval between injuries, and timing of care.

Transparency, Rigor and Reproducibility Summary

This study was not pre-registered online. The analysis plan was conceived and executed by the primary author and reviewed by a secondary author with biostatistics background. Data were pulled from electronic health records up to the date of study conception, which limits bias from treatment decisions regarding recovery time in the study. All outcome measures used are publicly available. Data and analytic code are potentially available upon reasonable request.

Footnotes

Authors' Contributions

JP conceived of the study, conducted statistical analyses, wrote the initial draft, revised based upon co-authors' comments, and approved the final draft. LM and SE contributed to writing the initial draft, critically reviewing the draft, and approving the final version. AKK, AF, BM contributed to data extraction and organization, critically reviewing the draft, and approving the final version. AT, MC, and AK critically reviewed the draft and approved the final version.

Funding Information

There was no funding provided for this research.

Author Disclosure Statement

No competing financial interests exist.

References

DePadilla

, Miller

, Jones

, et al. Self-reported concussions from playing a sport or being physically active among high school students—United States, 2017. MMWR Morb Mortal Wkly Rep, 2018; 67(24):682–685

Harmon

, Clugston

, Dec

, et al. American Medical Society for Sports Medicine position statement on concussion in sport. Br J Sports Med, 2019; 53(4):213–225

Broglio

, McAllister

, Katz

, et al. The natural history of sport-related concussion in collegiate athletes: findings from the NCAA-DoD CARE Consortium. Sports Medicine, 2022; 52(2):403–415

Eagle

, Okonkwo

. Telling the whole story: bibliometric network analysis to evaluate impact of media attention on chronic traumatic encephalopathy research. J Neurotrauma, 2023; 40(1-2):148–154; doi:10.1089/neu.2022.0266

Schatz

, Moser

, Covassin

, et al. Early indicators of enduring symptoms in high school athletes with multiple previous concussions. Neurosurgery, 2011; 68(6):1562–1567

Iverson

, Echemendia

, LaMarre

, et al. Possible lingering effects of multiple past concussions. Rehabil Res Pract, 2012; 2012

Collins

, Grindel

, Lovell

, et al. Relationship between concussion and neuropsychological performance in college football players. JAMA, 1999; 282(10):964–970

Broglio

, Ferrara

, Piland

, et al. Concussion history is not a predictor of computerised neurocognitive performance. Br J Sports Med, 2006; 40(9):802–805

Covassin

, Moran

, Wilhelm

. Concussion symptoms and neurocognitive performance of high school and college athletes who incur multiple concussions. Am J Sports Med, 2013; 41(12):2885–2889

10.

Guskiewicz

, McCrea

, Marshall

, et al. Cumulative effects associated with recurrent concussion in collegiate football players: the NCAA Concussion Study. JAMA, 2003; 290(19):2549-2555

11.

Macciocchi

S–

, Barth

, Littlefield

, et al. Multiple concussions and neuropsychological functioning in collegiate football players. J Athl Train, 2001; 36(3):303

12.

French

, Jennings

, Eagle

, et al. A within-subjects comparison of clinical outcomes for patients' first and second concussions. J Head Trauma Rehabil, 2021; 36(2):114–119; doi:10.1097/HTR.0000000000000612

13.

McCrory

, Meeuwisse

, Dvořák

, et al. 5th International Conference on Concussion in sSort (Berlin). Br J Sports Med, 2017; 51(11):837

14.

Lovell

Immediate Post-Concussion Assessment Testing (Impact) Test: Clinical Interpretive Manual Online ImPACT 2007–2012. ImPACT Applications Inc.: Pittsburgh, PA; 2013

15.

Lovell

, Iverson

, Collins

, et al. Measurement of symptoms following sports-related concussion: reliability and normative data for the post-concussion scale. Appl neuropsychol, 2006; 13(3):166–174

16.

Muthén

, Muthén

Mplus User's Guide v7. 3. Los Angeles, CA; 2015

17.

Brown

TA.

Confirmatory Factor Analysis for Applied Research. Guilford Publications: New York, NY; 2015

18.

Corrado

, Willer

, McPherson

, et al. Adolescents with more oculomotor and vestibular signs of sport-related concussion benefit from aerobic exercise: an exploratory analysis. J Neurotrauma, 2023; 40(15-16):1718–1729; doi: 10.1089/neu.2022.0225

19.

Eagle

, Puligilla

, Fazio-Sumrok

, et al. Association of time to initial clinic visit with prolonged recovery in pediatric patients with concussion. J Neurosurg Pediatr, 2020; 26(2):165–170; doi:10.3171/2020.2.PEDS2025

20.

Eagle

, Womble

, Elbin

, et al. Concussion symptom cutoffs for identification and prognosis of sports-related concussion: role of time since injury. Am J Sports Med. 2020; 48(10):2544–2551; doi:10.1177/0363546520937291

21.

McCrea

, Broglio

, McAllister

, et al. Return to play and risk of repeat concussion in collegiate football players: comparative analysis from the NCAA Concussion Study (1999–2001) and CARE Consortium (2014–2017). Br J Sports Med, 2020; 54(2):102–109

22.

Lange

, Iverson

, Rose

. Post-concussion symptom reporting and the “good-old-days” bias following mild traumatic brain injury. Arch Clin Neuropsychol, 2010; 25(5):442–450

23.

Iverson

, Gardner

, Terry

, et al. Predictors of clinical recovery from concussion: a systematic review. Br J Sports Med itish, 2017; 51(12):941–948