Sports-related head injuries are a public health issue, as they occur at all ages, from youth to professional levels. It is estimated that 1–2 million children in the United States receive a sports-related concussion every year,1 and the incidence remains high in college athletes across multiple sports.2 As epidemiologic research continually emerges regarding the long-term dangers of contact sports,3,4 there is considerable interest in reducing the risk of concussions, with many searching for a seemingly magic answer that minimizes the serious risks of inherently dangerous sports. Thus, the sports community may be vulnerable to adopting products and implementing policies based on untested, nonreplicable, and scientifically unsound solutions to the concussion crisis.5–9 Some concussion research has been further complicated by conflicts of interest.10,11 As new research findings and subsequent interventions rapidly appear, healthcare providers must carefully evaluate the available evidence to make informed recommendations to sports organizations, athletes, coaches, and parents to optimize safety in the athletic population.
There has been a recent spate of headline-grabbing epidemiologic research suggesting that concussion risk is reduced by approximately 30% at “higher altitude” due to fluid accumulation decreasing intracranial brain movement.12,13 This concept has inspired technologic innovations to mimic the supposed protective effects of altitude exposure.14 However, the physiologic justification for this theory is weak,15 and some inconsistent epidemiologic findings suggest the relationship may be due to chance.16,17,18 Thus, it is imperative to critically reconsider the epidemiologic evidence supporting this link. Here, we sought to determine the following: (1) the replicability of the epidemiologic link between altitude and concussion risk and (2) if a predictor without any scientific justification (team logo) could also yield a strongly protective association, presumably by chance.
This project received ethical approval as exempt status by the High Point University Institutional Review Board (#201607-500).
We performed a retrospective cohort analysis of four National Football League (NFL) seasons (2012–2013 through 2015–2016) of publicly available data from Public Broadcasting System (PBS) Frontline Concussion Watch.19 This resource has been used for other studies examining concussion epidemiology.12,20,21
Full details of our methods, including data verification and extraction, are reported in the eAppendix (https://links.lww.com/EDE/B219). In brief, we attempted to replicate the methods of the altitude study by Myer et al12 and analyzed concussions that occurred during games during the first 16 weeks of the NFL regular season (Figure). This period includes one “bye week,” during which each team does not play. Thus, 15 games per team-season were included in the analysis. Athlete-exposures were defined based on NFL rules that allow 46 individuals to dress for each game and potentially be exposed to a concussive event, resulting in a total of 690 athlete-exposures (46 players × 15 games) for each of the 32 teams. Thus, 22,080 athlete-exposures per season were analyzed.
For all four seasons, we determined the number of exposures and concussions which occurred at games played at “lower” (<196 m) and “higher” (≥196 m) altitude venues, based on the previously reported thresholds by Myer et al.12 Likewise, we compiled exposure and concussion data for teams with an animal (non-human) logos (n = 15 teams, e.g., Denver Broncos, Carolina Panthers) and non-animal logos (n = 17 teams, e.g., Tennessee Titans, Pittsburgh Steelers). Our dataset is available as Dataset1 (https://links.lww.com/EDE/B220).
We sought to determine whether different predictor variables, altitude category and team logo, were associated with an altered risk of concussion. For each analysis, relative risk (RR) and risk difference (RD) were computed using 2 × 2 contingency tables. We estimated 95% confidence intervals (95% CIs) to determine whether the predictors (higher altitude and animal logo) were statistically “protective” (fewer concussions than reference category across 95% CI) or “harmful” (more concussions than reference across 95% CI). If the 95% CI crossed 1.0, the effect was considered “near-null.” A comparison of proportions test was used to compare the differences in the risk difference of a concussive event between categories. All statistical analyses were performed using MedCalc Software (MedCalc Software bvba, Ostend, Belgium).
A summary of exposure, concussions, and risk statistics are presented in Table 1 (altitude) and Table 2 (logo). Our analysis of the 2012–2013 and 2013–2014 seasons ultimately produced results comparable to those reported by Myer et al12 (see further details in eAppendix; https://links.lww.com/EDE/B219), in that risk of concussion was reduced altitude ≥196 m when data from both seasons were combined (RR = 0.66 [0.50, 0.89]). Combined data from the following two seasons were near-null (RR = 0.92 [0.70, 1.22]). When data from all four seasons were combined, there was a protective effect for altitude (RR = 0.78 [0.64, 0.96]), but only one individual season (2013–2014) had a protective effect (RR = 0.48 [0.30, 0.77]). When that season was excluded, altitude was near-null across three combined seasons (RR = 0.90 [0.72, 1.12]) (eTable 2; https://links.lww.com/EDE/B219).
Teams with animal logos had a decreased risk of concussion when data from the 2012–2013 and 2013–2014 seasons were combined (RR = 0.72 [0.57, 0.91]), when 2014–2015 and 2015–2016 seasons were combined (RR = 0.78 [0.61, 0.99]), and when all four seasons were combined (RR = 0.75 [0.63, 0.89]) (Table 2). Two individual seasons (2012–2013 and 2015–2016) had protective effects for logo.
When data for all four seasons were combined, risk difference for higher altitude was 1.4 concussions per 1000 athlete-exposures. This equates to 1 less concussion per 16 team games. For animal logo, risk difference was 1.8 concussions per 1000 athlete-exposures, which equates to 1 less concussion per 13 team games.
When all four seasons are combined, the protective effect of logo is independent of altitude, as animal logo teams had a lower incidence of concussion sustained at <196 m (RR = 0.80 [0.65, 0.97]) and ≥196 m (RR = 0.67 [0.47, 0.95]) (eTable 1; https://links.lww.com/EDE/B219).
The replicability of the association between altitude and concussion risk was dependent on the comparisons performed. Though we replicated the results by Myer et al,12 combined data from the following two seasons (2014–2015 and 2015–2016) revealed no protective associations for higher altitude. Only “one” season had a protective association (2013–2014), which drove that for higher altitude when seasons were combined. In isolation, the protective effects from selected seasons could potentially justify “higher” altitude-inspired intervention strategies (especially if near-null results are not reported), such as jugular compression.21 However, animal logo had a similarly convincing protective association against concussion. Indeed, teams with animal logos were at reduced risk for concussion in “two” seasons. Team logo also showed a protective association for “both” two-season combined analyses, and showed comparatively more favorable risk reduction statistics than higher altitude when all four seasons were combined. Based on these findings alone, one could suggest that teams consider changing their logos, or maybe even implement “protective” animal stickers that players could place on their equipment.
Similar logic has been prematurely applied in developing interventions based on initial epidemiologic research linking higher altitude exposure with decreased risk of concussion.14,22 The mechanisms proposed to explain the protective effects of altitude12,13 are based on misunderstandings of physiologic responses to high altitude exposure.15 A thorough discussion of these misconceptions is available elsewhere,15 but noticeable physiologic responses to altitude in the brain only occur >2,000 m (>7,000 feet) at a very minimum. Instead of physiologically relevant thresholds, concussion studies have defined higher altitude as those above the median elevation of game venues; ≥196 m (645 feet) for the NFL study by Myer et al.12,13 The weak physiologic justification is concerning, as observational studies without a strong scientific foundation do not demonstrate cause-and-effect and therefore often have limited reproducibility and are subject to misinterpretation and misuse.23
Unexpected results from epidemiologic studies can sometimes inspire re-examination of pathophysiologic mechanisms. However, weak scientific rationale combined with unconvincing replicability suggest that associations between altitude and concussion are likely coincidental rather than causal, as they are for team logo. Extremely large studies may be more likely to find statistical associations where no meaningful differences exist,24 as exemplified by our team logo analysis. The relative risk reductions appear so dramatic (RR = 0.75–0.78) because the baseline risk is low (approximately 5–7 concussions per 1000 athlete-exposures). Although sometimes protective, the risk difference for higher altitude and animal logo were both quite small and unstable from season-to-season (range = 0.2–3.7 per 1000 athlete-exposures).
Sports medicine has been criticized for having a weak evidence basis,25 and this seems true for some aspects of concussion research. To minimize the chances of coincidental findings being mistaken for real effects,26 sports epidemiology studies should evaluate data across multiple seasons to determine whether consistent patterns exist and ensure that random fluctuations within a short time period do not spuriously inflate the overall long-term effect. Interpretation must also go beyond significant P values, and consider clinically relevant effect sizes.27 Datasets from sports injury studies should be made readily accessible so that other researchers can attempt to verify data accuracy, replicate the results, and perform further analyses. Although observational research is extremely valuable, it must be remembered that it is one component of an effective integrated approach to medicine,28 thus an interdisciplinary approach to concussion research and interventions is necessary.
The sports community is desperate to protect athletes from concussions, and there is considerable pressure for athletic clubs and academic institutions not to turn a blind eye to the issue.29 As coaches, athletic directors, and medical staff are rightfully pressured to do “something” to reduce athlete risk, it is necessary for healthcare providers to have an accurate understanding of concussion research, so they can make evidence-based recommendations to the general public. Thus, it is imperative that emerging theories regarding concussion are sufficiently scrutinized so that weak scientific claims with limited replicability do not divert resources or provide athletes with a false sense of safety. Our demonstration that higher altitude and animal logo both reduce concussion risk within this four-season NFL dataset serves as an exemplary warning that tenuous results from observational research can inaccurately imply causation, and that careful critique of studies is necessary before clinical and policy decisions are implemented.
1. Bryan MA, Rowhani-Rahbar A, Comstock RD, Rivara FSports- and recreation-related concussions in US youth. Pediatrics. 2016;138:e20154635.
2. Kerr ZY, Roos KG, Djoko A, et alEpidemiologic measures for quantifying the incidence of concussion in National Collegiate Athletic Association Sports. J Athl Train. 2017;52:167–174.
3. Bailes JE, Petraglia AL, Omalu BI, Nauman E, Talavage TRole of subconcussion in repetitive mild traumatic brain injury. J Neurosurg. 2013;119:1235–1245.
4. Pearce N, Gallo V, McElvenny DSports-related head trauma and neurodegenerative disease. Lancet Neurol. 2014;13:969–970.
5. Breech JRussell Wilson Says ‘Recovery Water’ Healed His NFC Title Game Head Injury. 26 August 2015. Available at: http://www.cbssports.com/nfl/eye-on-football/25281302/russell-wilson-says-recovery-water-healed-nfc-title-game-head-injury
. Accessed November 7, 2016.
6. Carson DTom Brady Once Endorsed a Sports Drink That Claimed Concussion ‘Protection’. Bleacher Report. 27 August 2015. Available at: http://bleacherreport.com/articles/2556219-tom-brady-once-endorsed-a-sports-drink-that-claimed-concussion-protection.html
. Accessed 7 November 2016.
7. Iannetta JU.S. Soccer’s Ali Krieger Wears a Concussion Headband. But Do They Work? 29 June 2015. Available at: http://www.denverpost.com/2015/06/29/u-s-soccers-ali-krieger-wears-a-concussion-headband-but-do-they-work/
. Accessed 7 November 2016.
8. Singal JThe University of Maryland Has a Burgeoning Chocolate-Milk Concussion Scandal on Its Hands. 20 January 2016. Available at: http://nymag.com/scienceofus/2016/01/chocolate-milk-concussion-scandal.html
. Accessed 7 November 2016.
9. United States Food and Drug Administration. Can a Dietary Supplement Treat a Concussion? No! 4 September 2015. Available at: http://www.fda.gov/ForConsumers/ConsumerUpdates/ucm378845.htm
. Accessed 7 November 2016.
10. Branch JN.F.L. Tried to Influence Concussion Research, Congressional Study Finds. 23 May 2016. New York, NY: New York Times; Available at: http://www.nytimes.com/2016/05/24/sports/football/nfl-tried-to-influence-concussion-research-congressional-study-finds.html?_r=0
. Accessed 7 November 2016.
11. Schwarz A, Bogdanich W, Williams JN.F.L.’s Flawed Concussion Research and Ties to Tobacco Industry. 24 March 2016.New York, NY: New York Times;
12. Myer GD, Smith D, Barber Foss KD, et alRates of concussion are lower in National Football League games played at higher altitudes. J Orthop Sports Phys Ther. 2014;44:164–172.
13. Smith DW, Myer GD, Currie DW, Comstock RD, Clark JF, Bailes JEAltitude modulates concussion incidence: implications for optimizing brain compliance to prevent brain injury in athletes. Orthop J Sports Med. 2013;1:2325967113511588.
14. Orcutt MNew Collar Promises to Keep Athletes’ Brains from “Sloshing” During Impact. 2016. Cambridge, MA: MIT Technology Review; Available at: https://www.technologyreview.com/s/600691/new-collar-promises-to-keep-athletes-brains-from-sloshing-during-impact/
. Accessed 7 November 2016.
15. Smoliga JM, Zavorsky GS“Tighter fit” theory-physiologists explain why “higher altitude” and jugular occlusion are unlikely to reduce risks for sports concussion and brain injuries. J Appl Physiol (1985). 2017;122:215–217.
16. Lawrence DW, Comper P, Hutchison MGInfluence of extrinsic risk factors on National Football League injury rates. Orthop J Sports Med. 2016;4:2325967116639222.
17. Zavorsky GS, Smoliga JMCorrect data and meta-analytic approaches show the reduced risk of concussion for athletes playing at higher altitudes-reply. JAMA Neurol. 2017;74:485–486.
18. Zavorsky GS, Smoliga JMRisk of concussion for athletes in contact sports at higher altitude vs at sea level: a meta-analysis. JAMA Neurol. 2016;73:1369–1370.
19. Frontline. Concussion Watch. 2015. Available at: http://apps.frontline.org/concussion-watch/
. Accessed 7 November 2016.
20. Nathanson JT, Connolly JG, Yuk F, et alConcussion incidence in professional football: position-specific analysis with use of a novel metric. Orthop J Sports Med. 2016;4:2325967115622621.
21. Teramoto M, Petron DJ, Cross CL, Willick SEStyle of play and rate of concussions in the National Football League. Orthop J Sports Med. 2015;3:2325967115620365.
22. Myer GD, Yuan W, Barber Foss KD, et alAnalysis of head impact exposure and brain microstructure response in a season-long application of a jugular vein compression collar: a prospective, neuroimaging investigation in American football. Br J Sports Med. 2016;50:1276–1285.
23. Young SS, Karr ADeming, data, and observational studies: a process out of control. Significance. 2011;8:116–120.
24. Ioannidis JPWhy most published research findings are false. PLoS Med. 2005;2:e124.
25. Lohmander LS, Roos EMThe evidence base for orthopaedics and sports medicine. BMJ. 2015;350:g7835.
26. Nuzzo RHow scientists fool themselves - and how they can stop. Nature. 2015;526:182–185.
27. Nuzzo RScientific method: statistical errors. Nature. 2014;506:150–152.
28. Willett WCThe search for truth must go beyond statistics. Epidemiology. 2008;19:655–656; discussion 657.
29. Lynall RC, Guskiewicz KMConcussion research: new horizons. Lancet Neurol. 2015;14:14–16.