Preliminary Examination of Guardian Cap Head Impact Kinematics Using Instrumented Mouthguards Open Access

Twenty-five NCAA Division I male American football players (average age = 20 ± 1 years) were recruited from the same American football team if they wore dental-scanned and -fitted iMGs to record head impacts (3200 Hz; Prevent Biometrics) during preseason workouts. Each iMG contained a triaxial accelerometer and gyroscope to measure linear and rotational kinematics. The players wore the iMGs for various numbers of the 6 workouts included in this study. For example, 13 players had valid data from their iMGs for the first workout included, whereas 17 did for the third workout (Table 1). Of the 25 participants, only 7 players (average age = 20 ± 0.76 years) engaged in all 6 practice sessions and had complete data for all recorded workouts (Table 1). Due to the fluctuating sample size for each workout, we analyzed these data in 2 ways. The first analysis used all available data from the full set of 25 participants, resulting in 83 individual observations, as some of the 25 participants completed multiple workouts. The 7 players with valid data from all 6 workouts underwent a second, separate repeated-measures analysis. The football players’ positions varied (offensive lineman, defensive lineman, running back, tight end, and linebacker), but all were identified by the coaching staff as having the potential to be high-dose players (Table 1). We defined a high-dose player as an individual who had the greatest opportunity for contact based on the position type and the offensive/defensive scheme of the respective university. This included players in positions such as offensive lineman, defensive lineman, running back, tight end, and linebacker, as players in these positions are known to receive more impacts per training week and are thus more susceptible to the effects of RHIs compared with quarterbacks, kickers, or punters.¹²  Identification of high-dose players occurred through close consultation with the coaching and sports medicine staff. Of the 25 participants, 12% (n = 3) of the athletes were starters, 36% (n = 9) were rotational players, and 52% (n = 13) were scout team players as denoted by the coaching staff (Table 1). The players were allowed to choose the brand of their traditional helmet; 84% (n = 21) of players wore the Riddell SpeedFlex (Riddell Sports Group), 4% (n = 1) wore the Vicis Zero2, and 12% (n = 3) wore the Schutt F7 2.0 (Schutt Sports). The players used the same helmet for each of the 6 recorded workouts.

Participants were excluded if they did not adhere to the iMG compliance standards (eg, chewed excessively) or if they had an injury that precluded them from practicing on the day of the selected practices. Of the 25 participants, 2 had been diagnosed with a concussion within 6 months of the study start date but were medically cleared to return to sport. One participant was diagnosed with a concussion during the study and was then excluded. Concussion history was not an exclusion criterion; however, all participants needed to complete the full workout as designated by the coaching staff. Athletes were not considered if they were performing modified workouts due to the postconcussion return-to-play process.

All individuals agreed orally and with written informed consent to participate. This study was approved by the university’s institutional review board (No. 1757959-5) and in accordance with the Declaration of Helsinki.

We selected 6 (3 wearing traditional helmets [TRAD] and 3 with GCs affixed to traditional helmets [GC]) nearly identical practice plans that contained similar levels of hitting and times for drills. These 6 practices were chosen because they depicted common NCAA American football workouts, were not deemed scrimmages, and had only slight variations in practice plans. We vetted all preseason practice plans and carefully picked these 6 workouts after video verification to ensure they had similar workout plans. The practices were during the middle of the preseason, as at the time of this study, the athletic department intended to use GCs only during the preseason. The practices lasted approximately 2 hours, with 15 minutes of tackling drills, 25 minutes of individual position drills (which included a tackling circuit), 60 minutes of team drills using THUD (initiation of contact at full speed with no predetermined winner but no takedown to the ground) tackling, and 20 minutes of team tempo play TAG (tackling to the ground).

The TRAD practices occurred within 24 hours of one another, whereas the GC practices occurred with 5 days between sessions. The last TRAD practice was 2 days before the first GC practice session; therefore, all 6 included workouts occurred within a single athletic season. The days between GC practices varied due to the sports season moving from the acclimation period and preseason to the regular season.¹³  During the regular season, full-contact TAG is limited to 1 full-contact day each week per NCAA recommendations.¹³  All participants in the GC portion of the study had GCs that were fitted by the equipment staff and verified as in working condition before each session.

Before the season, each participant underwent dental scanning and was provided with a custom mouthguard created by Prevent Biometrics. Based on preliminary data using the industry standards for head impact verification, the Prevent mouthguard’s custom head-impact filtering algorithm has a sensitivity of 0.75 for false-negative detection and a specificity of 0.95 for false-positive performance.¹⁴  When a participant incurred a blow ≥5g PLA, the sensor collected data for 16 pretrigger and 144 posttrigger samples. Each participant was instructed to always wear the iMG and to refrain from placing it in areas that might cause breakage. In addition, all reported true-positive head impacts were video verified using 3 camera (model 4k/HD AG-UX180 Handheld Camcorder; Panasonic) angles (both end zones and the 50-yard line) on a full-size practice field. One person, with NCAA Division I film review experience, video verified all the true-positive head impacts. Of the 828 recorded impacts, 19 could not be video verified and were removed from the study. The PLA, PAA, and total number of impacts were analyzed across the 6 practice sessions. Each impact was considered an individual event and ensemble averaged.

We examined all data for influential skewness and kurtosis. The head kinematics data were normally distributed, and an initial independent-samples t test showed no statistical difference between the TRAD and GC practices in all head kinematic data (Table 2). Independent-samples t tests were used for the 83 individual observations, as a varying number of the 25 participants engaged in each workout. Thus, we collapsed the TRAD and GC sessions separately and performed 1-way analyses of variance (ANOVAs) for PLA, PAA, and total impacts. One-way ANOVAs were used for the same reasoning as the independent-samples t tests: repeated measures are not warranted when sample sizes vary at each time point. In addition, the 7 players who had complete data sets were independently analyzed using repeated-measures ANOVAs across all 6 sessions for PLA, PAA, and total impacts. Follow-up paired-samples t tests were conducted in the event of a significant time effect for the entire sample (Table 3), as well as independent-samples t tests for the 7 consistent players (Table 3). The α level was set at .05 a priori.

The mean values for the average PLA, PAA, and total impacts for the collapsed TRAD and GC workouts using data from both the entire sample and only the 7 consistent players can be found in Table 4. The mean values for the average PLA, PAA, and total impacts for each individual workout are available in Table 2.

A 1-way between-participants ANOVA revealed no difference between the collapsed mean TRAD and GC PLA (F_1,83 = 1.67, P = .20), PAA (F_1,83 = 0.44, P = .51), or total impacts (F_1,83 = 0.13, P = .72). These data suggest that the head kinematics for all players did not differ before and after GC implementation.

A 1-way within-participants repeated-measures ANOVA yielded no differences between the collapsed mean PLA (F_1,6 = 1.16, P = .32), PAA (F_1,6 = 1.36, P = .29), or total impacts (F_1,6 = 0.03, P = .88) for the TRAD and GC conditions (Table 4). These data indicate that, in players who participated in all 6 practice sessions, the GC did not alter the head kinematics data.

To address the growing concern around RHIs and brain injury risk in sport, the purpose of our study was to evaluate the effect of GCs on head kinematics during similar on-field preseason practices in American football using iMGs. To our knowledge, this is the most diverse on-field measurement of iMGs using GCs, with an analysis of 809 unique video-verified head impacts across various players, positions, and playing time roles. We were able to obtain multiple practices’ worth of data from the same players throughout a single season, which other researchers have not been able to accomplish. The major outcome was that the GCs did not reduce or attenuate the PLA or PAA or alter the total number of head impacts during the 6 closely matched practice sessions. In addition, no changes in any variables were noted after implementation of the GCs when the same 7 athletes were compared across the 6 practice sessions. Despite reporting by media outlets that GCs greatly reduced the incidence of concussion, those findings were not peer reviewed and have not been backed by publicly available data, whereas our results suggested they did not affect overall head kinematic data on the field.^9,15  Although no direct link exists between head kinematic data and concussions, the total amount of head impact exposure (ie, frequency and magnitude) is generally higher in the days leading up to a concussion diagnosis.^16,17  Our outcomes did not indicate that the GCs reduced overall head impact exposure; however, we did not track concussions.

These results support prior laboratory research demonstrating that GCs did not reduce head kinematic data during use.^10,18  The majority of the published investigations on GCs reflects laboratory research, which has consisted of dropping the GCs from various heights and measuring the linear acceleration the helmet experienced when hitting the ground. Laboratory testing is highly recommended for preliminary investigations on safety concerns when implementing new technology in sport, yet it lacks ecological validity. The most effective line of testing is to apply the technology in a real-world setting, which is what we aimed to do by using GCs during closely matched practices. When we directly compared our data with previously published laboratory examinations of GCs, the PLA and PAA were similar to the results of existing research.¹⁸ 

In addition to supporting prior laboratory research on GC efficacy, our findings align with the 2022 National Athletic Trainers’ Association position statement regarding strategies for reducing headfirst contact behavior in American football.¹⁹  The statement proposed that companies producing helmet add-ons often amplified the reduction in head injuries, which may have caused an increase in high-contact play due to the perception of reduced risk.¹⁹  The statement also indicated that the efficacy of helmet add-ons has not been established; therefore, no evidence supports widespread use.¹⁹  We observed no differences in PLA, PAA, or total impacts with GC implementation, which suggests this particular cohort may not have engaged in riskier play due to the perception of decreased injury risk from GCs. If future researchers note a reduction in PLA or PAA with no change in total impacts when players are wearing GCs, that could be considered evidence that players engaged in riskier behavior due to perceptual changes. Nonetheless, our examination offered no evidence to imply any behavioral effects. That said, the nonsignificant results in this study provided on-field evidence that GCs may not be effective in mitigating head impact kinematics in American football.

To our knowledge, only a single study used on-field data to validate the use of GCs.¹¹  Cecchi et al used iMG analyzed data from 5 Division I linebackers collected from 13 practices using traditional helmets without GCs in 2019 and 14 practices using GCs attached to traditional helmets in 2021.¹¹  The 5 players in the earlier study were not consistent throughout the 2 recorded seasons. Though our research involved significant methodologic differences, as we analyzed data from a larger sample set over the course of a single season, the examination by Cecchi et al is currently the only published work available for comparison.¹¹  We found no differences in PAA, PLA, or total impacts, as did Cecchi et al.¹¹  They also reported no differences in diffuse axonal multi-axis general evaluation or head acceleration response metric; however, we did not evaluate these variables. The combined results from Cecchi et al and the present study strongly suggest that GCs were not effective in reducing the amount and the magnitude of head kinematics experienced by collegiate American football players.¹¹ 

Our outcomes suggested that GCs were ineffective in attenuating the linear and rotational head kinematics experienced by collegiate American football players, although they might be effective when applied to other helmets.

Our investigation had limitations related to the method of data collection, as it might have been more beneficial to obtain data during games as opposed to practices; still, GCs are not allowed during games. Previous researchers have shown that the game concussion rate for collegiate American football players was 3.74 per 1000 athlete-exposures versus the practice concussion rate of 0.53 per 1000 athlete-exposures.²⁰  Comparing GC and non-GC data from American football games would allow us to see the effectiveness of GCs in a game setting. An additional limitation of this study was that we did not consider behavioral differences while the players were wearing GCs. It is possible that the players felt more or less protected with the additional padding, which changed the way they played. Our data did not reveal a change in total impacts from TRAD to GC; however, we cannot firmly conclude that no behavioral differences in the players’ practice style occurred while wearing the GCs. Additionally, not all players wore the same helmet underneath the GC, nor did we analyze whether a certain helmet paired better with the GC to reduce the head kinematics experienced by American football players. Furthermore, GCs are affixed mainly by Velcro straps and tend to slide off the players’ helmets during high-contact drills or scrimmages. Though the workouts included in this study were video verified and data would have been excluded had the GC come off completely, we cannot guarantee that the GC was not dislodged during some of the impacts. Our work would have also benefited from a second video reviewer simply to ensure high levels of quality control when reviewing the practice film for impacts. Lastly, no concussion rates were tracked during this study, as the performance period was too short, and such data may be more suited to a larger multiteam study to achieve sufficient power.

Consistent with the limited literature on GCs, we found no reduction in PAA, PLA, or total impacts when the GCs were affixed to traditional helmets in collegiate American football players. As significant head injuries are prominent in American football and other contact sports and with only a small literature base examining the effects of RHIs, continued research into improvements to contact sports equipment is needed.

Funding was provided by the Neuroscience COBRE (P20GM103650), the Robert Z. Hawkins Foundation, and the Nevada DRIVE Scholarship. We thank all our participants who helped make this project possible. These results do not constitute endorsement by the American College of Sports Medicine.