Australian football (AF) is a contact game (Orchard et al., 1998) with tackling and collisions between players and the ground (Hrysomallis et al., 2006; Orchard et al., 1998). These collisions have the potential to result in head impacts that may lead to sports-related concussion (SRC); a mild traumatic brain injury associated with a range of symptoms (McCrory et al., 2013; Pearce et al., 2015). In the 9 to 17-year-old group, the overall incidence rate of SRC in AF has been reported at 0.6 (95% CI: 0.2 to 1.0) per 1,000 Athlete Exposures (A-E) with the older group (ages 14-17) recording 0.8 (95% CI: 0.1 to 1.5) per 1,000 A-E (Hecimovich and King, 2017). The exact forces resulting in SRC remains unknown and has been reported to be more likely due to rotational acceleration of the brain (Hoshizaki and Brien, 2004). It is still uncertain where in the brain a concussion occurs, or the exact origin of the symptoms of acute concussion (Hynes and Dickey, 2006). It is now apparent that direct impact to the head is not required and a concussion can occur with a blow to the chest, for example, that causes a whiplash effect on the neck and brain (Hynes and Dickey, 2006). Whiplash of the neck and brain and the incidence of concussion frequently co-exist (Hynes and Dickey, 2006). What is also known is that the young brain is more susceptible to concussion than the adult brain and may require more time to recover (Baillargeon et al., 2012). Therefore, a better understanding of impact metrics may help the recognition and recovery of SRC.
The quantification of head impacts in sport has been documented in the literature using a variety of devices. For example, mouthguard or head-mounted sensors (XPatch; X2biosystems, USA) have documented head impacts in AF (Hecimovich et al., 2018), rugby union (King et al., 2015; 2016), and rugby league (King et al., 2017). These studies have enabled the development of analytical risk functions (Pellman et al., 2003; Rowson et al., 2011; 2012; 2013), concussion risk curves (Rowson et al., 2012), and risk weighted exposure metrics (Urban et al., 2013) further assisting in the identification of athletes at risk of SRC.
Hecimovich et al. (2018) measured the frequency, magnitude and distribution of head impacts in juniorleague (aged 17-19) and senior-league (aged 20+) AF players, reporting that the resultant peak linear accelerations ranged from 10g to 158.8g with a median value of 15.3g. In the same study, the resultant peak rotational accelerations ranged from 2,996.6 to 1,286,748.6 deg/[s.sup.2] with a median value of 1,302,321.6 deg/[s.sup.2] (Hecimovich et al., 2018). There were no major differences observed between junior- and senior-league players, however there was only a small age gap between mean [+ or -] SD of cohorts (18.0 [+ or -] 0.7 yr. vs 21.0 [+ or -] 2.2 yr.).
Currently, there are no reported accelerations for AF players under the age of 17 yr. Therefore, the aims of this study were to: 1) Investigate the frequency, magnitude, and distribution of head impacts sustained by youth AF players on a single team over a season of games; and 2) compare the impact characteristics between players who self-reported sustaining a head impact during a game to those with no reported head impact.
A prospective observational cohort study was conducted on youth AF players on a single team competing during the 2017 competition season. All members of the team were invited to participate in the study. A total of 19 male youth AF players (age range 13-14 yr., mean [+ or -] SD 13.9 [+ or -] 0.3 yr.) were enrolled in the study. Consent was obtained from the players, parents and participating team before enrolling in the study. The researchers' University ethics committee approved all procedures (MUHREC 2016/012).
Over the course of the season (11 games) players were fitted with skin-mounted impact sensors to measure the impact frequency, magnitude and distribution of impacts and requested to respond to a post-game self-report logbook on recalling having a direct hit to their head with another player or striking their head to the ground.
Impact sensors and testing
All players enrolled in the study wore the XPatch impact-sensing skin patch (X2Biosystems Inc.) on the skin covering their right-side mastoid process for each game. The XPatch sensor, sampling at 1024 Hz, was placed behind the player's right ear just before participation in game activities and was removed immediately after completion of the game. The positioning of the XPatch over the mastoid process ensured that the sensor was not activated by enhanced soft-tissue effects when impacts occurred (Wu et al., 2016). The sensor contained a low-power, high-g triaxial accelerometer with 200g maximum per axis and a triaxial angular rate gyroscope to capture six degrees of freedom for linear and rotational time history accelerations of the heads center of gravity for all impacts that occurred during games. The time history incorporated three axes (x, y, z) of acceleration and velocity. While upright these planes describe the medial-lateral, anterior-posterior and vertical acceleration and deceleration. The Impact Management System (IMS) enabled the raw data to be transformed to the head center of gravity by using a rigid-body transformation for linear acceleration and a 5-point stencil for rotational acceleration (Wu et al., 2016). The biomechanical measures of head impact severity consisted of impact duration (ms), linear acceleration (g), and rotational head acceleration (deg/[s.sup.2]). Resultant linear acceleration is the rate of change in velocity of the estimated center of gravity of the head attributable to an impact and the associated direction of motion of the head (Mihalik et al., 2010). Resultant rotational acceleration is the rate of change in rotational velocity of the head attributable to an impact, and its direction in a coordinate system with the origin at the estimated center of gravity of the head (Mihalik et al., 2010). False impacts were removed by the X2Biosystems proprietary 'de-clacking' algorithm (King et al., 2015). Impacts with a resultant linear acceleration of
Head impact exposure including frequency, magnitude and location of impacts were quantified using previously established methods (Crisco et al., 2010; 2011). The impact variables were not normally distributed (Kolmogorov-Smirnov; p
Player head impacts exposure were assessed utilizing previously published levels for injury tolerance (linear >95 g and rotational acceleration >315,126.8 deg/s), impact (linear mild 106 g) and rotational acceleration (mild 452,636.7 deg/s) severity (Broglio et al., 2010; 2011a; Guskiewicz et al., 2007; Harpham et al., 2014; Ocwieja et al., 2012; Zhang et al., 2004). Two additional risk equations were included in the analysis of the head impact exposure data. The Head Impact Telemetry Severity profile ([HIT.sub.SP]) (Greenwald et al., 2008) is weighted composite score including linear and rotational accelerations, impact duration, as well as impact location. The Risk Weighted Exposure Combined Probability ([RWE.sub.CP]) (Urban et al., 2013) is a logistic regression equation and regression coefficient of injury risk prediction of an injury occurring based on previously published analytical risk functions. [RWE.sub.CP] combines resultant linear and rotational accelerations to elucidate individual player and team-based head impact exposure. The [HIT.sub.SP] and [RWE.sub.CP] were analyzed by player-position impacts utilizing a Friedman repeated measures ANOVA on ranks. A Wilcoxon signed-rank test post-hoc analysis was conducted with a Bonferroni correction applied if any significant differences were observed.
Resultant peak linear (PRA[g]) and rotational (PLA[deg/[s.sup.2]]) accelerations and impact locations (front, back, side and top) between player positions were assessed utilizing a Friedman repeated measures ANOVA on ranks. A post hoc analysis with a Wilcoxon signed-rank tests was conducted with a Bonferroni correction applied if any significant differences were observed. A one sample chi-squared ([chi square]) test and risk ratio (RR), with 95% confidence intervals (CI), were utilized to determine whether the observed impact frequency was significantly different from the expected impact frequency. Statistical significance was set at p
Qualitative data (logbooks)
Over the course of the season, participants completed an individual post-game logbook (Figure 1). Items included written feedback responses on recalling having a direct hit to their head with another player (their head, knee, elbow etc) or striking their head to the ground. Further, the logbook listed nine common concussion signs and symptoms (Meehan et al., 2010) for participants to indicate if during, or after, the game the player experienced. The logbooks were collected after its completion during the first training session following the weekend game.
Several steps were undertaken to exclude data identified as not representing on-field head impacts. First, the data contained on the XPatch were uploaded after each match onto the IMS provided by X2Biosystems. Next, the data were then downloaded and filtered through the IMS to remove any spurious linear acceleration that did not meet the proprietary algorithm for a head impact (Swartz et al., 2015). The data underwent a second filtering waveform parameter proprietary algorithm during data exporting to remove spurious linear acceleration data with additional layers of analysis (Swartz et al., 2015). This included the area under the curve, the number of points above threshold...