Article Text
PDF
Abstract
Objective Concussion symptoms can be clustered into domains and understanding how multiple symptom domains present clinically may guide more accurate interventions. We investigate the associations between concurrent symptom domains and clinical recovery outcomes, as well as the role of sex in these relationships.
Methods We analysed data from the Ivy League–Big Ten Epidemiology of Concussion Study and included sport-related concussions (SRC) across five academic years 2015–2016/2019–2020 with complete data (n=1160). We used symptoms from the Sport Concussion Assessment Tool 22-symptom evaluation, previously categorised into symptom domains. Symptom profiles characterise how athletes endorse concurrent symptom domains. Outcomes are time (in days) from SRC to symptom resolution, return to academics, and full play.
Results Females more commonly endorsed headache, sensory, and affective symptom domains. Four classes/symptom profiles emerged: (1) ‘low’ on all domains, (2) ‘high’ on headache and sensory domains, (3) ‘high’ on vestibulo-ocular, cognitive, and sleep domains, and (4) ‘high’ on all domains. Time to symptom resolution, return to academics, and return to play were consistently shorter among class/symptom profile 1 compared with other classes/profiles. Compared with class/profile 1, the chance of having symptoms resolve was lower among classes/profiles 2, 3, and 4 (HR 0.74, 95% CI 0.63 to 0.88; HR 0.74, 95% CI 0.60 to 0.92; HR 0.50, 95% CI 0.43 to 0.57, respectively). Results were similar for return to academics and full play outcomes. Interactions with sex were not statistically significant.
Conclusions Four symptom profiles characterised how concussion symptom domains co-occur. We found differences in recovery timelines among these groups, but not by sex. Findings inform and support targeted, symptom domain-specific interventions in concussion management.
- Brain Concussion
- Athletes
- Epidemiology
- Cohort Studies
- Neurology
Data availability statement
No data are available. Data are not publicly available due to current data sharing policies of the Ivy League-Big Ten Epidemiology of Concussion Study.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Evidence demonstrates that symptoms represent important clinical characteristics associated with concussion recovery timelines and may differ by sex. Investigations into how individual symptoms cluster into domains have been largely descriptive, and so concurrent symptom domains and associations with clinical and sport-related outcomes is not known.
WHAT THIS STUDY ADDS
This study contributes new knowledge about symptoms after concussion by describing how collegiate student-athletes fit into distinct classes, representing how they experience concurrent symptom domains (‘symptom profiles’). We found four classes/symptom profiles emerged to characterise the ways in which symptoms co-occur in student-athletes in our study. We also found some symptom profile experiences to be associated with lower chances of having recovery outcomes over time. Differences in how symptom profiles related to outcomes were not apparent by sex.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
These findings provide an evidence base to inform the development of targeted, symptom-specific interventions in the management of concussion. The results also suggest symptom subgroups in need of additional research.
Introduction
Sport-related concussion (SRC) is relatively common, accounting for 6.2% of injuries among athletes in the National Collegiate Athletic Association (NCAA).1 The symptoms athletes experience following a concussion vary and can be debilitating, and require costly resources and treatment modalities over weeks or months.2–4 A growing interest clinically and among researchers has emerged to better understand, develop, and direct targeted treatments based on the presence of specific concussion symptoms.5 Collegiate athletics specifically represents an important period for study and intervention on concussion symptom management since the resources and treatment athletes receive for injuries at this time may have important academic and sporting implications, and possibly more lasting health effects.6–10
Evidence demonstrates that symptoms represent the most important clinical characteristics associated with concussion recovery timelines.11–15 However, most prior studies have focused primarily on the reported symptom burden (ie, the number of symptoms) following concussion as a primary exposure,11–15 rather than symptom type. Individualised management of concussion guided by symptom domain presentation is an evolving strategy in SRC care.16 Understanding the nature of symptom experiences (eg, as affective vs vestibular) more precisely guides rehabilitation efforts to combat persisting postconcussion symptoms. Investigations into how individual symptoms may cluster together into symptom domains have been largely descriptive17–20 and have not considered associations between symptom domains and time to clinical, academic, and sport-related outcomes along the return-to-play trajectory.
The Sport Concussion Assessment Tool (SCAT) is widely employed by clinicians during initial evaluation, and to aid in return-to-learn and return-to-play decision-making,21–23 and yet only a few studies have investigated symptom domains among SCAT symptoms.17–19 24 25 Additional limitations of prior studies include small, heterogeneous cohorts and/or convenience samples, lack of adjustment for potential confounders, and lack of consideration of sex.24 26 27 Considering the role of sex in particular in these analyses is important, given prior and conflicting evidence that symptom prevalence and symptom resolution timelines may differ in male and female athletes,28–36 warranting further examination. Motivated by agreement that multiple, overlapping symptom domains are common in clinical presentations of concussion, prior studies have investigated the co-occurrence of symptom domains, but with descriptive approaches.5 The relationship between athletes’ symptom experiences and clinical recovery and return-to-learn and return-to-play outcomes following concussion remains unclear, warranting further study.
In this study, we were motivated by prior investigations which found that a six-factor symptom structure best described their symptom data, using a bifactor model,25 inclusive of all common concussion assessments.19 We also build on our recent work where we found that a similar six-domain symptom structure best represented the SCAT symptom data and that measured symptoms represent domains similarly for males and females.37 Here, we aim to describe how athletes may endorse multiple, co-occurring symptom domains, herein called ‘symptom profiles,’ among a large cohort of collegiate athletes with SRC. We consider the association between symptom profile and recovery timelines, as time from concussion injury to the outcomes of (1) symptom resolution, (2) return to academics, and (3) return to full play. We also consider the role of sex in these relationships.
Methods
Study design and study population
We used data from the Ivy League–Big Ten Epidemiology of Concussion Study (Ivy–B1G Study), a prospective, multisite, observational surveillance study of student-athletes with concussion across 20 participating Ivy League and Big Ten universities.38 Detailed methods and descriptions of the Ivy-B1G cohort study have been outlined previously.39 This study follows the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines. Analysis and presentation are also consistent with the CHecklist for statistical Assessment of Medical Papers (CHAMP) statement.40
For the present analysis, we included SRCs reported in varsity sports across five consecutive academic years; 2015–2016 to 2019–2020 (February 2020), as all variables of interest were collected during these years. As previously described, in the Ivy-B1G Study, SRC is defined by a clinical diagnosis according to International Consensus criteria using the most recent Concussion in Sport Group (CISG) consensus statement and SCAT at the time the SRC occurred.39 41 Of 1904 total SRCs, we excluded 739 missing outcome data, and 5 missing covariate data, resulting in 1160 complete SRC records included in analyses. We previously performed two sensitivity analyses to investigate the potential impact of missing data among the 739 cases with missing outcome data (since notably, symptom resolution and academic return are also conceptualised as covariates in the return to full play model). We examined HR estimates, 95% CI, and SEs, and found them to be similar. This reassured us against the influence of bias, and we moved forward with our complete case cohort (n=1160 SRCs). These methods to understand and address missing data in our sample have been published previously.36 Data collection for the 2019–2020 academic year was through February 2020 due to the COVID-19 pandemic. Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Demographic and clinical characteristics
Class year is defined as the academic year in school (freshman, sophomore, junior, senior, and fifth year), and age is reported as an integer as the student-athlete’s age at the time of the SRC.39 27 sports are represented in the Ivy-B1G study and we categorised them into high/low contact sport type, consistent with previous work33 36 and to create a binary variable with two groups of closer relative equal size for comparison. Concussion history is defined as the number of self-reported diagnosed concussions prior to the current SRC injury. Academic accommodations is defined as whether or not the athlete received academic accommodations (ie, adjustments or modifications to their academic work) during their concussion recovery.
Equity, diversity and inclusion statement
Our study included all identified cases of SRC in collegiate student-athletes inclusive of all genders, race and ethnicities, and socioeconomic levels. Of note, in the present study, we aimed to be mindful concerning the appropriate use of the terms sex and gender; whereas sex is based on biological or physiological attributes, gender refers to a social construct, and both characteristics affect health.42 Here, we present our results referring to sex (‘male; female’) and also using the terms ‘women; men’ and ‘athletes on men’s/women’s teams’ throughout to reflect how data were collected in the Ivy–B1G Study during the study period.39
SCAT symptom domains
Symptom data from the Ivy-B1G Study consist of the 22 postconcussion symptoms measured in the SCAT symptom evaluation.39 The Ivy-B1G Study kept apprised to new versions of the SCAT over time (SCAT3 and then SCAT5) as clinical practice adapted to new guidelines and tools across study sites. The symptoms collected were the same throughout the SCAT3 and SCAT5 symptom evaluations during the study period (2015–2020). In the Ivy-B1G Study, SCAT data are collected as a dichotomised symptom inventory. That is, whether the athlete experienced the symptom at any time throughout their recovery, and not individual symptom severity. We created a symptom count variable, defined as the total number of symptoms reported over the course of SRC recovery. Symptoms are reported from the time of reported injury through symptom resolution, consistent with clinical practice and the Ivy-B1G Study protocol.39
In previous work, we found that a six-domain structure best represented our symptom data,37 consistent and similar to prior investigations,19 25 despite different methodological approaches. The six-symptom domains are as follows: headache, vestibulo-ocular, sensory, cognitive, sleep, and affective19 25 37 (online supplemental appendix table 1). To determine the prevalence of each symptom domain, we created a binary variable by splitting each domain factor scores (z-scores) at 0. Scores greater than 0 represent endorsement of that symptom domain.
Supplemental material
Recovery outcomes
Our recovery outcomes are consistent with definitions from the Ivy-B1G Study.39 Time to the outcome symptom resolution was defined as the number of days between the date of SRC injury and the date when the athlete self-reported being symptom-free (ie, returned to baseline symptoms). Time to return to academics was defined as the number of days between SRC injury and the first day the athlete returned to academic coursework, in class.39 41 Time to return to full play was defined as the number of days between SRC injury and the date the athlete was medically cleared to return to full athletic participation.33 39 41 Both time to return to academics and full play are based on determinations made by athletic trainers and team physicians. Each recovery outcome is relevant to one of the stages in the return-to-sport and return-to-academics strategies in the 2017 CISG consensus statement and 2014 NATA position statement on concussion.4 21 41
Statistical analysis
We summarise participant characteristics overall and stratified by sex as median and 25th–75th percentiles for continuous variables, and as counts and proportions for categorical variables.
To identify symptom profiles, we used latent class analysis on our previously identified six symptom domains37 (online supplemental appendix table 1). We first estimated a two-class model and then added classes until we identified the model with the best fit. We examined model fit based on our understanding of concussion clinical symptom presentations and the statistical criterion BIC, with lower BIC indicating better model fit.43 44 The BIC metric is recommended as a parsimonious approach to determining latent classes, where strong penalties are imposed for model complexity, particularly in large samples.43 We report the prevalence and qualitative descriptions of each class/symptom profile. We determined the likelihood of membership in each class and estimated class marginal means. We then generated a new random variable from the posterior class membership probabilities.
We used Cox proportional hazards regressions to estimate HRs for the association between class/symptom profile and time from date of injury to outcomes of (1) time to symptom resolution, (2) return to academics, and (3) return to full play. Student-athletes were followed from the time of SRC to the outcome of interest, lost to follow-up or administrative censoring (n=7 (<1%)) on 1 June of the academic year of injury (1 March for 2019–2020 due to the COVID-19 pandemic).45 1 June of each academic year was chosen since athletic training care, and therefore, surveillance for the Ivy-B1G study is less consistent and reliable over the summer months.39 We assessed whether the proportional hazards assumption was satisfied graphically by inspecting log-log plots and Schoenfeld residuals.46–50 We also assessed whether the linearity assumption was satisfied by plotting and examining Martingale residuals against our continuous exposure variables.51 We use Kaplan-Meier curves with log-rank tests to show any overall differences in time from date of injury to each of the three outcomes by class/symptom profile. We then proceeded with our three models by first estimating the unadjusted association between class/symptom profile and the outcome. Second, we adjusted for sex, and we inspected for effect modification by sex using the Wald test on the regression coefficient estimate for the interaction between class/symptom profile and sex. Third, we adjusted for sex and additional available a priori-specified covariates and confounding variables including class year, symptom count, concussion history, and academic accommodations. We identified confounders using causal directed acyclic graphs. As an alternative approach, we evaluated for multiplicative and additive interactions between class/symptom profile, which we binarised—class/profile 1 (reference group) vs class/profiles 2, 3, 4—and sex.52 53 We evaluated for multiplicative interaction by inspecting for HR>1 on the interaction coefficient of class/symptom profile and sex, indicative of multiplicative interaction.52 54 We evaluated for additive interaction and calculated the relative excess risk due to interaction (RERI). An RERI>0 is indicative of additive interaction.52–55
To investigate if any particular symptom domain was driving the relationships between classes/profiles and outcomes, we evaluated the relationship between each symptom domain and outcome. We used Cox proportional hazards regressions to estimate HRs for associations between each of the six symptom domains and a symptom domain composite measure (endorsed one or more domains vs no domains) with the three recovery outcomes. Similar to our primary analysis, we estimated unadjusted associations first, then adjusted for sex and inspected for interaction between symptom domain and sex using the Wald test, and then adjusted for sex and a priori covariates and confounders. We again evaluated for multiplicative and additive interactions between symptom domain and sex by inspecting for HR>1 and then calculating and inspecting the RERI.52–55
A two-sided p<0.05 was set as the threshold of significance for each hypothesis test. Statistical analyses were performed using StataBE V.18.1 (StataCorp).
Results
Description of the study population
A total of 1160 SRCs (male, n=667; female, n=493) occurred between 2015 and 2020 across participating sites (table 1). The mean age at time of SRC overall was 19.8 (SD: 1.3) years. The majority of SRCs occurred among athletes playing a high-contact sport (78.0%), but this was different by sex. Of the SRCs in females, 67.5% were among high-contact sport athletes; whereas that was higher at 90.6% among males. For more than one-half of student-athletes (54.4%), this was their first SRC documented, including prior to collegiate play. Roughly one-half of student-athletes received academic accommodations (49.8%) after their SRC. Although the mean symptom count (SCAT symptom evaluation) was 11 (SD: 5.2) for both males and females, the prevalence of some symptom domains differed significantly between females and males. Compared with males, females were more likely to endorse the following domains: headache (70.8% vs 63.3%), sensory (58.2% vs 52.2%) and affective (50.5% vs 41.1%) (table 1).
Characteristics of SRC in Ivy-B1G Study student-athletes (2015–2016 to 2019–2020)
Time to recovery outcomes
Cumulative time to symptom resolution, return to academics and full return to play were consistently shorter among those endorsing class/symptom profile 1 compared with the other classes/profiles (figure 1). Median time to symptom resolution by class/symptom profile was as follows: class/profile 1: 4 days (IQR: 2–8); class/profile 2: 6 days (IQR: 4–12); class/profile 3: 6 days (IQR: 3–10); class/profile 4: 10 days (6–17). Median time to return to academics was as follows: class/profile 1: 4 days (IQR: 2–7); class/profile 2: 5 days (IQR: 3–11); class/profile 3: 5 days (IQR: 3–9); class/profile 4: 8 days (IQR: 5–15). Median time to return to full play: class/profile 1: 11 days (IQR: 8–18); class/profile 2: 13 days (IQR: 9–21); class/profile 3: 12 days (IQR: 9–19); class/profile 4: 18 days (IQR: 12–27).
Cumulative time from injury to (A) symptom resolution, (B) academic return, (C) full play return by class/symptom profile.
Description of classes/symptom profiles
Combined with our clinical understanding of concussion symptom presentations, the four-class model corresponded to the lowest BIC value (BIC=7014.39) compared with the three class (BIC=7062.38) and two class (BIC=7250.77), indicating the best fit. Student-athletes in class/profile 1 were ‘low’ on all symptom domains. Those in class/profile 2 were ‘high’ on headache and sensory domains. Those in class/profile 3 were ‘high’ on vestibulo-ocular, cognitive, and sleep domains. Those in class/profile 4 were ‘high’ on all symptom domains. Latent class/profile marginal probabilities and marginal means are presented in online supplemental appendix table 2. Full descriptions and the prevalence of each class/profile are shown in table 2.
Description and prevalence of classes/symptom profiles
Symptom resolution, return to academics and return to full play
Estimated HRs for the association between class/profile and time to symptom resolution for each model specification, as well as for covariates included to control for potential confounding are presented in table 3. In the unadjusted model, those in classes/profiles 2, 3, and 4 were less likely to have symptoms resolve over time compared with those in class/profile 1 (HR 0.72, 95% CI 0.61 to 0.85; HR 0.74, 95% CI 0.69 to 0.92; HR 0.50, 95% CI 0.43 to 0.57, respectively). Associations were similar with adjustment for sex, and we found no evidence of interaction by sex in the association of class/profile and time to symptom resolution (online supplemental appendix table 3). Associations were similar with adjustment for class year, sport contact level, and concussion history, and remained statistically significant (HR 0.74, 95% CI 0.63 to 0.88; HR 0.74, 95% CI 0.60 to 0.92; HR 0.50, 95% CI 0.43 to 0.57, respectively). For return to academics, those in classes/profiles 2, 3, and 4 were less likely to have returned to academics over time compared with class/profile 1 (HR 0.73, 95% CI 0.61 to 0.87; HR 0.74, 95% CI 0.59 to 0.94; HR 0.52, 95% CI 0.45 to 0.61, respectively). Associations were similar with adjustment for sex, and we again found no evidence of interaction by sex in the association between class/profile and time to return to academics. Associations were attenuated with adjustment for class year, sport contact level, concussion history, and academic accommodations (HR 0.82, 95% CI 0.69 to 0.98; HR 0.85, 95% CI 0.68 to 1.07; HR 0.60, 95% CI 0.52 to 0.70, respectively). For return to full play, those in classes/profiles 2, 3 and 4 were less likely to return to full play over time compared with those in class/profile 1 (HR 0.82, 95% CI 0.70 to 0.97; HR 0.92, 95% CI 0.74 to 1.14; HR 0.60, 95% CI 0.52 to 0.69, respectively). Associations were again similar with adjustment for sex, and we found no evidence of interaction by sex in the association of class/profile and time to return to full play. Associations were similar with adjustment for class year, sport contact level, and concussion history (HR 0.84, 95% CI 0.71 to 0.98; HR 0.92, 95% CI 0.74 to 1.14; HR 0.60, 95% CI 0.52 to 0.69, respectively). In our approach evaluating for multiplicative and additive interactions between symptom profile and sex, we found no evidence supporting either a multiplicative or an additive interaction (online supplemental appendix tables 4 and 5).
HRs for association between symptom profile and time to recovery outcomes
Secondary analysis, symptom domains
Associations of individual symptom domains with each of the three recovery outcomes were similar to our primary analysis, whereby the presence of symptoms across all domains was associated with a longer time from injury to symptom resolution, to return to academics, and to return to full play (table 4). We found no evidence of multiplicative or additive interaction by sex in the association of symptom domains with outcomes (online supplemental tables 6–8).
HRs for association between symptom domain and time to recovery outcomes
Discussion
This study investigated concurrent symptom domains (ie, ‘symptom profiles’) and was especially novel in considering associations between symptom profiles and recovery timelines among collegiate athletes with SRC. Four symptom profiles emerged as best representing how athletes endorse multiple symptom domains in our sample: 1: ‘low’ on all symptom domains, 2: ‘high’ on headache and sensory domains, 3: ‘high’ on vestibulo-ocular, cognitive, and sleep domains, and 4: ‘high’ on all symptom domains). Across our three outcomes, we found that compared with those in class/profile 1, those in classes/profiles 2, 3, and 4 were less likely to have each outcome over time, which persisted after adjusting for a priori clinical and demographic characteristics. Importantly, we also found no evidence of interaction between class/profile and sex in any of our models, suggesting that male and female athletes who report similar types of symptoms experience similar recovery trajectories following SRC.
Research implications
To date, a limited number of studies investigating how symptoms cluster onto domains have used symptom data reported via the ImPACT (Post-Concussion Symptom Scale (PCSS-22)) assessment, Post-Concussion Symptom Inventory (PCSI) and SCAT,20 56 57 all of which are well-known concussion assessment tools. However, few studies have included collegiate athletes in their sample and measured symptoms with the SCAT symptom inventory,17 24 25 as we do here. Our ability to leverage data from a large, homogeneous sample of collegiate student-athlete concussions (n>1,000) through the Ivy-B1G Study is a strength, given the high risk of concussion among collegiate athletes and the unique setting in which collegiate athletes experience care for injuries. This is critical, as the care that collegiate athletes receive following any injury, especially a concussion, has the potential to impact their health throughout their lifetime, specifically with regard to mental health, physical activity, mobility, and quality of life.6–9
Our study extends prior research exploring symptom domains by considering associations between endorsing particular symptom domains and recovery outcomes along return-to-play trajectories. We also assess and adjust for available, a priori clinical and demographic characteristics in our models, including sex, to reduce confounding. We additionally formally test for interaction by sex in our models. This is an important consideration, given some evidence that symptom burden (the number of symptoms) and time to symptom resolution may differ by sex.28–34 58 Recently, studies from larger cohorts have observed no statistically significant differences in timelines to recovery outcomes by sex,35 36 yielding a growing amount of conflicting evidence regarding sex differences in the literature. Here, we did not find evidence of a significant statistical interaction between symptom domain and sex or symptom profile and sex in our analysis. Therefore, sex did not appear to play a significant role in the relationship between symptoms and time to symptom resolution, return to academics, and return to full play. Future work should build off our approach and findings, and consider symptom severity and/or symptoms collected over time.
Clinical implications
Our study also adds to the literature by describing how student-athletes with SRC fit into distinct profiles, representing how they experience concurrent symptom domains (‘symptom profiles’). We then further consider associations between symptom profile and time to recovery outcomes along return-to-play trajectories. We found that four classes/profiles emerged to show the several characteristic ways in which symptoms co-occur in athletes in our study. We also found that, compared with symptom profile 1 (low on all symptom domains), those with symptom profile 2 (headache and sensory), 3 (vestibulo-ocular, cognitive, and sleep) and 4 (high on all symptom domains) were associated with a lower likelihood of having complete recovery outcomes over time. Together, these findings reveal symptom subgroups (ie, those with headache and sensory profiles; vestibulo-ocular, cognitive, and sleep profiles; and those endorsing multidomain symptom profiles) as areas to direct interventions to help inform personalised, symptom-targeted treatments following concussion. Such examples of personalised concussion treatment could include oculo-motor treatments targeting athletes experiencing vestibulo-ocular symptoms,59 60 or engaging psychologists and psychiatrists to treat existing and/or exacerbated affective symptoms.61 Findings may also inform efforts targeting prolonged recovery, including decisions regarding resource allocation in athletic medicine departments, and the organisation and mobilisation of clinical care teams aimed at personalised medicine for athletes with concussion. Outside of athletic medicine, results could even inform how the field views the inclusion and utilisation of academic resources and support as part of the recovery process. Of note, the four symptom profiles found here are categorised by symptom burden (low/high) across symptom domains. The latent classes we observed represent distinct symptom profiles and may also reflect previous findings that greater symptom burden (regardless of profile) is associated with clinical outcomes. Our findings also reinforce evidence of opportunities for intervention for those with many diverse symptoms.
Limitations
Our study makes novel contributions to the literature, but has some limitations. First, our findings may not be generalisable to other collegiate athletic conferences and levels of play. Future work in these and other diverse populations is needed. The Ivy-B1G Study procedures are standardised across the 20 participating sites via a central study protocol, but variability within and across sites in the study sites may remain. Second, for 54% of student-athletes in our population, this was their first concussion documented. However, we are unable to know if some individuals in the study had multiple concussions during the study period (2015–2020), meaning their concussion characteristics and outcomes may be correlated, and this is a limitation. Still, those athletes with more than one concussion likely comprise a very small portion of the sample. Third, symptom data from the Ivy-B1G Study consist of the 22 postconcussion symptoms measured in the SCAT and are collected dichotomously; that is, whether the athlete experienced the symptom at any time throughout their recovery. In clinical practice, the SCAT symptom inventory is administered as a 22-item, 7-point Likert scale ranging from none (0) to severe (6). Future studies could consider symptom severity using the SCAT. Finally, we were limited to those variables collected as part of the Ivy-B1G Study, and the possibility of residual confounding remains due to the observational nature of this study.
Conclusions
This study found four classes/symptom profiles that emerged to represent how symptom domains co-occur in our sample of collegiate athletes with SRC. We found statistically significant differences in the timelines to recovery outcomes among these symptom-profile groups, and we found no evidence of interaction between symptom profile and sex. These findings inform and underscore the need for the development of targeted, symptom-domain-specific interventions in the management of concussion.
Data availability statement
No data are available. Data are not publicly available due to current data sharing policies of the Ivy League-Big Ten Epidemiology of Concussion Study.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants, with the University of Pennsylvania Institutional Review Board (IRB) serving as the central IRB. The Federalwide Assurance (FWA) number is #00004028 and the IRB protocol number is #833372. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We acknowledge study personnel across participating Ivy League and Big Ten campuses, and we acknowledge the student-athletes participating in the Ivy League-Big Ten Epidemiology of Concussion Study. Key athletic staff, research staff, and faculty study contacts at each site include (listed alphabetically by institution) Bailey Lewis, MS, ATC (Brown University); Chad Mineo, MS, ATC and Theodore Cowling, MS, ATC (Columbia University); Lauren Rudolph, MS, ATC (Cornell University); Benjamin Schuler, MS, ATC and Yuri Fujioka, MS, ATC (Dartmouth College); Rebecca Coleman, MS, ATC, Samantha Yoke, MS, ATC, Yumi Kuscher, MS, ATC (Harvard University); Anne Danbury, PhD, MS, ATC and Aaron Anderson, PhD, MS (University of Illinois); Lauren Wilkins (Indiana University); Kathryn Berger, MA, ATC (University of Iowa); Kiersten Ann Janjigian, MA, MS, Angela Vaysman and Katelyn Engen, MEd, ATC (University of Maryland); Theo Belhomme, MAT, ATC and Vincent Delvalle, MS, ATC (Michigan State University); Leonard Navitskis, MS, ATC and David Millward, MD (University of Michigan); Suzanne Hecht, MD and Joi Thomas, MS, ATC (University of Minnesota); Heather C Bouchard, MA (University of Nebraska-Lincoln); Francesca Docters (Northwestern University); Jenna Ratka, MS, ATC, Philip Samko, MS, ATC, Anthony Braun, MS, ATC, Cory McMillen, MS, ATC, Christina Kossak, DAT, MBA, Mike Burkeitt, MS, ATC, Vic Szwanki, MS, ATC, Alessandro Vecchi, MS, ATC (University of Pennsylvania); Emily A Dorman, MEd, ATC and Bridget Hunt, MS, ATC (Princeton University); Carly Day, MD (Purdue University); Kyle Brostrand, MS, ATC, John Taggart, MSEd, ATC, and Evan Lobato, ATC (Rutgers University); Jordan Lockman, MS, ATC (Yale University). The authors acknowledge additional study leadership: Carolyn Campbell-McGovern, MBA and Robin Harris (the Ivy League), and Kerry Kenny and Jim Borchers, MD, MPH (Big Ten Conference). The authors give thanks to members of the Study Advisory Committee: Art Maerlender, PhD, ABPP-CN; Cary R. Savage, PhD; Emily A. Dorman, MEd, ATC; James C. Torner, PhD, MS; Jeffrey M. Mjaanes, MD; James M. Noble, MD, MS; Carrie Esopenko, PhD. The authors also thank Andrew Belfiglio, MPH, Ashley Rettmann, and Nichole Burnside, MBA for study management and coordination.
References
Footnotes
X @berndalonzo
Collaborators Ivy League-Big Ten Epidemiology of Concussion Study Investigators (listed alphabetically by institution): Beth Conroy (Brown University); Thomas Bottiglieri (Columbia University); Amy Sucheski-Drake and Kathryn J Harris (Cornell University); Kristine A Karlson and Jonathan D Lichtenstein (Dartmouth College); Arun J Ramappa (Harvard University); Randy Ballard (University of Illinois); Nicholas L Port (Indiana University); Andrew R Peterson (University of Iowa); Bradley D. Hatfield (University of Maryland); Mathew R Saffarian (Michigan State University); Abigail C Bretzin and James T Eckner (University of Michigan); Erin Moore and Suzanne Hecht (University of Minnesota); Cary R Savage and Kate Higgins (University of Nebraska-Lincoln); Matthew J Nerrie (Northwestern University); Anthony Erz and Brian J Sennett (University of Pennsylvania); Michael Gay (Pennsylvania State University); Sasha Steinlight (Princeton University); Scott Lawrance (Purdue University); Jason Womack and Carrie Esopenko (Rutgers University); Elizabeth C Gardner (Yale University).
Contributors BAD and DJW conceived of the idea for the manuscript. BAD performed analyses and interpreted results, and wrote the initial draft with input from ALCS and DJW. IJB, CLM, and RHH provided feedback and edited later drafts. All named Ivy League-Big Ten Study Investigators were given opportunities to provide feedback on the final draft. All authors read and approved the final version of the manuscript. BAD and DJW are the guarantors. BAD and DJW had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Funding The Ivy League-Big Ten Epidemiology of Concussion Study is funded by the Presidents of the Ivy League Universities and the Big Ten Athletic Conference. ALCS was supported by the National Institute of Neurological Disorders and Stroke K23NS123340.
Competing interests BAD none. ALCS reports being an Associate Editor at the journal Neurology outside of the submitted work. IJB none. CLM none. RHH reports being a trustee of the McKnight Brain Research Foundation outside of the submitted work. DJW has received funds for expert testimony on long-term consequences of sports concussions and TBI. ARP reports textbook royalties from McGraw-Hill. Other named Ivy League-Big Ten Study Investigators report no competing interests.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.