Interaction of surface type, temperature, and week of season on concussion risk American football: Bayesian analysis of 8 Seasons of National Football League Data ================================================================================================================================================================== * James M Smoliga * Sameer K. Deshpande * Zachary O Binney ## Abstract **Background** Artificial turf fields and environmental conditions may influence sports concussion risk, but existing research is limited by weak methodology and limited sample size. The purpose of this study was to examine how playing surface, time of season, and game temperature relate to concussion risk in the National Football League (NFL). **Methods** This retrospective cohort study examined data from the 2012-2019 NFL regular season. Bayesian negative binomial regression models were fit to relate how playing surface, game temperature, and week of the season independently related to concussion risk and any interactions among these factors. **Results** 1096 concussions were identified in 1830 games. There was a 99% probability that concussion risk was reduced on grass surface (median Incidence rate ratio (IRR) = 0.78 [95% credible interval: 0.68, 0.89], 99% probability that concussion risk was lower at higher temperatures (IRR=0.85 [0.76,0.95] for each 7.9°C), and 91% probability that concussion risk increased with each week of the season (IRR=1.02 [1.00,1.04]). There was an 84% probability for a surface × temperature interaction (IRR=1.01 [0.96, 1.28]) and 75% probability for a surface × week interaction (IRR=1.02 [0.99, 1.05]). **Conclusions** Concussion risk is increased on artificial turf compared to natural grass, and this is exacerbated in cold weather and, independently, later in the season. The complex interplay between these factors necessitates accounting for multiple factors and their interactions when investigating sports injury risk factors and devising mitigation methods. **Publication history** A preprint of a previous version of this manuscript was published on MedRxIV: [https://www.medrxiv.org/content/10.1101/2022.01.29.22270096v1](https://www.medrxiv.org/content/10.1101/2022.01.29.22270096v1) doi: [https://doi.org/10.1101/2022.01.29.22270096](https://doi.org/10.1101/2022.01.29.22270096) Keywords (MeSH) * Brain Concussion * Football * Athletes * Risk Factors * Sports Injuries * Bayes Theorem * Temperature ## Introduction Concerns regarding neurodegenerative diseases have prompted sports organizations to consider implementing policies and interventions aimed at reducing sport-related concussions.1 Identifying modifiable risk factors for concussions in American football is particularly important,2 due to the potential risk of chronic traumatic encephalopathy.3,4 There has been considerable interest in examining the role of playing surfaces on brain injuries, and artificial turf fields are commonly marketed with claims of reducing the risk of concussions.5-8 Conversely, the National Football League’s (NFL) Players Association (NFLPA) claims that artificial turf is detrimental to player safety and long-term health,9 but there is currently limited evidence to support this claim specific to brain injuries.An analysis of the 2012-2013 seasons of National Football League (NFL) data found no association between playing surface and concussion risk.2 A 2019 systematic review and meta-analysis concluded that concussion risk was slightly decreased when contact sports (soccer, American football, and rugby union) were played on an artificial surface compared to natural grass.10 The two included studies for American football found no differences in concussion risk between one specific brand of artificial turf product and natural grass over three years of college football,11 and a greater risk of concussion on natural grass over five seasons of high school football.12 There are also within-season and temperature trends in injury rates in football and similar sports. Examination of four years of NFL data revealed a significant increase in concussion risk during the second half of the season, compared to the first half,13 and for games played in colder weather.14 Due to the timing of the season, which begins in late summer and ends in early January, and the primarily outdoor nature of the sport, these two factors are substantially correlated and should be considered together using methods to disentangle their associations. Surface may also interact with these variables. While biological factors, such as a reduced concussion threshold (i.e., the magnitude of injurious stimulus needed to produce concussion symptoms, which may be dependent on dynamic neuroanatomic and physiological factors and thus vary between and within athletes during a season), 15-18 could explain within-season changes in injury risk, it is also possible that within-season changes in the material properties of the playing surface (e.g., friction and shock absorption) could influence concussion risk.19 These may be due to wear/damage to the field20-22 or temperature-dependent effects,23,24 both of which may differ between playing surfaces. Playing surface, time of season, and game temperature all potentially influence concussion risk in football, but these factors are inter-related and existing analyses do not disentangle their potential confounding effects upon one another. Additionally, many analyses of these factors have divided temperature and week of season into categories,2,13,14 and it is possible that arbitrary categorization could lead to spurious associations for concussion risk.25,26 Thus, the objective of this study was to holistically examine how playing surface, game temperature, and week of season influenced concussion risk in NFL games. ## Methods ### Study Design and Data Collection We performed a retrospective cohort analysis of eight NFL seasons (2012-13 through 2019-20) of concussion data from the PBS Frontline Concussion Watch27 and Football Outsiders28 injury databases. The Football Outsiders database includes data from all weekly NFL injury reports during this time period, including concussion and non-concussion injuries from weeks 1-16 of the regular season. To strengthen completeness, this concussion dataset was combined with data from the Frontline database, which provided an independently collected list of concussions incurred by an NFL player from 2012-2015. Previously identified erroneous data points within the Frontline data set25 were excluded. Injury report data from Week 17 are not available for teams that did not qualify for the playoffs, and therefore only the first 16 weeks of data (15 games per team) were analyzed, as is consistent with previous studies.25 We assumed that all concussions occurred in a game setting, unless 1) previous data from the Frontline database25,27 specified it occurred during a practice (in which case, it was excluded from analysis) or 2) if a concussion was listed for a player in the absence of him taking any snaps in games that week. Multiple analyses suggest that approximately 95% of concussions in the NFL occur during games.25,29 #### Stadium and Playing Surface Playing surface was categorized into natural grass, artificial turf, and hybrid (a surface which includes an artificial turf foundation, combined with natural grass, potential resulting in unique profile of material properties). Preliminary analysis revealed hybrid surfaces may be associated with a distinct difference in concussion risk compared to natural grass and artificial turf (SDC3: eTable 1, eFigure 1), but represented <5% of total games (Table 1). Thus, games played on hybrid surfaces (n=82) were excluded from analysis. View this table: [TABLE 1.](http://medrxiv.org/content/early/2022/10/28/2022.01.29.22270096/T1) TABLE 1. Summary of NFL games included in the primary model. Note, games which were played in stadia with a retractable roof closed, but doors and windows open (n=8) are excluded from the table. Aberrations from a team’s normal home stadium were accounted for in the dataset, including games played in international locations and games which were moved to a different stadium due to extenuating circumstances. Likewise, within-season and between-season changes in playing surface type within a stadium were accounted for (e.g., the Houston Texans played one game on natural grass before switching to artificial turf for the remainder of the 2015 season). #### Game Temperature The Pro-Football-Reference database30 provides official weather details for most games. For outdoor games where any of this data was missing, temperature was retrieved from official NFL game reports and annual team media guides. If temperature was not provided in these sources, data were retrieved from WeatherUnderground31 from the zip code of the stadium at the nearest time point prior to kickoff (generally within one hour). The official NFL summary occasionally provided the temperature for indoor games (generally 20-22°C), and when this was not provided, it was assumed to be 20°C. To determine game temperature in stadia with a retractable roof, the respective team’s media guide was examined to determine if the game was played under indoor conditions (i.e., roof closed, doors/windows closed) or outdoor conditions (i.e., roof open). If this information was not available in the media guide, Wikipedia game summaries, the official NFL summary, news records, and game video were used to determine whether the roof was open or closed. Games in which the doors/windows were open, but the roof was closed (n=8) were excluded from analyses, since game temperature was not certain. ## Data Analysis ### Primary Model Data analysis was performed in R Studio v1.4.1106.32 The data set and source code are available as Supplemental Digital Content (SDC) 1 and 2, respectively. Preliminary analysis of the complete dataset revealed the number of concussions within a game is well-represented by a negative binomial distribution (eFigure 2). Thus, negative binomial regression models were developed to determine the relationship between risk factors and the number of concussions within a game. The number of concussions in a game served as the dependent variable. Week of season (henceforth referred to as “week”), playing surface (artificial turf vs. natural grass), and game temperature served as independent predictor variables. As noted previously, surface wear occurs over the course of a season and this may differ between surface types. The material properties of playing surface also seem to be temperature dependent, and this may also vary by surface type. Thus, the two-way interactions of surface × week and surface × temperature were included in the model. We did not have any *a priori* justification to include the two-way interaction of week × temperature or the three-way interaction between all independent predictor in the model. However, we did fit these models in a supplementary analysis (SDC3: eTable 2A—B) and found similar results. The model intercept represents the log of the average of number of concussions in a game played on artificial turf in the middle of the season (between weeks 8 and 9) and at an average temperature (17.11 degrees). Previous research suggests that over a 12 year span, the average number of concussions in an NFL was about 0.40.33 We therefore specified the prior distribution on the model intercept to be normal with mean -0.92 (*ln* 0.4 = -0.92) and with standard deviation 0.3. This choice represents a strong prior belief that in average conditions, there is often 0 or 1 concussions per game. Previous research using Poisson regression models did not find any relationship between time of season (four categories of four weeks) or playing surface (natural grass versus artificial turf) across two NFL seasons.2 Thus, the priors for the model parameters corresponding to “week” and “playing surface” were set to 0 with a normal distribution with standard deviation 2.5. For simplicity, we specified identical normal priors for the remaining model parameters. Although these choices are somewhat informative, prior predictive checks34 revealed the priors on the parameters induced a fairly non-informative prior on the average number of concussions per game. To wit, the induced prior on average number of concussions per game placed 50% probability on the range [0.64, 4.39], 80% probability on the range [0.36, 20.23], 90% probability on the range [0.27, 69.14], and 95% probability on the range [0.21, 260.90]. Importantly, the induced prior does not concentrate on extremely small values near zero (which would suggest that there are essentially no concussions) or on implausibly large values (which would suggest that there are at least tens or hundreds of concussions per game). Instead, the induced prior covers a broad range, indicates that the prior regularization on the model parameters is not so strong as to bias our model parameters towards one set of conclusions. We use the posterior mean as the point estimate of each model parameter and report the 95% credible interval as a measure of uncertainty about these parameters. Since the posterior distribution is analytically intractable, we rely on Markov Chain Monte Carlo (MCMC) simulation to approximate these quantities. Our simulation was performed using the rstanarm package. We specifically ran 8 chains for 15,000 iterations each and discarded the first 5,000 iterations as warmup. and credible intervals for all model parameters. We exponentiated each sample of the model parameters to compute the incidence rate ratio (IRR) of all of the predictive variables in our model. We performed several checks on the reasonableness of our fitted model (see SDC3).34,35 The median value and 95% credible intervals (high density intervals) were computed for the IRR of each parameter.36 We additionally computed the posterior probability that the IRR exceeded 1.0, which would suggest a harmful effect. ### Sensitivity Analyses To ensure the correlation between week and temperature did not cause collinearity problems for our primary model, we used separate models with novel control methods for each variable. We then compared the coefficients from these models to those of the primary model. To determine if week of season was truly an independent predictor of concussion risk, we ran a model on only indoor games to control for temperature without including it in the model. This allowed for us to control for temperature (i.e., it would be similar for all games), without including it as a model parameter, and determine if week was still an independent predictor of concussion. We ran a similar model using both indoor games and outdoor games played within a narrow temperature range (17.8 to 25.6°C). To determine if temperature was an independent predictor, we ran a model on only games during weeks 13-16 without a term for week in the model. We would not expect a major change in concussion risk from week alone during this short 4-week span, but games during this time have a wide range of temperatures. Thus, this late season model would allow for us to determine if temperature was still a predictor of concussion within this short time span. ## Results ### Games Included There were 1920 games played during the first 16 weeks of the 2012-2019 seasons (Table 1). Of these, a total of 1830 games met the inclusion criteria described in the methods and were included in the primary model. Median temperature was similar between surfaces (∼20°C for both), but temperature distribution varied by playing surface and week of season. Games played on natural grass (n=1030) followed a left-skewed distribution, whereas games played on artificial turf (n=800) had a spike at 20°C (n=359 indoor games at 20°C, with n=441 outdoor games with median 13.9°C) (eFigure 3A-B). Median temperature decreased over the course of the season (eFigure 3C). ### Model Results #### Model diagnostics and fit Our diagnostics did not reveal problems with convergence or mixing or the simulated Markov Chains: all of the R-hat statistics were close to 1.0 (consistent with convergence to the stationary distribution) and visual inspection of the traceplots did not reveal issues with mixing. Additional checks revealed that there was no single game exerting undue influence on our posterior distribution. Detailed discussion of our model diagnostics are available in the SDC3. #### Model parameters A summary of model parameters are presented in Table 2 and probability density plots for the posterior distribution with IRR credible intervals for each parameter are provided in Figure 1. Grass playing surface (99% probability), early season games (91% probability), and higher temperature (99% probability) games are all independently associated with a decreased risk of concussion. View this table: [Table 2.](http://medrxiv.org/content/early/2022/10/28/2022.01.29.22270096/T2) Table 2. Summary of parameters from the primary negative binomial regression model, including relevant two-way interaction terms (n=1830 games). Grass surface was associated with a reduced risk of concussion compared to artificial surface (99% probability). Concussion risk was also reduced with greater temperatures (99% probability) and increased with each week of the season (91% probability). There is reasonable probability (75-84%) that the two-way interaction terms affect concussion risk. If the two-way interaction terms are interpreted as “no effect” and removed from the model, all three main effects have a >99% probability of influencing concussion risk (SDC3: eTable 3, eFigures 6A-B and 7A-B). (Note, the lower bound credible interval for Week is reported as 1.0, but is 0.997, so the 95% credible interval does cross 1.0.) ![](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/10/28/2022.01.29.22270096/F1/graphic-4.medium.gif) [](http://medrxiv.org/content/early/2022/10/28/2022.01.29.22270096/F1/graphic-4) ![](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/10/28/2022.01.29.22270096/F1/graphic-5.medium.gif) [](http://medrxiv.org/content/early/2022/10/28/2022.01.29.22270096/F1/graphic-5) Figure 1. Probability distributions for all parameters primary model. All coefficients were exponentiated to compute Incidence Rate Ratio. For all figures, the vertical bar within each curve represents the median value for prior distribution, gray shading represents the 89% credible interval and the tails represent the 97% credible interval for each parameter coefficient. The intercept represents the concussion rate (number of concussions per game). The posterior distribution for the IRR of natural grass surface is below 1.0 (dotted red line, for frame of reference), representing a decreased risk for concussion. **Figure 1A. Probability distribution for natural grass playing surface**. The overall probability that concussion risk was reduced on natural grass compared to artificial turf was 98%, with a median incident rate ratio of 0.53 (∼47% reduced risk). However, positive values for two-way interactions (Figure 1C) suggest that the protective effect of natural grass may diminished with greater temperatures and later in the season. **Figure 1B. Probability distribution for game temperature, week of season, and the two-way interactions included in the model**. There was a high probability (99%) that concussion risk was reduced at greater temperatures, and increased later in the season (91%). Much of the density of the posterior distributions for week and the two-way interactions were >1.0, suggesting that these values may influence concussion risk, however cautious interpretation of these parameters may be warranted as their probabilities are a bit lower (Grass × Week: 75%; Grass × Temperature: 84%). If the two-way interactions are interpreted as “non-significant” and removed from the model, all main effect parameters increase to 99% probability of association with concussion risk (SDC3: eTable 3). The point estimate for grass surface IRR was 0.78, meaning that for mid-season games (weeks 8 and 9) played at the average temperature for the dataset (∼17°C), the risk of concussion was 22% lower on grass than on artificial turf. The 95% credible intervals (0.68, 0.89) indicated the true effect is likely is between a 11-32% reduction on grass compared to artificial turf. The point estimates for week IRR were 1.02 on artificial turf, and 1.04 on grass. In other words, for each one-week change, there was a 2% increased risk of concussions on artificial turf and 4% increased risk on grass, when temperature was held constant at the average for the dataset. The point estimates for temperature IRR were 0.84 on artificial turf and 0.94 on grass. This translates to a 16% reduced/increased risk of concussion on artificial turf and a 6% reduced/increased risk of concussion on natural grass, for each one standard deviation (corresponding to 7.9°C) increase/decrease in temperature during the middle of the season. The two-way interactions of surface × week (75% probability) and surface × temperature (84% probability) also appeared likely to be associated with concussion risk (Figure 1C). The results indicate the protective association of grass surface compared to artificial turf was somewhat attenuated in warmer conditions and late-season games (Figures 2 and 3). While the risk of concussion increases over the course of the season regardless of surface, this weekly increase in risk seems to be greater on natural grass than it does on artificial turf when temperature is held constant (Figure 2). The risk of concussion seems to increase substantially on artificial turf in colder temperatures, whereas cold temperatures do not seem to exacerbate the risk of concussion on natural grass as much holding week constant (Figure 3). ![Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/10/28/2022.01.29.22270096/F2.medium.gif) [Figure 2.](http://medrxiv.org/content/early/2022/10/28/2022.01.29.22270096/F2) Figure 2. Marginal effects of concussion rate by week of season, at a fixed temperature of 20°C. The black spline (solid = artificial turf, dashed = grass) represents predicted concussion rate and grey shading represents the 95% credible intervals. These represent reasonable probable temperatures for outdoor games in temperate climates across the entire football season, and 20°C represents the approximate temperature for indoor games (see SDC3: eFigure 3A-C). At any given week of the season, the risk of concussion on artificial turf is magnified during colder temperatures. ![Figure 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/10/28/2022.01.29.22270096/F3.medium.gif) [Figure 3.](http://medrxiv.org/content/early/2022/10/28/2022.01.29.22270096/F3) Figure 3. Marginal effects of concussion rate by temperature, fixed to Week 9 of the NFL season. The black spline (solid = artificial turf, dashed = grass) represents predicted concussion rate and grey shading represents the 95% credible intervals. Week 9 is utilized for visualization because it represents a time of the season with a wide range of game temperatures. The red and blue vertical lines represent the temperature extremes observed on at Week 9 within our dataset for artificial turf (n=44 games, mean temperature = 13.4°C) and natural grass (n=57 games, mean temperature = 17.4°C), respectively. For games as cold as 1.5°C, our model is quite certain that there will be more concussions on artificial turf than grass. At the other end of the temperature spectrum, however, our model is much more uncertain, as indicated by the substantial overlap in the credible intervals at temperatures above 20°C. This overlap is to be expected due to low sample size at higher temperatures at Week 9, typically falls in the early part of November. For Week 9 games, the average temperature in outdoor stadiums (n=88 games, artificial turf and natural grass combined) is 15.0°C (standard deviation = 6.8°C). Though the two-way interactions were likely associated with concussion risk (≥75% probability), we also ran the models without these interactions to see if the results for the main effects remained stable. When this main effects model is used, all three independent predictors have a >99% probability (SDC3: eTable 3A, eFigures 6A-C) and produced nearly identical point estimates to the primary model (SDC3: eTable 3B). #### Concussion rates For an average-temperature (∼17°C), mid-season game (Week 8-9), the number of concussion predicted per game was 0.51 on natural grass and 0.66 on artificial turf. Early season (Week 1) warm weather (25°C) games have a concussion rate of 0.37 on natural grass and 0.48 on artificial turf. Late season (Week 16) cold weather (5°C) games have concussion rates of 0.72 and 0.96, on the respective surfaces. #### Sensitivity Analyses Models investigating only indoor games or indoor and outdoor games in a moderate temperature range found similar IRRs for week and its interaction with surface as the main model (Table 2 and SDC3: eTable 4B). A model investigating only games in weeks 13-16 found similar IRRs for temperature and its interaction with surface as the main model (Table 2 and SDC3: eTable 5B). This confirmed both as independent predictors. ## Discussion To our knowledge, this is the first study to comprehensively examine how the interaction of surface, temperature, and time in season factors interact to influence concussion risk and the first to use a Bayesian approach to evaluating concussion risk. We found that the risk of concussion is reduced on natural grass playing surfaces, increased later in the season, and increased in colder temperatures. While all of these factors independently modify concussion risk, week and temperature both modify the effect of playing surface. These findings provide compelling evidence that artificial turf is associated with greater concussion risk in the NFL, and support the NFL Player’s Association’s advocacy for transitioning to natural grass fields to improve player safety.9 Additionally, these results suggest that extending the NFL season will likely increase concussion risk. The large sample size combined with the Bayesian negative binomial regression model provide us with a high-level of confidence in our results, which conflict with some existing literature with smaller sample sizes.2 In an industry-funded study, Meyers found a slightly greater risk of concussion on natural grass compared to one specific manufacturer’s (FieldTurf) artificial turf over three seasons of college football.11 We ran a separate analysis comparing natural grass surfaces to FieldTurf surfaces, and found natural grass was still associated with lower concussion risk (SDC3: eTables 6A-B, Figures 8A-B). Discrepancies may be attributable to greater sample size in our study, differences between college versus professional football, or differences in time period studied. The evidence that week and temperature interact with surface effects may also explain conflicting findings within the literature regarding the role of playing surface in athletic injuries. ### Explanatory Mechanisms The most likely explanation for lower concussion risk on natural grass compared to artificial turf is rooted in the biomechanics of helmet-to-ground impacts. Video review of injuries across five NFL seasons (1996-2001), suggests helmet-to-ground impacts account for 16%37 to 22%38 of concussions. A more recent study across two seasons (2015-2016) found 18% of concussion with a clear source of impact were due to this mechanism.39 The biomechanics of helmet-to-ground impacts are substantially different than player-to-player impacts, and football helmets are not tested (or designed) for these types of impacts.40-42 These impacts are potentially more severe, due to unusually high angular velocities and accelerations, and the potential for rebounds (both of the head within the helmet, and with the helmet and head combined).41 Grass tears in relation to translational force, whereas artificial turf does not – which allows the latter to have greater force transmission between the player and the surface.43,44 Increased friction between the helmet and surface could lead to greater torque placed about the head upon a helmet-ground interaction.41 Indeed, laboratory reconstruction of helmet-to-ground impacts has revealed the severity of impact is highly sensitive to the compliance and frictional properties of the surface.40 Reconstruction of 10 representative impacts causing concussion from NFL games demonstrated greater severity index and head injury criterion for all helmet-to-ground impacts on artificial turf (n=2) compared to helmet-to-helmet impacts (n=8).45 If our results are indeed driven by surface materials, this has implications for other athletes, since head-to-ground impacts account for 5-20% of concussions across various sports.46-51 Week of season and game temperature were independently found to influence concussion risk, which are also likely rooted in helmet impact biomechanics. Both surfaces degrade over the course of the season,20,21 which can influence friction and energy absorption.22,52 Likewise, surface temperature influences shock absorption properties of artificial turf,23,24 and soil hardness and grass quality of soccer fields (moisture dependent, with the greatest effect in winter months).52 Artificial surface temperature is also highly influenced by solar radiation,53 which varies with season. As demonstrated by the interaction effects (Figures 1C, 2A-B), there is a reasonably high probability that natural grass fields may also be prone to greater degradation within a season compared to artificial turf (e.g., soil compaction, significant decrease in grass density, etc.),52,54 and lose some of it apparent protective effect later in the season. Various biological factors could independently account for increased concussion risk later in the season. It has been theorized that prior history of repeated sub-concussive head impacts could change one’s biological threshold for a concussion,55 and there is mixed evidence to support this.15-18,56-59 Additionally, cross-sectional data suggests that brain volume changes seasonally in the general population,60 and this could theoretically influence concussion risk over the course of a season, regardless of game conditions or cumulative impact exposure. ### Limitations This study also has a number of limitations. There is considerable heterogeneity within the broad categories of artificial turf and natural grass,44 and we did not attempt to differentiate between different them (with the exception of Field Turf, SDC3: eTable 6A-B). In other sports, certain species/cultivars may be associated with differing risks of musculoskeletal injuries independent of ground hardness (attributable to differences in thatch),61,62 though not concussions.62 Surface moisture can also influence surface material properties.63 One could attempt to account for this by including weather in analysis, however, pre-existing precipitation influence the surface even if game-time weather as dry and weather can vary considerably within a game. We were also unable to quantify field “wear” given different environmental conditions (e.g., sunlight) and non-game activity (e.g.., concerts, setup and maintenance),64 and these may vary between stadia. Given the increased risk of concussion later in the season and at colder temperatures, we suspect that concussion risk would be greatest during the playoff season. However, accurate injury reports are not available for teams which fail to advance to the next round of playoffs (or after the championship game), which precludes the possibility of a valid analysis of concussion risk during this time. ## Conclusions Eight years of data demonstrated that concussion risk is substantially increased in NFL games played on artificial turf. While concussion risk also increases with cold weather regardless of playing surface, the risk is particularly amplified on artificial turf. There is also an increased risk of concussion later in the season, even when temperature is controlled for, but it remains uncertain if is this is due to biological factors, field degradation, or a combination. These data suggest that extending the NFL season may provide greater health risk to players, and that player concerns over the use of artificial turf are warranted. Further research is necessary to determine if these findings extend to non-professional levels of American football, and other team sports. ## Supporting information Supplemental Methods and Results [[supplements/270096_file07.docx]](pending:yes) ## Data Availability All data produced in the present study will be available upon acceptance into a peer-review journal. ## Footnotes * **Competing Interest Statement:** * **No competing interests:** All authors have completed the ICMJE uniform disclosure form at [www.icmje.org/coi\_disclosure.pdf](http://www.icmje.org/coi_disclosure.pdf) and declare: no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work. * **Sources of Financial Support:** The authors did not receive any financial support for this study. * **Data Sharing Statement:** The source code and full dataset used for this study are attached as Supplemental Digital Content (SDC) and will be made publicly available as SDC upon acceptance of this paper for publication. * This manuscript was revised after including input from an additional co-author. Key revisions include: 1. Performing a Prior Predictive checks on the primary model 2. Scaling and centering the temperature and week data 3. Adding point estimates for IRR 4. Providing additional reference lines to the figures (i.e., showing the maximum and minimum temperatures used in the models) 5. Updating methods, results, and discussion to reflect these changes. Note, these changes improved the statistical rigor of the study, but only had a trivial effect on the result values, and did not change the conclusions in any way. * Received January 29, 2022. * Revision received October 28, 2022. * Accepted October 28, 2022. * © 2022, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution-NonCommercial 4.0 International), CC BY-NC 4.0, as described at [http://creativecommons.org/licenses/by-nc/4.0/](http://creativecommons.org/licenses/by-nc/4.0/) ## References 1. 1.Emery CA, Black AM, Kolstad A, et al. What strategies can be used to effectively reduce the risk of concussion in sport? A systematic review. Br J Sports Med 2017;51(12):978–984. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6ODoiYmpzcG9ydHMiO3M6NToicmVzaWQiO3M6OToiNTEvMTIvOTc4IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjIvMTAvMjgvMjAyMi4wMS4yOS4yMjI3MDA5Ni5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 2. 2.Lawrence DW, Comper P, Hutchison MG. Influence of Extrinsic Risk Factors on National Football League Injury Rates. Orthop J Sports Med 2016;4(3):2325967116639222. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/2325967116639222&link_type=DOI) 3. 3.Mez J, Daneshvar DH, Kiernan PT, et al. Clinicopathological Evaluation of Chronic Traumatic Encephalopathy in Players of American Football. JAMA 2017;318(4):360–370. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2017.8334&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=28742910&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) 4. 4.Binney ZO, Bachynski KE. Estimating the prevalence at death of CTE neuropathology among professional football players. Neurology 2019;92(1):43–45. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1212/wnl.0000000000006699&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30487144&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) 5. 5.Field_Turf. Proven Safety. [https://fieldturf.com/en/why-fieldturf/proven-safety/](https://fieldturf.com/en/why-fieldturf/proven-safety/) Accessed August 26, 2021. 6. 6.AstroTurf. Football Turf. [https://www.astroturf.com/artifical-sports-turf/football-turf/](https://www.astroturf.com/artifical-sports-turf/football-turf/) Accessed August 26, 2021. 7. 7.Act_Global. Artificial Grass for American Football. [https://www.actglobal.com/american-football/](https://www.actglobal.com/american-football/) Accessed August 26, 2021. 8. 8.Matrix_Turf. Pads for synthetic turf. [http://www.matrix-turf.com/products/pads/](http://www.matrix-turf.com/products/pads/) Accessed August 26, 2021. 9. 9.Tretter J. Only Natural Grass Can Level The NFL’s Playing Field. [https://nflpa.com/posts/only-natural-grass-can-level-the-nfls-playing-field](https://nflpa.com/posts/only-natural-grass-can-level-the-nfls-playing-field). 10. 10.O’Leary F, Acampora N, Hand F, O’Donovan J. Association of artificial turf and concussion in competitive contact sports: a systematic review and meta-analysis. BMJ Open Sport Exerc Med 2020;6(1):e000695. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NzoiYm1qb3NlbSI7czo1OiJyZXNpZCI7czoxMToiNi8xL2UwMDA2OTUiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMi8xMC8yOC8yMDIyLjAxLjI5LjIyMjcwMDk2LmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 11. 11.Meyers MC. Incidence, mechanisms, and severity of game-related college football injuries on FieldTurf versus natural grass: a 3-year prospective study. Am J Sports Med 2010;38(4):687–97. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/0363546509352464&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20075177&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000276167200005&link_type=ISI) 12. 12.Meyers MC. Incidence, Mechanisms, and Severity of Game-Related High School Football Injuries Across Artificial Turf Systems of Various Infill Weights. Orthop J Sports Med 2019;7(3):2325967119832878. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/2325967119832878&link_type=DOI) 13. 13.Teramoto M, Cushman DM, Cross CL, Curtiss HM, Willick SE. Game Schedules and Rate of Concussions in the National Football League. Orthop J Sports Med 2017;5(11):2325967117740862. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/2325967117740862&link_type=DOI) 14. 14.Haider S, Kaye-Kauderer HP, Maniya AY, et al. Does the Environment Influence the Frequency of Concussion Incidence in Professional Football? Cureus 2018;10(11):e3627. 15. 15.Beckwith JG, Greenwald RM, Chu JJ, et al. Head impact exposure sustained by football players on days of diagnosed concussion. Medicine and science in sports and exercise 2013;45(4):737–746. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1249/MSS.0b013e3182792ed7&link_type=DOI) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000316410600017&link_type=ISI) 16. 16.Broglio SP, Lapointe A, O’Connor KL, McCrea M. Head Impact Density: A Model To Explain the Elusive Concussion Threshold. J Neurotrauma 2017;34(19):2675–2683. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1089/neu.2016.4767&link_type=DOI) 17. 17.O’Connor KL, Peeters T, Szymanski S, Broglio SP. Individual Impact Magnitude vs. Cumulative Magnitude for Estimating Concussion Odds. Ann Biomed Eng 2017;45(8):1985–1992. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s10439-017-1843-3&link_type=DOI) 18. 18.Stemper BD, Shah AS, Harezlak J, et al. Comparison of Head Impact Exposure Between Concussed Football Athletes and Matched Controls: Evidence for a Possible Second Mechanism of Sport-Related Concussion. Annals of Biomedical Engineering 2019;47(10):2057–2072. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s10439-018-02136-6&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) 19. 19.Orchard J. Is there a relationship between ground and climatic conditions and injuries in football? Sports Med 2002;32(7):419–32. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.2165/00007256-200232070-00002&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12015804&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000176246400002&link_type=ISI) 20. 20.Naunheim R, Parrott H, Standeven J. A Comparison of Artificial Turf. Journal of Trauma and Acute Care Surgery 2004;57(6):1311–1314. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/01.Ta.0000136154.36483.98&link_type=DOI) 21. 21.Wannop JW, Luo G, Stefanyshyn DJ. Footwear traction at different areas on artificial and natural grass fields. Sports Engineering 2012;15(2):111–116. 22. 22.Sánchez-Sánchez J, Felipe JL, Burillo P, del Corral J, Gallardo L. Effect of the structural components of support on the loss of mechanical properties of football fields of artificial turf. Proceedings of the Institution of Mechanical Engineers, Part P: Journal of Sports Engineering and Technology 2014;228(3):155–164. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/1754337114527276&link_type=DOI) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000345341200001&link_type=ISI) 23. 23.Gallardo L, García-Unanue J, Haxaire P, et al. Effect of extrinsic factors and structural components on sport functionality of artificial turf surfaces. Proceedings of the Institution of Mechanical Engineers, Part P: Journal of Sports Engineering and Technology 2018;233(1):135–144. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/1754337118809229&link_type=DOI) 24. 24.Charalambous L, von Lieres und Wilkau HC, Potthast W, Irwin G. The effects of artificial surface temperature on mechanical properties and player kinematics during landing and acceleration. Journal of Sport and Health Science 2016;5(3):355–360. 25. 25.Smoliga JM, Zavorsky GS. Team Logo Predicts Concussion Risk: Lessons in Protecting a Vulnerable Sports Community from Misconceived, but Highly Publicized Epidemiologic Research. Epidemiology 2017;28(5):753–757. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/EDE.0000000000000694&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=28570384&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) 26. 26.Binney Z, Smoliga JM. Bad Altitude: Categorizing Elevation Produces Spurious Association With Concussions in the National Football League (NFL). Journal of Orthopaedic & Sports Physical Therapy 2022:1–16. 27. 27.Frontline. Concussion Watch. [http://apps.frontline.org/concussion-watch/](http://apps.frontline.org/concussion-watch/), 2016. 28. 28.Football_Outsiders. [http://www.footballoutsiders.com](http://www.footballoutsiders.com). 29. 29.National\_Football\_League. Injury Data Since 2015. [https://www.nfl.com/playerhealthandsafety/health-and-wellness/injury-data/injury-data](https://www.nfl.com/playerhealthandsafety/health-and-wellness/injury-data/injury-data) Accessed August 26, 2021. 30. 30.Pro\_Football\_Reference. [https://www.pro-football-reference.com/](https://www.pro-football-reference.com/) Accessed August 26, 2021. 31. 31.Weather_Underground. Historical Weather. [https://www.wunderground.com/history](https://www.wunderground.com/history). 32. 32.RStudio_Team. RStudio: Integrated Development Environment for R. Boston, MA: RStudio, PBC, 2021. 33. 33.Casson IR, Viano DC, Powell JW, Pellman EJ. Twelve years of national football league concussion data. Sports Health 2010;2(6):471–83. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/1941738110383963&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=23015977&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) 34. 34.Vehtari A, Gabry J, Magnusson M, et al. loo: Efficient leave-one-out cross-validation and WAIC for Bayesian models. R package version 2.4.1, 2020. 35. 35.Vehtari A, Gelman A, Gabry J. Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Statistics and Computing 2017;27:1413–1432. 36. 36.McElreath R. Statistical Rethinking: A Bayesian Course with Examples in R and STAN (2nd Ed.). Boca Raton, FL: Chapman and Hall/CRC Press, 2020. 37. 37.Pellman EJ, Viano DC, Tucker AM, Casson IR, Waeckerle JF. Concussion in professional football: reconstruction of game impacts and injuries. Neurosurgery 2003;53(4):799-812; discussion 812-4. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1227/01.NEU.0000054218.50113.40&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=14519212&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000185898200020&link_type=ISI) 38. 38.Pellman EJ, Viano DC, Tucker AM, Casson IR. Concussion in professional football: location and direction of helmet impacts-Part 2. Neurosurgery 2003;53(6):1328-40; discussion 1340-1. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1227/01.NEU.0000093499.20604.21&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=14633299&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000187176500020&link_type=ISI) 39. 39.Lessley DJ, Kent RW, Funk JR, et al. Video Analysis of Reported Concussion Events in the National Football League During the 2015-2016 and 2016-2017 Seasons. Am J Sports Med 2018;46(14):3502–3510. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/0363546518804498&link_type=DOI) 40. 40.Kent R, Forman J, Bailey A, et al. Surface Contact Features, Impact Obliquity, and Preimpact Rotational Motion in Concussive Helmet-to-Ground Impacts: Assessment via a New Impact Test Device. Annals of Biomedical Engineering 2020;48(11):2639–2651. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s10439-020-02621-x&link_type=DOI) 41. 41.Kent R, Forman J, Bailey AM, et al. The biomechanics of concussive helmet-to-ground impacts in the National Football league. Journal of Biomechanics 2020;99:109551. 42. 42.National\_Operating\_Committee\_on\_Standards\_for\_Athletic_Equipment. Standard method of impact test and performance requirements for football faceguards. 2018. 43. 43.Kent R, Forman JL, Crandall J, Lessley D. The mechanical interactions between an American football cleat and playing surfaces in-situ at loads and rates generated by elite athletes: a comparison of playing surfaces. Sports Biomech 2015;14(1):1–17. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1080/14763141.2015.1024277&link_type=DOI) 44. 44.Kent R, Yoder J, O’Cain CM, et al. Force-limiting and the mechanical response of natural turfgrass used in the National Football League: A step toward the elimination of differential lower limb injury risk on synthetic turf. J Biomech 2021;127:110670. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jbiomech.2021.110670&link_type=DOI) 45. 45.Viano DC, Pellman EJ, Withnall C, Shewchenko N. Concussion in professional football: performance of newer helmets in reconstructed game impacts--Part 13. Neurosurgery 2006;59(3):591-606; discussion 591-606. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1227/01.NEU.0000231851.97287.C2&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=16823324&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000240623900023&link_type=ISI) 46. 46.Boden BP, Kirkendall DT, Garrett WE, Jr.. Concussion incidence in elite college soccer players. Am J Sports Med 1998;26(2):238–41. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/03635465980260021301&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=9548117&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000072713500013&link_type=ISI) 47. 47.Fuller CW, Dick RW, Corlette J, Schmalz R. Comparison of the incidence, nature and cause of injuries sustained on grass and new generation artificial turf by male and female football players. Part 1: match injuries. Br J Sports Med 2007;41 Suppl 1(Suppl 1):i20–6. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6ODoiYmpzcG9ydHMiO3M6NToicmVzaWQiO3M6MTQ6IjQxL3N1cHBsXzEvaTIwIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjIvMTAvMjgvMjAyMi4wMS4yOS4yMjI3MDA5Ni5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 48. 48.Hinton-Bayre AD, Geffen G, Friis P. Presentation and mechanisms of concussion in professional Rugby League Football. J Sci Med Sport 2004;7(3):400–4. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1440-2440(04)80035-5&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15518305&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000224394600015&link_type=ISI) 49. 49.Makdissi M, Davis G. Using video analysis for concussion surveillance in Australian football. J Sci Med Sport 2016;19(12):958–963. 50. 50.Marar M, McIlvain NM, Fields SK, Comstock RD. Epidemiology of concussions among United States high school athletes in 20 sports. Am J Sports Med 2012;40(4):747–55. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/0363546511435626&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=22287642&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000302285100002&link_type=ISI) 51. 51.Putukian M, Echemendia RJ, Chiampas G, et al. Head Injury in Soccer: From Science to the Field; summary of the head injury summit held in April 2017 in New York City, New York. Br J Sports Med 2019;53(21):1332. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6ODoiYmpzcG9ydHMiO3M6NToicmVzaWQiO3M6MTA6IjUzLzIxLzEzMzIiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMi8xMC8yOC8yMDIyLjAxLjI5LjIyMjcwMDk2LmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 52. 52.Biraderoglu M, Kaplan S, Basaran M. Spatio-temporal investigation of surface soil hardness on professional football field. Environmental Monitoring and Assessment 2020;192(2):151. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s10661-020-8087-7&link_type=DOI) 53. 53.Thoms AW, Brosnan JT, Zidek JM, Sorochan JC. Models for Predicting Surface Temperatures on Synthetic Turf Playing Surfaces. Procedia Engineering 2014;72:895–900. 54. 54.Carrow RN, Martin Petrovic A. Effects of Traffic on Turfgrasses. Turfgrass, 1992;285–330. 55. 55.Guskiewicz KM, Mihalik JP, Shankar V, et al. Measurement of head impacts in collegiate football players: relationship between head impact biomechanics and acute clinical outcome after concussion. Neurosurgery 2007;61(6):1244–52; discussion 1252-3. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1227/01.neu.0000306103.68635.1a&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=18162904&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000252044400036&link_type=ISI) 56. 56.Oliver JM, Anzalone AJ, Stone JD, et al. Fluctuations in blood biomarkers of head trauma in NCAA football athletes over the course of a season. Journal of Neurosurgery JNS 2019;130(5):1655–1662. 57. 57.Oliver JM, Jones MT, Kirk KM, et al. Serum Neurofilament Light in American Football Athletes over the Course of a Season. J Neurotrauma 2016;33(19):1784–1789. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1089/neu.2015.4295&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26700106.&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) 58. 58.Rowson S, Campolettano ET, Duma SM, et al. Accounting for Variance in Concussion Tolerance Between Individuals: Comparing Head Accelerations Between Concussed and Physically Matched Control Subjects. Annals of Biomedical Engineering 2019;47(10):2048–2056. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s10439-019-02329-7&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F10%2F28%2F2022.01.29.22270096.atom) 59. 59.Zonner SW, Ejima K, Bevilacqua ZW, et al. Association of Increased Serum S100B Levels With High School Football Subconcussive Head Impacts. Frontiers in Neurology 2019;10(327). 60. 60.Book GA, Meda SA, Janssen R, et al. Effects of weather and season on human brain volume. PLoS One 2021;16(3):e0236303. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pone.0236303&link_type=DOI) 61. 61.Orchard JW, Chivers I, Aldous D, Bennell K, Seward H. Rye grass is associated with fewer non-contact anterior cruciate ligament injuries than bermuda grass. British Journal of Sports Medicine 2005;39(10):704–709. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6ODoiYmpzcG9ydHMiO3M6NToicmVzaWQiO3M6OToiMzkvMTAvNzA0IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjIvMTAvMjgvMjAyMi4wMS4yOS4yMjI3MDA5Ni5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 62. 62.Orchard JW, Waldén M, Hägglund M, et al. Comparison of injury incidences between football teams playing in different climatic regions. Open access journal of sports medicine 2013;4:251–260. 63. 63.Alcántara E, Gámez J, Rosa D, Sanchis M. Analysis of the influence of rubber infill morphology on the mechanical performance of artificial turf surfaces for soccer. Proceedings of the Institution of Mechanical Engineers, Part P: Journal of Sports Engineering and Technology 2009;223(1):1–9. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1243/17543371JSET27&link_type=DOI) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000282935300001&link_type=ISI) 64. 64.Aldahir PCF, McElroy JS. A Review of Sports Turf Research Techniques Related to Playability and Safety Standards. Agronomy Journal 2014;106(4):1297–1308.