Skip to main content
NCBI home page
European Journal of Sport Science logoLink to European Journal of Sport Science
. 2025 Dec 19;26(1):e70105. doi: 10.1002/ejsc.70105

Wearing Regulation Soft‐Padded Headgear Does Not Reduce the Risk of Head Injuries in Professional Men's Rugby Players: An Observational Cohort Study

James Brown 1,2,3,✉, Marc Douglas 4, Ben Hester 4, Manish Mohan 1, Sharief Hendricks 1,2,5, Brady Wiseman 3, Matthew Boshoff 3, Stephanie Phillips 3, Michael Bromfield 3, Amy Howard 3, Lindsay Starling 1,4, Ben Jones 1,2,6,7,8, Ross Tucker 1,4
PMCID: PMC12717021  PMID: 41419445

ABSTRACT

There is no empirical evidence that soft‐padded headgear is protective against head injury risk in rugby. However, studies that have assessed purported protective effects have not accounted for rates of contact. The aim of this study was to compare head injury rates while considering tackle‐event exposure in players with and without headgear. In the 2018 and 2019 professional men's SuperRugby season, video analysts recorded headgear use, playing position, match time and head injury assessments (proxy for head injury risk) for each player. Tackle‐event involvements for each player were obtained from third‐party video analysis provider. Tackle‐related head injury rates were calculated per 1000 h (incidence) and per 1000 tackle‐events (propensity), and compared between headgear and non‐headgear wearers using incidence rate ratios (IRRs) with 95% confidence intervals and Poisson regression models. Players wearing headgear were involved in more tackles per match than players without headgear (IRR 1.07, 95% CI 1.05–1.09). Head injury incidence (IRR: 1.78 95% CI: 1.11–2.70) and propensity (IRR: 1.66 95% CI: 1.04–2.52) were higher in players wearing headgear. However, statistical models found no difference in this risk between positional groups. A lack of protective effect is consistent with previous studies and could be explained by World Rugby's headgear design regulations while increased risk may be a result of greater injury susceptibility. As World Rugby's headgear regulations change and further advancements in headgear is made, it is important to continue to examine their effect on head injury risk at an individual level.

Keywords: football, injury and prevention, team sport, technology

Highlights

  • This study extended previous studies that examined the protective effect of soft‐padded headgear in rugby union by quantifying suspected head injury risk at an individual level. This was achieved by measuring player game time (with and without headgear) and tackle event (tackling and ball carrying) involvements.

  • Even at an individual level, this study found no evidence of a protective effect of soft‐padded headgear for suspected head injuries in elite rugby players,

  • Certain positions, such as Hookers, showed an elevated risk of suspected head injury when wearing soft‐padded headgear, in comparison to the same positional group not wearing headgear.

1. Introduction

Rugby union (rugby) is a popular collision sport played by an estimated eight million players globally (https://www.world.rugby/organisation/about‐us/overview). Rugby is a sport in which players' roles and actions differ between positional groups of forwards and backs (Roe et al. 2025). On average, forwards are involved in more collisions per minute of play, whereas backs cover greater distances with more high‐speed running (Lindsay et al. 2015; Quarrie et al. 2013). At the elite level, each rugby match has 15 starting players and 8 replacements who may replace players either due to an injury or for tactical reasons (Lindsay et al., 2015).

Rugby has a relatively high risk of all injuries, including sports‐related concussion, to the player (Williams et al. 2021). Personal protective equipment (mouthguards and helmets) are recommended for concussion prevention (Eliason et al. 2023). However, the protective effects of helmets/headgear have mainly been shown in sports such as American Football that have hard‐shell helmets (Rowson et al. 2014). A recent meta‐analysis of that included studies from soccer and rugby found no protective effect of headgear use against concussion (Al Attar et al. 2024). Specifically in rugby, two studies have compared concussion rates in adult men wearing and not wearing regulation soft‐padded headgear (Marshall et al. 2005; Stokes et al. 2021). Although the earlier study (Marshall et al. 2005) found a reduction in minor head injuries (abrasions), neither study found a protective effect of wearing the soft‐padded headgear on concussion rates. The lack of protective effect in these two rugby studies could be partly explained by the strict regulation on design by World Rugby, the sport's global governing body (https://www.world.rugby/the‐game/facilities‐equipment/equipment/specifications/headgear) (McIntosh et al. 2009). The protective effect of thicker soft‐padded headgear, which was not compliant with World Rugby regulations, has also been investigated (McIntosh et al. 2009). Although this study also found no protective effect on concussion risk, the authors acknowledge a lack of compliance with wearing the thicker headgear could have affected the study analyses (McIntosh et al. 2009).

Importantly, all of the aforementioned studies examining the effect of headgear on concussion risk in rugby have compared concussion rates per minute of play at a group level (Marshall et al. 2005; McIntosh et al. 2009; Stokes et al. 2021). In rugby, collision rates differ substantially between positions, with flankers, hookers and eighth‐men having higher collision rates (Lindsay et al. 2015; Roe et al. 2025). Players in positions that are involved in more collisions might be more likely to wear headgear, either to reduce abrasions or more serious injuries. Therefore, approaches that assess concussion risk only as a function of time, and not exposure to risk events (tackles), may miss the true effect of headgear.

Accordingly, this study aimed to compare head injury risk in players with and without headgear, as a function of tackle‐event involvements.

2. Materials and Methods

2.1. Study Design

Observational cohort study comparing tackles and head injury numbers and rates in players who were and were not wearing headgear.

2.2. Setting

Video files of all matches for two seasons of the men's professional rugby competition—SuperRugby ‐ in 2018 and 2019. There were 253 matches in total: 127 in 2018 and 126 in 2019.

2.3. Participants

Elite adult male professional players from Argentina, Australia, Japan, New Zealand and South Africa. All players were included in the study.

2.4. Ethics

Ethics approval was received from Stellenbosch University (approval number U21/05/120).

2.5. Variables

The main outcome variable was a suspected tackle‐event related head injury to a player, which we used as proxy for head injury risk in tackle‐events. The main predictor variable was headgear use (yes or no) and position of the player was a secondary predictor variable. Exposure variables were minutes of play and tackle involvements (tackling and ball‐carries) for each player. Tackle‐event involvements was also used as a secondary outcome variable, with the predictor variables being headgear use and player position (exposure = minutes of play).

2.6. Data Sources/Measurement

All data except for tackle‐event involvements were collected by video analysis of the 2018 and 2019 match footage.

For each of the 253 matches that occurred during these time periods, the video analysts needed to record a complete list of any player who entered the match, either as a starting player or substitute.

Video analysts noted a suspected head injury if a player was confirmed or suspected to have entered World Rugby's head injury assessment (HIA) process (Raftery et al. 2016). Confirmation of the player entering the HIA process could be from the match commentators and/or broadcaster when the player was substituted. Outside of this match/broadcaster confirmation, video analysts were only told to suspect that a player entered the HIA process if there were clear signs (Criteria 1 signs) of a concussion to a player (Raftery et al. 2016).

Minutes of play was noted for each player for on‐field time. If a player was temporarily substituted (e.g., for a blood injury or yellow card) and returned to the field, the time that the player was off the field was not included. Excluding temporary substitutions, minutes of play was from the start of the match until substitution or match end (for match starters) or from the time a substitute entered the field until they were substituted or the match end (for substitution players).

Position was recorded as either the position in which the player started the match (match starters) or the position of the substituted player. For substituted players, positions were confirmed by the match commentators/broadcaster in the match footage, or from available online match reports. If this was not possible, it was simply assumed that the substituted player replaced the exact position of the player that they replaced. Player positional changes were not noted outside of this. Finally, the analysts also noted the match event the player was involved in (tackle, ball‐carry, maul and aerial collision) when the suspected head injury occurred.

Match‐play event data was obtained from Opta data (Stats Perform, London, UK) and downloaded online from their portal (https://www.optaprorugby.com/index.php) in extensible markup language (XML) format for each match (n = 253). The XML files were subsequently converted to comma‐separated value (CSV) files using Python programing language (version 3.10.9) and the ElementTree (etree) package. The resulting CSV files included key fields such as the match date, a unique fixture ID, player ID, position category, event ID, event timestamp and event type (e.g., ball carries or tackle involvements). A tackle‐event was defined by Opta as “when a player has attempted to halt the progress or dispossess an opponent in possession of the ball.” Although collisions/impacts are possible in other aspects of rugby (e.g., the scrum, ruck and maul), the tackle‐event accounts for the majority of contact involvements across all players (Quarrie et al. 2013). Data were filtered to isolate only tackle‐events, which were then grouped by player ID and fixture ID to calculate the number of tackle involvements for each player in each match for each video analysed headgear usage. The event timestamps from Opta and video analysis were used to account for each player's on‐field time, ensuring an accurate assessment of tackle events involvement.

2.7. Bias

As with any video analysis study, this study was subject to misclassification bias by the video analysts or Opta data. This is especially true if there are multiple coders, as we had in the present study or as is the case for Opta data generation. To reduce the chances of misclassification bias in the video analysis component, we split the coders, one of whom identified all players who entered the match and whether they were wearing headgear, while a second coder watched the same matches and noted the playing time of each player. The division of roles reduced the chances of errors being recorded. The noting of suspected head injuries was also subject to error, but the total numbers were compared against World Rugby's recorded HIAs for in these two seasons and there was less than 5% difference in the overall counts. Opta data demonstrated excellent reliability (kappa values: 0.92–0.94) in football (soccer) (Hongyou et al. 2015). Although the reliability of Opta data is yet to be empirically tested in rugby, there are multiple studies that have used it, both in the realm of performance (Scott et al. 2023) and more clinical/sports medicine outcomes (Tooby et al. 2024).

This study is also subject to selection bias in that players wearing headgear might be different to players not wearing headgear in ways that we were unable to measure. Although we are measuring and comparing positions and tackle‐event involvement rates between headgear and non‐headgear wearing players, we did not have information on previous injuries or injury history.

Finally, two different sources were used for our exposure variables (match involvements) and this could also be a source of bias: the number of minutes on the field was captured by the video analysts, while tackle‐event involvements was captured by a third‐party commercial entity (Opta) as described in the previous paragraph. However, this also served as an additional check of the players recorded as playing in each match by the video analysts.

2.8. Study Size

As our main outcome of interest was suspected head injuries and main predictor variable was headgear use, we made some estimations for a suitable number of matches. Based on the finding of Stokes that 16% of concussed players were wearing headgear at the time of their injury (Stokes et al. 2021), we expected that a higher proportion of our entire player cohort would wear headgear (20%). Also, we assumed that there would be one suspected head injury per match and that this suspected head injury would be equally likely to a player wearing headgear, and equally likely in any of the playing positions.

Based on these three assumptions, 200 matches would provide 200 suspected injuries, and 40 of these suspected head injuries would be to players wearing headgear, a sample we considered sufficient to investigate our main predictor variable (headgear use). The 2018 and 2019 SuperRugby seasons offered one competition with more than 200 matches (253 in total) with high quality broadcaster footage and contributed to our selection.

2.9. Quantitative Variables

The main outcome of suspected head injury was a binary variable for each player who took to the field for every match: yes (suspected head injury) or no (no suspected head injury). Only suspected injuries that took place in a tackle event were included in the analyses.

The main predictor variable (headgear use) was a binary outcome for each player who took to the field for every match: yes (headgear) or no (non‐headgear).

The secondary predictor variable was the 15 rugby positions which were grouped in two ways, according to previous research describing positional groupings based on match involvements (Quarrie et al., 2013).

  1. Binary: (1) Forwards (loose‐head prop, tighthead prop, hookers, right lock, left lock, blind‐side flanker, open‐side flanker and eighth man) or (2) backs (scrum‐half, flyhalf, inside centre, outside centre, right wing, left wing and full‐back).

  2. 10 categories: (1) Props (2) props: loosehead and tighthead props), (2) hookers, (3) locks (2 locks: right lock and left lock), (4) Flankers (2 flankers: open‐side and blind‐side flanker), (5) eighth man, (6) scrumhalf, (7) Flyhalf, (8) midfield backs/centres (2 centres: inside and outside centre), (9) wings (2 wings: right wing and left wing) and (10) Full‐back.

The two exposure variables were both continuous in nature: minutes of match play and tackle‐event involvements. Minutes of match play was also converted to hours of match played, wherever convenient to do so.

Tackle‐event involvements per player were obtained from Opta and included both tackler and ball‐carrying events. Opta's definition of a tackle is “A player has attempted to halt the progress or dispossess an opponent in possession of the ball.” This is comparable with the consensus statement definition for coding a tackle event in rugby (Hendricks et al. 2020) In other words, a tackle is observed regardless of whether the tackled player was brought to ground or not. Ball‐carrying events are called “Carries” by Opta is defined as: “A player touching the ball has deemed to make a carry if they have made an obvious attempt to engage the opposition with the ball in hand.” The consensus statement definition of ball‐carrying is simply the “player carrying the ball in a tackle event” (Hendricks et al. 2020) which is very similar to Opta's. Tackle involvements were the sum of both tackles and ball‐carrying activities that could be attributed to a player.

2.10. Statistical Methods

Tackle event rates were calculated per 80 min (1 regular match in professional rugby). Suspected head injuries were calculated as an incidence (per 1000 match hours) and propensity (per 1000 tackle event involvements) (Bahr et al. 2020). To compare the incidence and propensity, unadjusted rates between headgear and non‐headgear wearing groups and relative rate ratios along with their 95% confidence intervals, were calculated. For overall relative rate ratios for headgear versus non‐headgear (e.g. for all positions), the finding was considered statistically significant if both the upper and lower 95% confidence intervals didn't cross 1.00 (Kirkwood and Sterne 2003).

A Poisson regression was run using the glm function of the R base package (R Core Team 2021; Vienna, Austria). An interaction term was performed between headgear and positional grouping (either forward vs. back or the 10 positional groupings) and the p‐value for the interaction term recorded. The relevant exposure variable was included as an offset with a log link function.

In total 6 different Poisson regression models were run.

  1. Outcome: count of tackle‐events. Predictor variables: headgear (non‐headgear vs. headgear) * positions (10 positions as listed above). Exposure: minutes of match time.

  2. Outcome: count of tackle‐events. Predictor variables: headgear (non‐headgear vs. headgear) * positional groupings (forwards/backs). Exposure: minutes of match time.

  3. Outcome: count of suspected HIAs. Predictor variables: headgear (non‐headgear vs. headgear) * positions (10 positions as listed above). Exposure: hours of match time.

  4. Outcome: count of suspected HIAs. Predictor variables: headgear (non‐headgear vs. headgear) * positional groupings (forwards/backs). Exposure: hours of match time.

  5. Outcome: count of suspected HIAs. Predictor variables: headgear (non‐headgear vs. headgear) * positions (10 positions as listed above). Exposure: count of tackle‐events.

  6. Outcome: count of suspected HIAs. Predictor variables: headgear (non‐headgear vs. headgear) * positional groupings (forwards/backs). Exposure: count of tackle‐events.

For each model, the p‐value of likelihood ratio Chi‐squared test is reported in the associated Tables (footnotes of Tables 2 and 3).

TABLE 2.

Comparison of rates and incidence rate ratios (including 95% confidence intervals) of tackling per 80 min in players. p‐values are from Poisson regression interaction term: headgear x position (“Prop” is the base/comparison position for the model using all 10 positions; “Backs” are the base/comparison for the model using the grouped version of positions [forwards vs. backs]). NA was for “All” positions as there is no comparison.

Positions Tackle‐event rate, non‐headgear (tackles per 80 min) Tackle‐event rate, headgear (tackles per 80 min) IRR headgear versus. non‐headgear p‐value
All 17.6 (17.5–17.7) 18.8 (18.5–19.2) 1.07 (1.05–1.09) NA
Forwards 19.4 (19.3–19.6) 19.2 (18.8–19.5) 0.99 (0.97–1.01) 0.208
Prop 17.0 (16.7–17.2) 15.9 (14.9–16.9) 0.94 (0.88–1.00) base
Hooker 19.6 (19.2–20.0) 19.2 (18.3–20.1) 0.98 (0.93–1.03) 0.318
Lock 19.4 (19.1–19.7) 18.9 (18.4–19.5) 0.98 (0.94–1.01) 0.313
Flanker 21.1 (20.8–21.4) 20.8 (20.0–21.7) 0.98 (0.94–1.03) 0.232
Eighth man 21.0 (20.6–21.4) 20.2 (19.2–21.3) 0.96 (0.91–1.02) 0.540
Backs 15.8 (15.6–15.9) 16.2 (15.3–17.1) 1.03 (0.97–1.09) base
Scrumhalf 14.3 (13.9–14.6) 14.2 (8.0–20.4) 0.99 (0.62–1.49) 0.809
Flyhalf 16.7 (16.3–17.1) 16.5 (15.2–17.9) 0.99 (0.91–1.08) 0.316
Centre 18.3 (18.0–18.5) 19.6 (17.1–22.0) 1.07 (0.94–1.21) 0.073
Wing 14.1 (13.8–14.3) 15.7 (12.8–18.7) 1.12 (0.92–1.35) 0.087
Fullback 14.7 (14.4–15.0) 13.4 (11.7–15.1) 0.91 (0.80–1.04) 0.692
Open in a new tab

Note: Overall Poisson model fit (p‐value of likelihood ratio Chi‐squared test) for all forward positions: < 0.001. Overall Poisson model fit (p‐value of likelihood ratio Chi‐squared test) for forwards versus backs: 0.013. Tackle event includes involvements in both tackling and ball‐carrying activities. Italics indicate a grouping of positions. For example, Forwards includes Prop, Hooker, Lock, Flanker, and Eightman; Backs includes Scrumhalf, Flyhalf, Center, Wing, and Fullback. All refers to all positions—that is, both the Forwards and Backs groupings.

Abbreviation: IRR, incidence rate ratio.

TABLE 3.

Comparison of rates and incidence rate ratios (including 95% confidence intervals) of suspected head injuries (a) per 1000 match hours, and (b) per 1000 tackle‐events. p‐values are from Poisson regression interaction term: headgear x position (“Prop” is the base/comparison position for the model using all 10 positions; “Backs” are the base/comparison for the model using the grouped version of positions [forwards vs. backs]). NA was for “All” positions as there is no comparison.

Suspected head injury incidence per 1000 h mean (95% CI) Suspected head injury propensity per 1000 tackle‐events ‐ mean (95% CI)
Positions Non‐headgear Headgear IRR headgear versus. non‐headgear p‐value Non‐headgear Headgear IRR headgear versus. non‐headgear p‐value
All 15.1 (12.6–18.0) 26.8 (15.9–38.0) 1.78 (1.11–2.70) NA 1.1 (1.0–1.3) 1.9 (1.1–3.0) 1.66 (1.04–2.52) NA
Forwards 17.0 (13.3–20.8) 28.9 (16.8–41.0) 1.70 (1.03–2.67) 0.134 1.2 (0.9–1.4) 2.0 (1.2–2.9) 1.72 (1.05–2.71) 0.781
Prop 13.5 (0–20.0) 25.2 (0–60.0) 1.87 (0.30–6.51) Base 1.1 (0–2.0) 2.1 (0–5.0) 1.99 (0.32–6.93) base
Hooker 14.2 (0–24.0) 53.4 (10.7–96.0) 3.75 (1.24–10.80) 0.901 1.0 (0.3–2.0) 3.7 (0.7–7.0) 3.83 (1.26–11.01) 0.477
Lock 24.7 (15.0–34.0) 30.4 (11.6–49.0) 1.23 (0.56–2.49) 0.055 1.7 (1.0–2.0) 2.1 (0.8–3.0) 1.26 (0.58–2.55) 0.588
Flanker 17.3 (0–25.0) 20.6 (0–44.0) 1.19 (0.28–3.45) 0.447 1.2 (0–2.0) 1.3 (0–3.0) 1.21 (0.29–3.50) 0.607
Eighth man 13.7 (0–23.0) 10.7 (0–32.0) 0.78 (0.04–4.26) 0.974 0.9 (0–1.0) 0.7 (0–2.0) 0.81 (0.04–4.42) 0.490
Backs 13.2 (0–16.5) 10.3 (0–30.6) 0.78 (0.04–3.54) Base 1.1 (0–1.4) 0.9 (0–2.5) 0.76 (0.04–3.45) base
Open in a new tab

Note: Overall Poisson model fit (p‐value of likelihood ratio chi‐squared test) for (a) for all forward positions: 0.170. Overall Poisson model fit (p‐value of likelihood ratio chi‐squared test) for (a) forwards versus backs: 0.025. Overall Poisson model fit (p‐value of Likelihood Ratio chi‐square test) for (b) for all forward positions: 0.159. Overall Poisson model fit (p‐value of Likelihood Ratio chi‐square test) for (b) forwards versus backs: 0.014. Italics indicate a grouping of positions. For example, Forwards includes Prop, Hooker, Lock, Flanker, and Eightman; Backs includes Scrumhalf, Flyhalf, Center, Wing, and Fullback. All refers to all positions—that is, both the Forwards and Backs groupings.

Abbreviation: IRR, incidence rate ratio.

When all 10 positional groupings were added to the model, “prop” was arbitrarily chosen as the base/reference position. Similarly, when the 10 positions groupings were compared within the binary grouping variables (forwards vs. backs), “backs” was arbitrarily chosen as the base/reference positional grouping.

There were five specific players who featured in seven matches for which no data were available from Opta. As it was confirmed by the video analysts that these players were indeed playing in the match in question, it was considered that these data were missing at random.

3. Results

The two seasons in 2018 and 2019 had 171 unique players in a total of 253 matches. The majority of players (91%, n = 155) did not wear headgear, 8% (n = 8) always wore headgear and the remaining 2 players (1%) wore headgear occasionally. These 253 matches resulted in 607,040 min of match play (approximately 10,117 h) logged against individual players by the video analysts (Table 1).

TABLE 1.

Tackle‐events (tackling and ball carrying), minutes of play and suspected head injury assessments (HIAs) in headgear and non‐headgear players.

Combined (headgear and non‐headgear) Non‐headgear Headgear
Positions Tackle‐events Minutes of play Suspected HIAs Tackle‐events Minutes of play Suspected HIAs Tackle‐events Minutes of play Suspected HIAs
All 134,275 607,040 163 122,171 555,631 140 12,104 51,409 23
Forwards 78,438 323,815 101 67,509 278,211 79 10,929 45,604 22
Prop 16,988 80,410 19 16,039 75,643 17 949 4767 2
Hooker 9876 40,493 14 8259 33,751 8 1617 6742 6
Lock 19,418 80,541 35 14,751 60,809 25 4667 19,732 10
Flanker 21,529 81,664 24 19,252 72,909 21 2277 8755 3
Eighthman 10,627 40,707 9 9208 35,099 8 1419 5608 1
Backs 55,837 283,225 62 54,662 277,420 61 1175 5805 1
Scrumhalf 7227 40,498 8 7207 40,385 8 20 113 0
Flyhalf 8485 40,699 12 7906 37,900 11 579 2799 1
Centre 18,595 81,290 18 18,355 80,309 18 240 981 0
Wing 14,143 80,396 17 14,036 79,852 17 107 544 0
Fullback 7387 40,342 7 7158 38,974 7 229 1368 0
Open in a new tab

Note: Italics indicate a grouping of positions. For example, Forwards includes Prop, Hooker, Lock, Flanker, and Eightman; Backs includes Scrumhalf, Flyhalf, Center, Wing, and Fullback. All refers to all positions—that is, both the Forwards and Backs groupings.

Abbreviation: HIA, head injury assessment.

Players not wearing headgear (non‐headgear) accounted for the majority of tackle‐events (122,171/134,275; 91%), match minutes (555,631/607,040; 92%) and suspected tackle‐event HIAs (140/163; 86%).

4. Comparison of Tackle and Suspected Head Injury Rates Between Players Wearing and Not Wearing Headgear

For all positions grouped together (Table 2, “all” positions), tackle‐events (tackling and ball‐carrying) per 80 min were significantly higher in players wearing headgear than not wearing headgear (IRR: 1.07, 95% CI: 1.05—1.09). None of the specific positional groupings (including “forwards” and “backs”) had significantly different tackle‐event rates per 80 min between players wearing and not wearing headgear. Centres (IRR: 1.07, 95% CI: 0.94—1.21) and Wings (IRR: 1.12, 95% CI: 0.92—1.35) wearing headgear had slightly higher tackle event rates than the same positional groups not wearing headgear, although not significant. Props wearing headgear had a slightly lower rate of tackle‐event involvement per match than Props not wearing headgear (IRR: 0.94, 95% CI: 0.88—1.00). However, the Poisson regression models that examined the interaction effect of headgear and positional groupings confirmed that there were no statistically significant differences in tackle‐event rates between specific positions (p‐values in Table 2).

Only one suspected HIA occurred in a back player wearing headgear (Table 1). Thus, the rate (per 1000 match hours) and propensity (per 1000 tackle‐events) of suspected HIAs is compared between players wearing and not wearing headgear using the positional grouping of backs and forwards and within specific forward positions only (Table 3).

Overall, both the suspected tackle‐event head injury incidence (per 1000 match hours, IRR: 1.78, 95% CI: 1.11—2.70) and propensity per 1000 tackles (IRR: 1.66, 95% CI: 1.04—2.52) were higher in players wearing headgear (Table 3). According to the unadjusted rate ratios, forwards wearing headgear had a higher suspected head injury incidence per 1000 h (IRR: 1.70, 95% CI: 1.03—2.67) and propensity per 1000 tackles (IRR: 1.72, 95% CI: 1.05—2.71). In the Poisson regression models, backs were not significantly different to forwards for suspected head injury incidence or propensity, however the back positional grouping only had one suspected head injury.

Although the rate ratios suggested that Hookers wearing headgear had a higher incidence (IRR: 3.75, 95% CI: 1.24—10.80) and propensity (IRR: 3.83, 95% CI: 1.26—11.01) of suspected head injuries than Hookers not wearing headgear, the Poisson regression model did not find this positional group to be significantly different to any other (incidence p‐value = 0.318, propensity p‐value = 0.284). Indeed, the Poisson regression models that examined the interaction effect of headgear and positional groupings on suspected head injury rates and propensities confirmed that no specific positional groupings (including forwards and backs) were statistically different from one another (p‐values in Table 3).

5. Discussion

This study assessed tackle‐event head injury risk in rugby players, wearing and not wearing soft‐padded head gear, as a function of their exposure to tackle‐events (propensity), in addition to time played (incidence). This approach enables assessment of the potential effects of soft‐padded headgear on head injury risk accounting for positional differences in tackle‐event involvements that might expose players to more risk as a result of more frequent contacts during play.

We found that players who wear headgear are involved in a higher rate of tackle‐events than players not wearing headgear (Table 2), and that players wearing headgear have a higher propensity, per 1000 tackle involvements, to sustain head injuries (Table 3). There are three possible, not mutually exclusive, explanations for this elevated risk in those wearing headgear: (1) players wearing headgear have a recent history of concussion, making them predisposed to another concussion (Stokes et al. 2021); or (2) the higher rate of tackle‐event involvements in players wearing headgear leads to a higher absolute count of tackle‐events per match, increasing their risk of a head injury within that match or (3) players wearing headgear are also more likely to have inadequate technique, as this has previously been shown to be associated with concussions (Hendricks et al. 2015).

However, the increase in exposure is not universal among positions. Hookers, Flankers and Eighth Men, who have the highest tackle‐event rates by position, had similar tackle‐event rates with and without headgear. The two positional groupings that were closest to showing an effect of headgear on increased tackle‐event rates were the Centres and Wings (Table 2). However, due to an absence of HIAs in these positional groupings while wearing headgear, we could not assess whether the greater tackle‐event rates contributed to differences in head injury incidence. In contrast to Centres and Wings, Props tended to show the opposite effect: a lower tackle‐event involvement per match in players wearing headgear (Table 2). Thus, although players wearing headgear did, overall, have a higher tackle‐event rate per 80 min than players not wearing headgear (IRR 1.07, 95% CI: 1.05—1.09, Table 2), there was no specific positional grouping that had a higher tackle‐event rate when wearing headgear, making it difficult to speculate on the reason for this finding when all positions are grouped together (“All” positions in Table 2).

These findings, at an individual level, are supportive of similar studies in rugby that have found no evidence of a protective effect at a group level (Marshall et al. 2005; McIntosh et al. 2009; Stokes et al. 2021), and advance those findings by providing evidence that players wearing headgear may have elevated tackle‐event head injury risk, particularly among forwards. As we and other authors (McIntosh et al. 2009) have described, the absence of any protective effect may be the result of the low impact attenuation efficacy of the headgear that is permitted by World Rugby's headgear regulation. A 2009 randomised controlled trial in youth players compared concussion risk in regulation headgear against modified headgear that was thicker (and said to have superior force attenuation properties) than World Rugby regulations permitted at the time (McIntosh et al. 2009). Although the modified headgear group did not have a lower concussion rate than regulation headgear, these authors acknowledge that there was low compliance with wearing the modified headgear. Indeed, the compliance was so poor that it may have hidden any actual effect of the modified headgear. It is important to note that World Rugby's recent headgear trial (https://www.world.rugby/the‐game/facilities‐equipment/equipment/law‐4‐headgear‐trial) may result in a greater variety of headgear devices, with potentially larger protective effects, in time. However, it should also be noted that force dampening is only one component of head injury risk. Previously identified concussion risk factors such as prior concussion history (Stokes et al. 2021) and specific mechanisms (such as those that involve large rotational forces (Tierney 2024)) might reduce the effectiveness of force‐attenuating headgear.

This observational cohort study had a number of limitations. We assumed that headgear (and non‐headgear) wearing remained the same for the duration of the player's time on the field. There is a chance that this could have misclassified some of the playing time and head injury assessments. Despite our best estimates, one would need a far larger sample of matches to be able to examine the effect of headgear on head injury rates and propensity in backs, where infrequent headgear use and low HIA incidence combine to produce very few cases for analysis. This low use of headgear has been described in other professional male cohorts (Stokes et al. 2021). Even within forwards, the current study might be underpowered due to the low proportion of players wearing headgear, compounded by the relatively low number of players who sustain head injuries. In backs, the rarity of head injuries in players wearing headgear resulted in only one HIA case, negating the possibility for further analysis. This also accounts for the wide confidence intervals for the incidence rate ratio for specific positions (Table 3). However, given the low proportion of overall match time played by backs wearing headgear (5805/283,225; 2%), examining this effect would require a very large cohort of matches that poses other epidemiological study challenges such as temporal differences and homogeneity of participants. Although we feel that these findings are generalisable to other elite men rugby cohorts, we are not able to say the same of the women's and/or youth game. Another limitation is that we did not describe or account for the specific soft‐padded headgear which was worn by the players during these 2018 and 2019 rugby seasons. Whereas it is assumed all were compliant with World Rugby regulation, there may be differences in impact attenuation capability between those worn in this study, and those commercially available. Furthermore, given the focus within rugby to reduce head acceleration events, as well as concussions, the ability of soft‐padded headgear to reduce both the magnitude and frequency of head acceleration events was not investigated. Our Poisson regression models were not well fitted (overall Poisson model fit p‐value > 0.05) for suspected head injury rate and propensities that were compared between forwards and backs. However, trying alternative Poisson regression models (negative binomial and zero‐inflated) did not improve this model fit and this is probably further indication of the inability to compare the effect of backs given that only one back player wearing headgear had a suspected head injury. It was also assumed that head injury assessments were a proxy for head injury risk throughout this study, and recognise that these two variables might not be directly related. Finally, as mentioned above, we acknowledge that we could not include important confounders of concussion such as previous history of concussion or recent concussion or total years of exposure to contact sport, as has been described in other studies (Stokes et al. 2021).

Nonetheless, the major strength of this observational study is that it extends our understanding of soft‐padded headgear wearing and head injury risk in rugby by comparing propensity (head injury per tackle involvement) and incidence at an individual level rather than at a group level. This is the first paper, to the authors’ knowledge, that has assessed head injury risk at this level of detail.

6. Conclusion

Soft‐padded headgear use was not only found to not be protective of head injury risk in this elite men's rugby cohort, there was also some evidence that it was associated with increased risk of suspected head injury. Whereas the lack of protective effect is consistent with previous findings (Marshall et al. 2005; McIntosh et al. 2009; Stokes et al. 2021), our study expands on those findings, since we examine's head injury risk at an individual level. It is important to note that the headgear under study was highly regulated and thus could be assumed to be homogenous at the time of study. Recent trials by World Rugby might increase the variety and potentially, effectiveness, of approved headgears in time. Future studies should continue to re‐examine the effect of headgear on head injuries, accounting for collision involvements, tackle technique and previous concussion or head acceleration event exposure, where possible.

Funding

The authors have nothing to report.

Ethics Statement

Ethics approval was received from [removed for confidentiality during peer‐review] Stellenbosch University (approval number U21/05/120).

Consent

This study involved the collection of secondary analyses and thus no consent was required.

Conflicts of Interest

J.B. has received researching funding, travel funding and part of his university salary from World Rugby. S.H. consults to World Rugby. R.T. is a consultant to World Rugby. The rest of the authors declared no conflicts of interest.

Acknowledgements

Thank you to the late Professor Wilbur Kraak for all of his assistance with this project over many years and with many of his students. Wilbur would have been an author on this paper, but he sadly passed away suddenly in July 2024.

Brown, James , Douglas Marc, Hester Ben, et al. 2026. “Wearing Regulation Soft‐Padded Headgear Does Not Reduce the Risk of Head Injuries in Professional Men's Rugby Players: An Observational Cohort Study,” European Journal of Sport Science: e70105. 10.1002/ejsc.70105.

Data Availability Statement

De‐identified data is available upon request.

References

  1. Al Attar, W. S. A. , Mahmoud H., Alfadel A., and Faude O.. 2024. “Does Headgear Prevent Sport‐Related Concussion? A Systematic Review and Meta‐Analysis of Randomized Controlled Trials Including 6311 Players and 173,383 Exposure Hours.” Sports Health 16, no. 3: 473–480. 10.1177/19417381231174461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bahr, R. , Clarsen B., Derman W., et al. 2020. “International Olympic Committee Consensus Statement: Methods for Recording and Reporting of Epidemiological Data on Injury and Illness in Sports 2020 (Including the STROBE Extension for Sports Injury and Illness Surveillance (STROBE‐SIIS)).” Orthopaedic Journal of Sports Medicine 8, no. 2: 1–33. 10.1177/2325967120902908. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Eliason, P. H. , Galarneau J.‐M., Kolstad A. T., et al. 2023. “Prevention Strategies and Modifiable Risk Factors for Sport‐Related Concussions and Head Impacts: A Systematic Review and Meta‐Analysis.” British Journal of Sports Medicine 57, no. 12: 749–761. 10.1136/bjsports-2022-106656. [DOI] [PubMed] [Google Scholar]
  4. Hendricks, S. , O Connor S., Lambert M., et al. 2015. “Contact Technique and Concussions in the South African Under‐18 Coca‐Cola Craven Week Rugby Tournament.” European Journal of Sport Science 15, no. 6: 557–564. 10.1080/17461391.2015.1046192. [DOI] [PubMed] [Google Scholar]
  5. Hendricks, S. , Till K., Den Hollander S., et al. 2020. “Consensus on a Video Analysis Framework of Descriptors and Definitions by the Rugby Union Video Analysis Consensus Group.” British Journal of Sports Medicine 54, no. 10: 566–572. 10.1136/bjsports-2019-101293. [DOI] [PubMed] [Google Scholar]
  6. Hongyou, L. , Hopkins W. G., Gomez M. M., and Molinuevo J. S.. 2015. “Inter‐Operator Reliability of Live Football Match Statistics From OPTA Sportsdata Inter‐Operator Reliability of Live Football Match Statistics From OPTA Sportsdata.” Journal of Sports Performance 13, no. August: 803–821. 10.1080/24748668.2013.11868690. [DOI] [Google Scholar]
  7. Kirkwood, B. , and Sterne J.. 2003. Essential Medical Statistics (Second). Blackwell Science Ltd. [Google Scholar]
  8. Lindsay, A. , Draper N., Lewis J., Gieseg S. P., and Gill N.. 2015. “Positional Demands of Professional Rugby.” European Journal of Sport Science 15, no. 6: 480–487. 10.1080/17461391.2015.1025858. [DOI] [PubMed] [Google Scholar]
  9. Marshall, S. W. , Loomis D. P., Waller A. E., et al. 2005. “Evaluation of Protective Equipment for Prevention of Injuries in Rugby Union.” International Journal of Epidemiology 34, no. 1: 113–118. 10.1093/ije/dyh346. [DOI] [PubMed] [Google Scholar]
  10. McIntosh, A. S. , McCrory P., Finch C. F., Best J. P., Chalmers D. J., and Wolfe R.. 2009. “Does Padded Headgear Prevent Head Injury in Rugby Union Football?” Medicine & Science in Sports and Exercise 41, no. 2: 306–313. 10.1249/MSS.0b013e3181864bee. [DOI] [PubMed] [Google Scholar]
  11. Quarrie, K. L. , Hopkins W. G., Anthony M. J., and Gill N. D.. 2013. “Positional Demands of International Rugby Union: Evaluation of Player Actions and Movements.” Journal of Science and Medicine in Sport 16, no. 4: 353–359. 10.1016/j.jsams.2012.08.005. [DOI] [PubMed] [Google Scholar]
  12. Raftery, M. , Kemp S., Patricios J., Makdissi M., and Decq P.. 2016. “It is Time to Give Concussion an Operational Definition: A 3‐Step Process to Diagnose (Or Rule Out) Concussion Within 48 H of Injury: World Rugby Guideline.” British Journal of Sports Medicine 50, no. 11: 642–643. 10.1136/bjsports-2016-095959. [DOI] [PubMed] [Google Scholar]
  13. Roe, G. , Sawczuck T., Collins N., et al. 2025. “Spot the Difference? Contact Event Frequency During > 30,000 Women s and Men s Rugby Union Player Matches Across Top Domestic and International Competitions.” European Journal of Sport Science 25, no. 5: 1–9. 10.1002/ejsc.12307. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Rowson, S. , Duma S. M., Greenwald R. M., et al. 2014. “Can Helmet Design Reduce the Risk of Concussion in Football?” Journal of Neurosurgery 120: 1–4. 10.3171/2014.1.JNS13916. [DOI] [PubMed] [Google Scholar]
  15. Scott, G. A. , Edwards O., Bezodis N. E., et al. 2023. “Classifying Winning Performances in International Womens Rugby Union.” International Journal of Sports Physiology and Performance 18, no. 9: 1072–1078. 10.1123/ijspp.2023-0086. [DOI] [PubMed] [Google Scholar]
  16. Stokes, K. A. , Cross M., Williams S., et al. 2021. “Padded Headgear Does Not Reduce the Incidence of Match Concussions in Professional Mens Rugby Union: A Case‐Control Study of 417 Cases.” International Journal of Sports Medicine 42, no. 10: 930–935. 10.1055/a-1345-9163. [DOI] [PubMed] [Google Scholar]
  17. Tierney, G. 2024. “Concussion Biomechanics, Head Acceleration Exposure and Brain Injury Criteria in Sport: A Review.” Sports Biomechanics 23, no. 11: 1888–1916. 10.1080/14763141.2021.2016929. [DOI] [PubMed] [Google Scholar]
  18. Tooby, J. , Woodward J., Tucker R., et al. 2024. “Instrumented Mouthguards in Elite‐Level Mens and Womens Rugby Union: The Incidence and Propensity of Head Acceleration Events in Matches.” Sports Medicine 54, no. 5: 1327–1338. 10.1007/s40279-023-01953-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Tucker, R. , Raftery M., Kemp S., et al. 2017. “Risk Factors for Head Injury Events in Professional Rugby Union: A Video Analysis of 464 Head Injury Events to Inform Proposed Injury Prevention Strategies.” British Journal of Sports Medicine 51, no. 15: 1152–1157. 10.1136/bjsports-2017-097895. [DOI] [PubMed] [Google Scholar]
  20. Williams, S. , Robertson C., Starling L., et al. 2021. “Injuries in Elite Mens Rugby Union: An Updated (2012–2020) Meta‐Analysis of 11,620 Match and Training Injuries.” Sports Medicine 52, no. 5: 0123456789–1140. 10.1007/s40279-021-01603-w. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

De‐identified data is available upon request.

Articles from European Journal of Sport Science are provided here courtesy of Wiley

RESOURCES