Head Acceleration Events Measured by Instrumented Mouthguards in Elite Australian Football

Abstract

Background

The consequences of exposure to head acceleration events (HAEs) in collision sports, including Australian Rules football, are a growing concern. Instrumented mouthguards (iMGs) now enable accurate quantification of HAEs in sport; however, little is known about HAE incidence and mechanisms in Australian Rules football. We aimed to measure HAEs in elite Australian Rules football players, and evaluate the influence of playing position and match event on the incidence and magnitude of HAEs in men and women. iMG data were collected from Australian Football League (AFL) and AFL Women’s (AFLW) players in 2023 utilising the HIT-IQ Nexus Gen III iMG, targeting 500 player-matches per sex. Match footage was synchronised with iMG data for video verification and match play coding.

Results

Three hundred and eleven players contributed iMG data in the analyses, including 126 men (median matches per player: 3 (interquartile range [IQR: 2–6]) and 185 women (median: 3 [IQR: 1–4]). Men had a higher mean incidence of HAEs ≥ 8 g per match than women (5.7 [95% confidence interval CI: 5.0–6.4] vs. 3.0 [95% CI: 2.8–3.3]), however the incidence was comparable when adjusted for time (3.6/hour [95% CI: 3.2–4.1] vs. 3.1/hour [95% CI: 2.9–3.4]). Men recorded higher maximum single-impact peak linear acceleration (PLA; median 31 g vs. women: median 22 g; p < 0.001) and peak rotational acceleration (PRA; median 3.0 krad/s² vs. women: 2.4 krad/s²; p < 0.001) per match. Conversely, women experienced greater PRA per impact (median 1.4 krad/s² vs. men: 1.2 krad/s²; p < 0.001), while PLA was similar (men: median 14 g, women: median 13 g; p = 0.330). The leading match events resulting in HAEs were marking contests (1.0/hour) in men and contested ball situations (0.8/hour) in women. HAE incidence and magnitude were generally comparable across playing positions, except for men’s rucks, who experienced significantly higher incidence of HAEs (6.5/hour) than forwards (3.6/hour), midfielders (3.2/hour), and defenders (2.7/hour).

Conclusions

This study highlights differences in HAE incidence, magnitude, and mechanisms by sex, playing position, and match events in elite-level Australian Rules football. These findings provide insight into impact exposure patterns and may inform targeted skill development and policy adjustments to enhance player safety.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40798-026-00982-6.

Key Points

• In elite-level Australian Rules football, men sustained more instrumented mouthguard-measured head acceleration events (HAEs) per match compared to women; however, incidence was similar after adjusting for time played.

• Men experienced higher maximum single-impact peak linear and rotational accelerations values per match, while women experienced greater peak rotational acceleration per HAE.

• Men’s rucks had a significantly higher incidence of HAEs than defenders, midfielders, and forwards, primarily due to more frequent low-magnitude impacts from ruck and marking contests.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40798-026-00982-6.

Introduction

Head acceleration events (HAEs) resulting from contact to the head or body in collision sports can cause concussion [1], and there is growing concern that repeated exposure to non-concussive HAEs may adversely affect long-term brain health. This concern has driven research into wearable sensors for quantifying HAEs, with instrumented mouthguards (iMGs) emerging as a key technology. These devices have demonstrated high accuracy in laboratory settings [2] and superior on-field reliability compared to earlier wearable technologies, with their use already enabling a greater understanding of the mechanisms of HAEs in rugby league and rugby union [3–9].

Australian Rules football is a high-speed, non-helmeted collision sport characterised by frequent physical contests [10, 11]. These include tackling (where one player physically restrains another, with potential HAEs for either player), contested ball situations (contact while competing for possession), marking contests (attempting to catch or spoil a ball kicked over a minimum distance, often involving aerial collisions), and ruck contests (where players jump or make contact to tap the ball). Playing positions are broadly categorised as defenders, midfielders, forwards, and rucks, each associated with different patterns of play [10, 11]. Women’s participation in Australian Rules football has grown significantly [12, 13], with the elite-level women’s competition (AFLW) commencing in 2017 under similar rules and field dimensions as the men’s elite competition (AFL), though with approximately 40% shorter matches and two fewer players per team on the ground [14–16]. Recent analyses found concussion rates of 6–7 and 7–10 per 1000 player hours in the AFL and AFLW respectively, with marking contests the leading cause in men, and bumps and tackles the primary contributors in women [17, 18].

While concussion incidence in elite-level Australian Rules football is well documented [17, 19–21], the incidence, magnitude, and causes of HAEs—regardless of clinical signs and symptoms—remain less understood than in the rugby codes. The last elite-level study on HAEs, conducted on 2017 AFL/AFLW data using skin-mounted sensors, was limited by a small sample size, and an analysis restricted to impacts within a single match and peak linear acceleration (PLA) ≥ 30 g, with an absence of peak rotational acceleration (PRA) metrics due to device reliability issues [22]. In addition, ruck contests—where frequent and inherent collisions occur—were not specifically analysed, leaving a gap in understanding the HAE burden in this position. Since then, the game has continued to evolve, particularly in the AFLW, further emphasising the need to reassess HAE mechanisms using improved methodologies. Advances in iMG technology, along with access to high-definition multi-camera video and timekeeper records for precise alignment of impact events ≥ 8 g, now provide an opportunity to gain new insights into HAE characteristics in modern elite-level men’s and women’s Australian Rules football.

The primary aim of this study was to assess the incidence and magnitude of iMG-measured HAEs in men’s and women’s Australian Rules football, and, within each sex, evaluate variation by playing positions, match events, and impact locations. The secondary aim was to assess the positive predictive value of the HIT-IQ iMG.

Methods

Data Collection

This prospective observational cohort study used iMG data available from the 2023 AFL and AFLW seasons. All participating players provided written informed consent. In total, 154 AFL-listed (43% cohort; player matches, n = 1193 [58%]) and 206 AFLW-listed players (57% cohort; player matches, n = 881 [42%])) played at least one game wearing the HIT-IQ Nexus Gen III iMG (HIT-IQ Pty. Ltd., Melbourne, VIC, Australia) in 2023. These players were from a broader cohort of players who signed informed consent for a multi-year iMG study. Each participating team (15 AFL teams, 18 AFLW teams) managed their iMG units and uploaded match data, which the research team accessed in its processed form via HIT-IQ’s Nexus portal. Matches were randomly selected based on the availability of match footage, iMG data (i.e. at least one player with iMG data in the match), and timekeeper records. Where possible, 2–3 matches per round were analysed, except in rounds without timekeeper data. Randomisation was applied until 500 player matches with iMG data were reached for both the AFL and AFLW.

Athletes wore the Nexus Gen III iMG which includes an on-tooth proximity sensor to reduce false positive captures, a triaxial accelerometer (Analog Devices ADXL372: sampling rate of 3,200 Hz, measuring range: ±200 g, 12-bit), and a gyroscope (Bosch BMI270: sampling rate of 800 Hz, measuring range ± 2,000 degrees per second, 16-bit). Impacts with a transformed value of ≥ 8 g at the centre of gravity (CoG) of the head were analysed, whereby the recording trigger threshold was ≥ 8 g at the iMG location (i.e. mouth).

Player age, height, and primary playing position (condensed into four categories: forwards, defenders, midfielders, or rucks) were obtained from public player profiles. Time on ground was obtained from public match statistics per player, allowing for incidence calculations to be relative to actual time played, and accounting for the difference in match duration between the AFL and AFLW matches. Match footage and timekeeper records were provided by the Australian Football League. Total number of ruck and marking contests, tackles, and contested balls per match were obtained from official league statistics and reflect all matches and players across the competitions; these data were used for contextual reference only, as the study sample comprised a subset of this broader population.

Video-Verification

Timekeeper records and multi-angle video footage (AFL: Broadcast, Behind-the-Goal Left/Right, Side View; AFLW: Broadcast, Behind-the-Goal, while Side View was available for 78% of matches) were used to align timestamped iMG data, using Catapult Focus (Version 7.7.0; Catapult, Melbourne, VIC, Australia). iMG captures reaching the Nexus portal (i.e., data passed through impact classifiers, including on-tooth detection) were reviewed using video footage and coded as either verified true positive, verified false positive, inconclusive, or inactive (see Online Resource 1). For true positive impacts, further coding detailed the match event in which the HAE occurred and impact location. Match events were categorised as tackles, marking contests, ruck contests, or contested ball. From here, tackles were separated into tackled or tackling, and ruck contests were divided into jumping or wrestling contests. Impact location was coded as either head (direct contact) or body (all other body locations). Impacts were verified and coded by two raters (LJE, JWH, WTO), with an agreement rate (total no. impacts with agreement/total no. impacts coded) of 94%. Any discrepancies in impact verification (i.e., true positive, false positive) were reviewed by all three raters, and a consensus decision was made.

Statistical Analysis

The positive predictive value (PPV) was defined as the number of true positives divided by the sum of true positives and false positives. Three PPV values were calculated: (i) an overall PPV, which was calculated for all captures for the entire season, irrespective of player or match; (ii) a per player match PPV, in which the PPV was calculated for each player match; (iii) a per player PPV, in which the PPV was calculated for each player across all matches participated in throughout the season. Median and interquartile range (IQR) were reported for the per player match and per player PPV, whilst the overall PPV was reported as a percentage. All continuous data were inspected for normality through visualisation of the distribution and the Shapiro-Wilk test. Parametric or non-parametric tests were subsequently used as appropriate. Mann-Whitney U tests were used to assess for differences in age and height between the men’s and women’s cohorts, whilst the Chi-square test was used to compare the breakdown of playing positions between competitions.

The mean incidence of HAEs per player from ≥ 8 g through to ≥ 60 g was calculated relative to their number of matches and total season hours using a bootstrapping procedure, featuring resampling with 2,500 iterations. Mean incidence was calculated using the number of impacts that occurred above each PLA threshold for every player match, before calculating the bootstrapped mean and 95% confidence interval (95% CI), with the lower and upper bounds defined as the 2.5th and 97.5th percentiles, respectively. Overlap of these CIs represents no difference in the incidence of HAEs between the respective groups, whereas a statistically significant difference is observed when the CIs do not overlap. This approach was similarly performed per hour on ground within the men’s and women’s competitions. Additional analyses were performed separated by match event, playing position, impact location (head vs. body), and within men’s rucks for match event, ruck contest type, and impact location (head vs. body). Ad hoc analyses of HAE incidence in men’s rucks were performed given the results of the group-level playing position analyses. The same results were not observed in women’s playing positions, and therefore no further analyses were performed.

Mann-Whitney U tests were used to assess PLA and PRA magnitudes for each impact, the median PLA and PRA per player match, and maximum PLA and PRA per player match. Kruskal Wallis tests were used to assess for overall effects of (i) match event and (ii) playing position on PLA and PRA, with Dunn’s multiple comparison test used where appropriate. The Mann-Whitney U test was used to assess for differences in PLA and PRA in HAEs sustained to the head or body. Wilcoxon effect sizes were calculated to assess differences in head impact kinematics between men and women overall, and by impact location. Pearson’s correlation analyses were used to assess the association between HAE incidence and the number of centre bounces and ruck contests attended within men’s rucks. All statistical tests were two-tailed, with significance set at p < 0.05. Analyses were conducted using RStudio (V.4.2.2) and GraphPad Prism (V.10.2.0).

Results

Three hundred and eleven players were included in the analysis (men: n = 185 [60%]; women: n = 126 [40%]. There was no difference in age between the competitions (men, median [IQR]: 25 years [22–28]; women: 24 years [21–28]; p = 0.052), whilst males were taller than females (men: median [IQR]: 188 cm [183–194]; women: 170 cm [166–177]; p < 0.0001) (Online Resource 2). A total of 500 player matches for the men’s competition, and 513 player matches for the women’s competition were analysed. For one match in the men’s competition, verification of more than one verified true positive impact was not possible (which is required for alignment of the footage), and the match was subsequently excluded. As such, a final number of 499 player matches (79 matches, 37% of season) were analysed for the men’s competition, and 513 player matches (56 matches, 57% of season) were analysed for the AFLW.

iMG Performance

The overall PPVs were 84.8% and 94.0% for the AFL and AFLW, respectively. Accounting for the potential influence of individual player behaviours and malfunctioning or ill-fitting iMGs, improved performance with perfect to near-perfect PPVs was observed when calculated per player (men, median: 90.5% [IQR: 77.8–100]; women, median: 100% [94.6–100]) and per player match for both competitions (men, median: 100% [IQR: 80–100]; women, median: 100% [100–100]; Online Resource 3).

Incidence and Distribution of HAEs

Men had a higher mean incidence of HAEs per player match across all PLA thresholds from ≥ 8 g (men: 5.7 [95% CI 5.0–6.4]; women: 3.0 [95% CI 2.8–3.3]) through to ≥ 60 g (Men: 0.1 [95% CI 0.1–0.2]; women: 0.1 [95% CI 0.1–0.1]; Fig. 1A). When adjusting for time on ground, the mean incidence per player match hour was similar for men and women across all thresholds from ≥ 8 g (men: 3.6/h [(95% CI 3.2–4.1]; women: 3.1/h [95% CI 2.9–3.4] to ≥ 60 g (men: 0.2/h [95% CI 0.1–0.3]; women: 0.1/h [95% CI 0.1–0.1]; Fig. 1B).

Fig. 1.

Incidence of HAEs measured by instrumented mouthguards in men’s and women’s elite Australian Rules football. Men experienced a greater incidence of HAEs ≥ 8 g through to ≥ 60 g per player match (A). No difference in HAE incidence was observed between AFL and AFLW players after adjusting for time played (B). Solid lines represent the bootstrapped mean incidence per player, while shaded areas denote the corresponding bootstrapped 95%CI. Non-overlapping CIs denote significant difference between men and women players. 2023 M: HAEs, n = 2914. Players, n = 126. Player matches, n = 499. 2023 W: HAEs, n = 1596. Players, n = 185. Player matches, n = 513. HAE = head acceleration event. 95%CI = 95% confidence interval. 2023M = Men’s 2023 season. 2023W = Women’s 2023 season

There was no difference between men and women in the distribution of PLA of impacts overall (Fig. 2A), or the median PLA per player match (Fig. 2B); however, men sustained higher maximum PLA per player match compared to women (men, median: 31 g [IQR: 19–50]; women, median: 22 g [IQR: 14–37]; r = 0.23 [95% CI 0.17–0.29]; p < 0.001; Fig. 2C). AFLW players experienced higher overall PRA per HAE compared to men (men, median: 1.2 krad/s² [IQR: 0.8–2.0]; women, median: 1.4 krad/s² [IQR: 0.9–2.2]; r = 0.08 [95% CI 0.05–0.11]; p < 0.001; Fig. 2D). Furthermore, women sustained a greater median PRA per player match (men, median: 1.3 krad/s² [IQR: 1.0–1.7]; women, median: 1.4 krad/s² [IQR: 1.1–2.0]; r = 0.14 [95% CI 0.08–0.20]; p < 0.001; Fig. 2E), but a lower maximum PRA per player match (men, median: 3.0 krad/s² [IQR: 1.8–5.4]; women, median: 2.4 krad/s² [IQR: 1.5–3.9]; r = 0.16 [95% CI 0.10–0.23]; p < 0.001; Fig. 2F). Further HAE incidence and distribution results are available in Online Resource 4.

Fig. 2.

Head impact kinematics measured by instrumented mouthguards in men’s and women’s elite Australian Rules football. No significant differences were observed in overall impact PLA (A) or median PLA per match (B) between men and women. Men experienced significantly greater maximum single-impact PLA (C) and PRA (F), while women experienced significantly higher overall PRA (D) and median PRA per match (E). ***p < 0.0001. NS = p≥0.05. PLA and PRA are presented on a log transformed scale. Boxes represent median (solid black line), and 25th and 75th percentiles, respectively. Whiskers represent Tukey’s limits (1.5x IQR). PLA = peak linear acceleration. PRA = peak rotational acceleration. 2023M = Men’s 2023 season. 2023W = Women’s 2023 season. AFL: HAEs, n = 2914. Players, n = 126. Player matches, n = 499. AFLW: HAEs, n = 1596. Players, n = 185. Player matches, n = 513instrumented mouthguards

Impact Location

In the AFL, the mean incidence of HAEs ≥ 8 g occurring due to contact to the head was similar (1.57/h, 95% CI 1.36–1.81) compared to the body (1.59/h, 95% CI 1.36–1.84) (Fig. 3A). Similarly, in the AFLW, the mean incidence of HAEs ≥ 8 g occurring due to contact to the head was comparable (1.35/h, 95% CI 1.18–1.53) to that resulting from contact to the body (1.38/h, 95% CI 1.19–1.59) (Fig. 3B). An overall effect of impact location on PLA was observed for both the AFL (r = 0.19, 95% CI 0.16–0.23, p < 0.001) (Fig. 3C) and AFLW (r = 0.19, 95% CI 0.14–0.24, p < 0.001) (Fig. 3D), with a greater incidence of HAEs sustained by contact to the head. Likewise, an overall effect of impact location on PRA was seen in both the men’s (r = 0.13, 95% CI 0.10–0.17, p < 0.001) (Fig. 3E) and women’s (r = 0.14, 95% CI 0.09–0.20, p < 0.001) competitions (Fig. 3F), with a higher number of HAEs sustained via contact to the head.

Fig. 3.

The incidence and head impact kinematics of HAEs by impact location. The incidence of HAEs resulting from contact to the head and body was comparable in both men (A) and women (B). HAEs caused by contact to the head led to significantly higher PLA and PRA in men (C, E) and women (D, F) compared to HAEs caused by body impacts ***p < 0.0001. NB: HAEs with undetermined impact location are not represented here (AFL, n = 378, 13%; AFLW: n = 241, 15%). Solid lines represent the bootstrapped mean incidence per player, while shaded areas denote the corresponding bootstrapped 95%CI. Non-overlapping CIs denote a significant difference between playing positions. PLA and PRA are presented on a log transformed scale. Boxes represent median (solid black line), and 25th and 75th percentiles, respectively. Whiskers represent Tukey’s limits (1.5x IQR). PLA = peak linear acceleration. PRA = peak rotational acceleration. 2023M = Men’s 2023 season. 2023W = Women’s 2023 season. 95%CI = 95% confidence interval. AFL: Players, n = 126. HAEs, n = 2914. Head, n = 1271 (44%); Body n = 1265 (43%). AFLW: Players, n = 185. HAEs, n = 1596. Head, n = 711 (45%); Body, n = 644 (40%). The locations of impact for the remaining HAEs (AFL: 13%, AFLW: 15%) were unable to be determined and are not presented here

Match Event Causes of HAEs

The highest incidence of HAEs ≥ 8 g in the men’s competition occurred in marking contests (1.0/h [95% CI 0.8–1.2]), followed by contested ball (0.8/h [95% CI 0.7–0.9]), tackling an opposition player (0.6/h [95% CI 0.5–0.7]), and being tackled by an opposition player (0.5/h [95% CI 0.4–0.7]), with the lowest incidence occurring from ruck contests (0.3/h [95% CI 0.1–0.4]; Fig. 4A). In contrast, in the women’s competition, the highest incidence of HAEs ≥ 8 g were sustained in contested ball collisions (0.8/h [95% CI 0.7–0.9]), followed by tackling an opposition player (0.7/h [95% CI 0.6–0.9]), marking contests (0.6/h [95% CI 0.5–0.7]), and being tackled by an opposition player (0.5/h [95% CI 0.4–0.6]). Similarly, the lowest incidence occurred from ruck contests (0.2/h [95% CI 0.1–0.3]; Fig. 4B). The per hour distribution of match events (irrespective of HAEs; as per official league statistics) was comparable in the AFL and AFLW in 2023, with marking contests the most frequently occurring (AFL: 4.3/h, AFLW: 5.1/h), followed by tackling (AFL: 2.6/h, AFLW: 4.5/h), being tackled (AFL: 1.7/h, AFLW: 3.3/h), ruck contests (AFL: 1.3/h, AFLW: 1.9/h) and contested balls (AFL: 0.8/h, AFLW: 1.3/h) (Online Resource 7).

Fig. 4.

The incidence and head impact kinematics of HAEs by match event. Differences in the incidence of HAEs across common match play events were observed between men (A) and women (B). Men experienced the greatest number of HAEs in marking contests and contested ball events, while contested balls and tackling another player were the primary causes in women. No differences were observed in the distribution of PLA and PRA in men (C, E) and women (D, F) between match events. NB: Only the five main match events are presented here, with the remaining impacts accounting for 305 HAEs (10%) in AFL, and 169 HAEs (11%) in AFLW – see Online Resource 7 for all match event data. Solid lines represent the bootstrapped mean incidence per player, while shaded areas denote the corresponding bootstrapped 95%CI. Non-overlapping CIs denote significant difference between match play events. PLA and PRA are presented on a log transformed scale. Boxes represent median (solid black line), and 25th and 75th percentiles, respectively. Whiskers represent Tukey’s limits (1.5x IQR). AFL: HAEs, n = 2914. Players, n = 126. Player matches, n = 499. AFLW: HAEs, n = 1596. Players, n = 185. Player matches, n = 513. HAE = head acceleration event. PLA = peak linear acceleration. PRA = peak rotational acceleration. 95%CI = 95% confidence interval

A significant effect of match event was observed for PLA in the AFL (Kruskal Wallis H-statistic = 14.15, 95% CI 0.00–0.01, 02p = 0.007) (Fig. 4C), but not the AFLW (H = 5.65, 95% CI 0.00–0.01, p = 0.227) (Fig. 4D). In contrast, an overall effect of match event was observed for PRA in both the AFL (H = 66.09, 95% CI 0.01–0.04, p < 0.001) (Fig. 4E), and the AFLW (H = 33.30, 95% CI 0.01–0.04, p < 0.001) (Fig. 4F). For the AFL, post-hoc analyses revealed that contested ball collisions resulted in a greater median PLA (15 g [IQR 10–24]) than being tackled by an opposition player (13 g [IQR 10–20]; z = 3.50, p = 0.005), but no other differences between match events. Furthermore, in the AFL there was a greater median PRA for HAEs sustained from contested ball collisions (1.4 krad/s² [IQR 0.9–2.3]) compared to marking contests (1.1 krad/s² [IQR 0.7–1.8]; p < 0.001), ruck contests (1.2 krad/s [IQR 0.8–1.9]; p = 0.006), and being tackled (1.1 krad/s² [IQR 0.7–1.7]; p < 0.001). Furthermore, the median PRA for HAEs sustained from tackling an opposition player (1.3 krad/s² [IQR 0.8–2.1]) was higher compared to marking contests (1.1 krad/s² [IQR 0.7–1.8]; p < 0.001) and being tackled by an opposition player (1.1 krad/s² [IQR 0.7–1.7]; p = 0.002). In the AFLW, the median PRA for HAEs sustained during contested ball collisions (1.5 krad/s² [IQR 1.0–2.4]) was greater than marking contests (1.2 krad/s² [IQR 0.7–1.9]; p < 0.001) and being tackled by an opposition player (1.3 krad/s² [IQR 0.8–2.0]; p = 0.014). Furthermore, the median PRA for HAEs sustained whilst tackling an opposition player (1.5 krad/s² [IQR 1.0–2.3]) was greater than marking contests (1.2 krad/s² [IQR 0.7–1.9]; p < 0.001) and being tackled (1.3 krad/s² [IQR 0.8–2.0]; p = 0.042). Full post-hoc comparisons are available in Online Resources 10–12.

Positional Differences

Men’s rucks experienced the greatest mean incidence of HAEs ≥ 8 g (6.46/h, 95% CI 4.53–8.55) compared to forwards (3.63/h, 95% CI 3.04–4.33), midfielders (3.17/h, 2.57–3.85), and defenders (2.72/h, 95% CI 2.27–3.24) (Fig. 5A). In the AFLW, rucks (3.46/h, 95% CI 2.72–4.28) and midfielders (3.47/h, 95% CI 2.87–4.11) experienced the highest mean incidence of HAEs ≥ 8 g, compared to forwards (2.93/h, 95% CI 2.49–3.42) and defenders (2.64/h, 95% CI 2.24–3.09) (Fig. 5B). No overall effect of playing position on PLA per impact was observed for either the men’s (H = 0.87, 95% CI 0.00–0.00, p = 0.832) (Fig. 5C), nor women’s (H = 3.29, 95% CI: 0.00–0.01, p = 349) competitions (Fig. 5D). Similarly, no overall effect of PRA was observed in men’s (H = 3.26, 95% CI 0.00–0.00, p = 0.353) (Fig. 5E) nor women’s (H = 4.80, 95% CI 0.00–0.01, p = 0.187) competitions (Fig. 5F).

Fig. 5.

The incidence and head impact kinematics of HAEs by playing position. Men’s rucks experienced the greatest incidence of HAEs compared to defenders, forwards, and midfielders (A). No difference was observed in the incidence of HAEs across playing positions in women (B), nor in the distribution of PLA and PRA in men (C, E) and women (D, F). Solid lines represent the bootstrapped mean incidence per player, while shaded areas denote the corresponding bootstrapped 95%CI. Non-overlapping CIs denote significant difference between playing positions. PLA and PRA are presented on a log transformed scale. Boxes represent median (solid black line), and 25th and 75th percentiles, respectively. Whiskers represent Tukey’s limits (1.5x IQR). AFL: HAEs, n = 2914. Players, n = 126. No. players per position: Rucks n = 12.Defenders n = 38. Forwards n = 42. Midfielders n = 34. AFLW: HAEs, n = 1596. Players, n = 185. No. players per position: Rucks n = 19. Defenders n = 58. Forwards n = 62. Midfielders n = 46. HAE = head acceleration event. PLA = peak linear acceleration. PRA = peak rotational acceleration. 95%CI = 95% confidence interval

Men’s rucks sustained the highest mean incidence of HAEs ≥ 8 g in ruck contests (2.41/h, 95% CI: 1.56–3.36), followed by marking contests (2.08/h, 95% CI 1.39–2.70) (Fig. 6A). We observed a significant association between the number of HAEs sustained by men’s rucks and the number of ruck contests (Pearson’s r = 0.27, p = 0.048) (Fig. 6B), and centre bounces (Pearson’s r = 0.30, p = 0.024) attended (Fig. 6C). The mean incidence of HAEs ≥8 g in men’s rucks was higher in wrestling ruck contests (1.9/h, 95% CI 1.25–2.54) compared to jumping ruck contests (0.7/h, 95% CI 0.39–1.09) (Fig. 6D). Finally, men’s rucks experienced a greater mean incidence of HAEs ≥8 g via contact to the head (1.18/h, 95% CI 0.86–1.53) compared to the body (1.17/h, 95% CI 0.48–2.01) (Fig. 6E).

Fig. 6.

The characteristics of HAEs in men’s rucks. Men’s rucks experienced the greatest incidence of HAEs in ruck and marking contests (A). An association was observed between incidence of HAEs within a match and the number of ruck contests (B) and centre bounces (C) attended by men’s rucks. Wrestling ruck contests resulted in a greater incidence of HAEs compared to jumping ruck contests (D). Incidence of HAEs resulting from head and body contact were comparable at all PLA thresholds (E). NB: Only the five main match events are presented here, with the remaining impacts sustained during game celebrations (n = 3, 0.5%), during disposal (n = 11, 2%), or off the ball contact (n = 7, 1%). Solid lines represent the bootstrapped mean incidence per player, while shaded areas denote the corresponding bootstrapped 95%CI. Non-overlapping CIs denote significant differences between match play events. Figure A: HAEs, n = 659; Players, n = 12. Figure B: 3504 ruck contests attended by 12 ruckmen across 56 player matches. Figure C: 1049 centre bounces attended by 12 ruckmen across 56 player matches. Figures D-E: Players, n = 11; One ruck man contributed only one HAE that was not during a ruck contest. Figure D: Jumping (n = 64, 26%), Wrestling (n = 183, 74%). Figure E: HAEs in ruck contests resulting from contact to the head, n = 115 (47%). HAEs in ruck contests resulting from contact to the body, n = 107 (43%). HAEs in ruck contests where the impact location was unable to be determined on footage (n = 25, 10%) are not represented here. HAE = head acceleration event. 95%CI = 95% confidence interval

Full statistical details can be found in Supplementary Materials.

Discussion

This study provides the most comprehensive analysis of HAEs in elite-level Australian Rules football, utilising iMG data with high-definition, multi-camera angle footage for impact verification and coding across 79 AFL and 56 AFLW matches in 2023. Contested ball and marking contests were the main causes of HAEs in men, while contested ball collisions and tackling were the most common causes in women. Analysis of over 1000 player matches showed that while men had a higher mean incidence of HAEs per match, time-adjusted rates were similar between sexes. Men recorded higher maximum single-impact PLA and PRA per match, whereas women experienced greater PRA per impact. Impact incidence and magnitude were generally consistent across playing positions, except for men’s rucks, who sustained twice as many HAEs as other positions, due to frequent low-magnitude impacts in ruck and marking contests. Herein, we explore these findings in greater depth, examining their implications for impact exposure, player safety, and future research directions.

iMG HAE Data Reliability and Trigger Threshold Considerations

One aim of this study was to assess the reliability of data collected using HIT-IQ’s iMG technology. In 2023, athletes wore the upgraded Nexus Gen III iMG, which introduced a proximity sensor to reduce false positive recordings when the mouthguard was not worn or decoupled from the upper dentition. We found that this addition significantly reduced false positives compared to previous studies using the Gen II model [4, 23], with overall PPVs of 85% and 94% for the men’s and women’s competitions, respectively. These values improved further when median PPVs were calculated per player and per player match, with overall values skewed by a small minority of participants with particularly low PPVs, likely reflecting differences in player behaviour and iMG fit quality. Considering that most false positives were low-magnitude impacts (Online Resource 13), these findings suggest strong promise for using the updated iMG model without the need for video verification. However, as false negatives were not systematically coded, we cannot assess their occurrence with the iMG. Given the time-intensive and subjective nature of identifying false negatives, we prioritised validating true positive impacts across a large number of matches, and future studies should explore negative predictive value.

Related to this, false negatives are likely influenced by linear acceleration trigger bias, where acceleration at the iMG location does not always reflect acceleration at the CoG of the head [24, 25]. As a result, some impacts that exceed relevant acceleration thresholds at the CoG may not be captured if acceleration at the iMG remains below the trigger threshold. The Gen III iMG had a lower resultant linear acceleration trigger threshold (8 g) than the Gen II model (13 g), reducing the likelihood of missed impacts, but under-detection remains a consideration for lower-magnitude impacts [24]. Notably, a recent simulation study found that for a 10 g CoG acceleration, only 25% of impacts exceeded a 10 g single-axis threshold at the mouth location, compared to 32% using a 10 g resultant threshold [26]. For a 20 g CoG acceleration, these values increased to 86% for a 10 g single-axis threshold and 98% for a 10 g resultant threshold. This suggests that HAE incidence in our dataset, which used a resultant 8 g trigger, as well as in recent rugby studies using a single-axis 8 g trigger (using Prevent Biometrics iMG) [9, 27], may still be underestimated for lower-magnitude HAEs (see Tooby et al. [27] for a detailed discussion). To address this, we not only described overall impact incidence, but also examined HAE incidence across multiple thresholds to better characterise impact distributions and best enable comparison with other studies.

Overall HAE Incidence and Magnitude

The incidence of HAEs per match was higher in men than women across all linear acceleration recording thresholds; however, after accounting for time on ground, incidence rates were comparable. At the ≥ 10 g threshold, incidence was 4.5 and 2.4 HAEs per match in men and women, respectively, with adjusted rates of 2.9 and 2.4 per hour. At ≥ 20 g, rates were 1.6 and 0.9 per match, adjusting to 1.1 and 0.9 per hour. At ≥ 30 g, incidence was 0.8 and 0.4 per match, with adjusted rates of 0.6 and 0.4 per hour.

While differences in iMG models, trigger thresholds, and data processing must be considered, our study shares methodological and technological similarities with recent iMG studies of elite-level rugby, allowing for meaningful comparisons of HAE incidence and magnitude. For example, two recent studies of elite rugby union found HAE incidence in men ranged from approximately 15–20/h for impacts exceeding 10 g, 5–7/h above 20 g, and 1.5–3/h over 30 g, while in women, incidence was approximately half [9, 27]. As such, while the overall number of impacts appears to be significantly higher in elite rugby union than in elite Australian Rules football for both sexes—particularly in men—the difference is driven primarily by greater proportional differences observed at lower thresholds (i.e., 10–20 g). In contrast, the disparity in incidence narrows considerably at higher thresholds, with impacts ≥ 30 g showing a much smaller fold difference between the two sports. This pattern aligns with concussion incidence rates, which, while higher in elite rugby union (12 per 1,000 h in men [28]; 8–16 in women [29]) than in elite Australian Rules football (6–7 in men; 7–10 in women) [18], more closely mirror the disparity in high-magnitude HAEs rather than the much larger difference seen in lower-magnitude impacts.

Our study indicates that, unlike in elite-level rugby union, the incidence of HAEs per playing hour in elite Australian Rules football is comparable between men and women. However, AFL matches and seasons are considerably longer than in AFLW, leading to greater cumulative HAE exposure in men. We found that men recorded higher maximum PLA and PRA per match than women, likely reflecting both longer match durations—which increase the chance of exposure to high-magnitude impacts—and differences in player size and speed. Of relevance, recent preliminary evidence suggests that even in the absence of concussion, iMG-measured match cumulative total and maximum single-impact head acceleration magnitudes are associated with post-match blood levels of glial fibrillary acidic protein (GFAP), an objective marker of brain cell injury [23]. These findings are consistent with broader evidence from American Football and water polo, where repeated non-concussive HAEs over longer time frames have been linked to elevations in brain injury biomarkers [30–35]. However, further studies are needed to understand these associations. Moreover, it remains unclear whether greater exposure in men necessarily translates to a greater risk of brain injury compared to the lesser exposure in women. Longitudinal research is required to determine whether and how such exposure affects brain structure and function in both sexes, and whether monitoring thresholds for cumulative or single-impact HAE burden could help mitigate potential effects.

A notable finding in our study was that per-impact PRA, but not PLA, was higher in women than men. While the difference was relatively small (median 1.4 krad/s² in women vs. 1.2 krad/s² in men), it raises important considerations. Anatomical differences, including lower neck strength and muscle stiffness in women, may influence head kinematics by reducing the ability to stabilise the head and absorb rotational forces [36–38]. While some studies suggest greater neck strength may reduce concussion risk, others have found no significant association between neck strength and impact kinematics [36, 39, 40]. For a given level of rotational acceleration, female axons—due to their smaller diameter—may be more vulnerable to damage than male axons, as experimental models suggest they sustain greater strain under the same applied load [37]. Psychological factors, including anticipatory responses and perceived impact risk, may also contribute to sex differences in head impact kinematics [41]. These findings highlight the need for further research into how anatomical, biomechanical, and neurophysiological factors interact to influence HAE exposure and injury risk in female athletes, and whether neck strengthening and neuromuscular training interventions could help mitigate rotational head acceleration and improve player safety.

HAE Incidence and Magnitude by Match Events and Playing Position

While contested ball collisions resulted in the greatest frequency of HAEs ≥ 8 g in both sexes, men players sustained more HAEs in marking contests, whereas AFLW players experienced more HAEs when tackling another player. These findings are consistent with the main causes of concussion reported in the AFL and AFLW [17], and likely reflect variations in match play between the two competitions. Although the incidence of HAEs varied by match event, no significant differences in PLA or PRA magnitude were observed between these skills within each sex.

While our study did not assess relative propensity (i.e., risk per instance of each skill), we drew on season-wide data provided by the league to add context to the frequency of key match events in 2023 (Online Resource 8). These data reflect all matches and players across the AFL and AFLW, not just those included in our sample. Notably, marking contests occurred slightly more often in the AFLW than in the AFL (5.1 vs. 4.3 events/player/hour). When considered alongside our finding that men sustained more HAEs per hour in marking contests (0.8 in AFL vs. 0.5 in AFLW for impacts > 10 g), this suggests that the propensity for HAEs in marking contests may be somewhat higher in men. Tackling (4.5 vs. 2.6 events/player/hour) and being tackled (3.3 vs. 1.7 events/player/hour) were also more frequent in the AFLW and given the highly comparable HAE incidence between sexes for these events, this may indicate a higher per-event HAE risk during tackles in men’s matches. However, future studies are required to formally assess HAE propensity by accounting for the number of specific match events per player, enabling more accurate comparisons of per-event risk.

A prior study using the X-Patch in elite AFL and AFLW found no differences in HAE frequency among forwards, midfielders, and defenders, but grouped rucks and midfielders together, limiting insights into ruck-specific exposure [22]. The most notable difference in the present study was observed among men’s rucks, who experienced approximately twice the number of HAEs ≥ 8 g compared to other positions. This difference was largely driven by lower-magnitude impacts, as incidence in rucks was comparable to other positions at higher thresholds (> 16 g). These HAEs were primarily sustained during ruck and marking contests, with wrestling ruck contests contributing more impacts than jumping contests. Lastly, a significant association was found between the number of HAEs and the number of ruck contests and centre bounces attended. However, while this trend was not observed in AFLW rucks, interestingly, league statistics show that in 2023, women players participated in a greater mean number of ruck contests per hour than men (1.9 vs. 1.3 events/player/hour; Online Resource 8), suggesting that the elevated HAE exposure in male rucks may not be due to contest frequency, but rather to differences in how ruck contests are performed between competitions.

Limitations

Our study has limitations to consider. First, as previously noted, we did not assess false negative HAEs, limiting our ability to evaluate the negative predictive value of the iMG and HAE propensity. However, propensity values do not provide information surrounding the frequency of an event, and therefore did not align with our aim of examining the incidence of HAEs. Second, due to the recording trigger threshold of the iMG, many impacts that were actually ≥ 8 g at the CoG may not have been captured, preventing accurate estimation of true HAE incidence. However, this limitation is common across iMG studies and primarily affects very low-magnitude impacts. Furthermore, lowering the trigger threshold introduces other challenges, such as increased data noise. Finally, although multi-angle high-definition video review was a strength of the study, some AFLW matches were limited to two camera angles, which may have increased the number of inconclusive events.

Conclusion

HAEs are common in elite Australian Rules football, with important differences in incidence, magnitude, and mechanisms by sex, playing position, and match event. These insights improve understanding of impact exposure in the sport and may inform targeted skill refinement and safety strategies to minimise the risk of short- and long-term adverse neurological consequences of participation in elite Australian Rules football.

Abbreviations

AFL

Australian Football League

AFLW

Australian Football League Women’s

CI(s)

Confidence interval(s)

CoG

Centre of gravity

HAE

Head acceleration event

iMG

Instrumented mouthguard

PLA

Peak linear acceleration

PRA

Peak rotational acceleration

Author Contributions

SJM, BM, MM, CW, JR, TJO, and SRS were involved in study design and conceptualisation. LJE, JWH, JR, BX, KGJ, LPG, and WTO were involved in data acquisition and preparation. LJE, WTO, and SJM were involved in data analysis and interpretation. LJE, WTO, and SJM were involved in manuscript preparation. All authors contributed to manuscript review and approved the final manuscript.

Funding

Open Access funding enabled and organized by CAUL and its Member Institutions. Funded by Monash University and the Australian Football League.

Data Availability

The investigators do not have permission to disclose the data under the terms of the ethics approval and participant consent.

Declarations

Ethics Approval and Consent to Participate

This study was approved by the Monash University Human Research Ethics Committee (#31310). All participants provided written informed consent.

Consent for Publication

Not applicable.

Competing Interests

SRS sits on the AFL Players Association Health, Safety, and Wellbeing Committee. MM fulfils a paid consultancy role with the Australian Football League. JR, KGJ, and CW are employed by the Australian Football League.