Motorcyclists and other powered two-and three-wheeler (PTW) riders are among the most vulnerable road users (VRUs), making over 30% of the 1.19 million annual road traffic deaths globally (WHO, 2023). Additionally, the number of motorcycles and other PTWs nearly tripled in the decade to 2021, with an associated increase in fatality numbers. Helmets can provide protection to the motorcyclist in collisions, reducing serious head injury as well as fatality (Organization, 2023; WHO, 2023; Yasin et al., 2022). For this reason, the World Health Organization (WHO) has identified helmet use as a key legislation area to reduce the global casualty burden (Organization, 2023; WHO, 2023; Yasin et al., 2022).

Many global regions require that helmets meet a certain standard (shown in blue in Figure 1). Work to mandate standards in other countries is ongoing, which will improve helmets (WHO, 2023). Helmet safety improvements can be made by an improved understanding of head impact characteristics in real-world crashes (such as impact location and its relationship to specific head injury patterns). The profile of injuries associated with certain impact regions, including the face, is under-characterised in the context of informing standards. As such, the facial region (comprising the upper face, visor and chinbar) undergoes minimal impact testing in current standards, as shown in Table 1. The exceptions are Snell M2025 (and similarly M2020), BSI 6658 and UN ECE R22.05 & R22.06 which include chinbar impacts (Smith, 2025). ECE R22.06 introduced three upper face impact points, but the visor region remains entirely untested across all major standards (Figure 2). The January 2024 implementation of ECE 22.06 saw an increase from one to nine pre-defined impact points on the crown (BP, BXPL, BXPR, P, RP, RXPL, RXPR, XPL, XPR), two to six on the side (non-facial: BXR, BXL, RXL, RXR, XL and XR), one to three on the upper face (BXL, B and BXR) with a single chinbar impact maintained (though at an increased speed of 6m/s in ECE 22.06).

Only a small number of previous studies have reported helmet damage or impact location together with injury outcomes for motorcyclists (Dowdell et al., 1988; Kraus et al., 2003; Meng et al., 2023; Rice et al., 2016; Whyte et al., 2016). One study investigating motorcyclist head injury in 1980s identified that over 50% of severe impacts occurred to the facial region (Dowdell et al., 1988). Notably, impacts to the frontal region, which often was not protected by 1980s helmets, frequently led to frontal skull fractures and basilar skull fracture (BSF). The comprehensive pan-European COST 327 study analysed data between July 1996 and June 1998, included 226 motorcyclists of whom two thirds sustained a head injury (Chinn et al., 2001; Chinn et al., 1999). The most frequent intracranial injuries were subdural hematomas (22.4%, 47/226) and subarachnoid haemorrhage (25.2%, 53/226). Another study, included an analysis of 47 helmeted riders who were fatally injured in the Adelaide metropolitan region between 1983 and 1994. This found 43% (20/47) of the impacts were to the facial region, leading to BSF in 13 cases out of 20 (Whyte et al., 2016). Case studies have also proposed facial impact and the chinbar’s role in transmitting force from impact as a mechanism for basilar skull fracture (Cooter et al., 1988; Huelke et al., 1988).

The studies reviewed above were conducted using data from more than two decades ago. Since then, factors affecting motorcyclist injury, including improvements to opponent vehicle crashworthiness such as front-end vehicle geometry and stiffness, helmet type and helmet technology, have improved motorcyclist and other VRU protection (Baker et al., 2024a; Cini et al., 2014; ECE, 1999; Lee et al., 2014; Perticone et al., 2024; Xiao et al., 2020). These developments motivate a contemporary assessment of motorcyclist head injury. In addition, facial impacts and associated head injuries are minimally reported in these previous studies, with the exception of some references to basilar skull fracture mechanism (Huelke et al., 1988). The most detailed study that reports injuries caused by facial impacts is limited to fatal cases only (Whyte et al., 2016). This limits our ability to understand survivable injuries and protection improvements in the wider population of injured motorcyclists. There is a related body of work assessing whether helmet type has a role to play in facial injury outcome. Motorcyclists wearing full-face helmets have been found to have reduced head injury prevalences, including lower facial fracture rates, and were three times less likely to sustain facial injury compared to those with another type (Aires et al., 2022; Chaichan et al., 2020; Kraus et al., 2003; Wu et al., 2019). These studies however did not assess the interaction between helmet type, injury outcome and impact location, limiting our ability to interpret the findings in the context of informing helmet assessment methods.

To address these gaps, this study characterises helmet impact location and head injury patterns, with a specific focus on facial impacts and their relevance to current test standards. We use contemporary, in‑depth investigation data with detailed clinical coding from the UK Road Accident In-Depth Studies (RAIDS) database between 2013 and 2025 and aim to:

An overview of the methods is provided in Figure 3. We describe motorcycle collisions from the RAIDS database, including demographics, collision type and helmet use. Distinct types of head injury are identified using a combination of Abbreviated Injury Scale (AIS) and free-text search method (Baker et al., 2022; Gennarelli, 1985; Gennarelli & Wodzin, 2006; Medicine, 2016). Concurrently, we identify a subset of the cohort with categorised primary head impact locations as inferred by reported helmet damage. Finally, the impact location and clinical head injury outcomes are combined to analyse their interaction to better understand implications for future motorcycle helmet testing methods.

This study provides the first analysis of post-2010 motorcyclist helmet damage data, providing researchers and policy makers with motorcyclist head impact locations and their implications for injury prevention, potentially informing the development of more effective helmet test methods. Our analysis of 353 motorcycle collisions which occurred from 2013-2025 showed that a third of motorcyclists with confirmed head impacts sustained a head injury, nearly 20% sustained a moderate-severe TBI and 12% sustained a skull fracture. These results indicate areas for helmet improvement. Importantly, almost half of the primary impacts to the helmet were in the facial region among cases with determinable primary helmet impact location. Primary facial impacts led to statistically similar injury patterns and severities as non-facial impacts.

The most striking finding of this study is that 58.7% of all helmet damage, and 49.6% of primary impacts, occurred to the facial region. This is particularly notable given that the facial region is the least tested area in current helmet standards: no mandated standard includes visor impact testing, and chinbar testing is limited to a single central impact point in ECE R22.05/06, BSI 6658 and Snell (Table 1). ECE R22.06 introduced three upper face test points, representing a step forward, but the visor region and lateral facial impacts remain entirely unaddressed. Our findings therefore provide direct epidemiological justification for expanding facial impact coverage in future standards and ratings testing.

Half of all primary impacts were to the facial region, even though this region comprises less than half of the helmet’s coverage (Figure 5). Our collision configuration analysis showed that motorcyclists often collide head-on with another vehicle that is turning or crossing. This results in impacts between the rider’s front and the vehicle’s side, providing an explanation for the high prevalence of facial and left-sided impacts. Direct comparison with previous work is limited, as COST327 reports all damage (as opposed to primary and secondary damage) and does not separate front and rear left or right, grouping all damage together (Chinn et al., 2001). Existing standards minimally test the facial region of helmets, despite the high exposure of this region in our cohort of confirmed head impact locations.

Severe head and brain injury occurs in both primary facial and non-facial impacts. There was not a statistically significant difference between the head injury rates in primary facial vs non-facial impact locations, despite facial injuries having higher overall percentages, reflecting the limited size of the subset with determinable impact locations. Our interpretation of the results therefore focuses primarily on the high absolute burden of head injury in facial impact cases and the qualitative pattern of impacts (e.g. predominance of upper face and chinbar). Although our findings are derived from a subsample of collisions with sufficiently documented helmet damage, the high proportion of facial impacts in this group suggests that facial regions are frequently involved in real‑world crashes and merit consideration in helmet test protocols.

The only standards that include facial impacts are ECE R22 and Snell, which prescribe a linear impact to the centre of the chinbar (ECE, 2020; ECE, 1999; Snell, 2020). From ECE 22.05 to ECE 22.06, 12 additional impact points were included to select from, introducing variation in the impact conditions that are considered to be realistic in the context of real-world crash variability. However, most of the 12 impact points are on the non-facial region, excluding three upper face points: one left, one right, one central (Figure 2). ECE R22.06 also introduced oblique impact testing to assess helmet performance under rotational loading conditions, representing an important advance over the purely linear assessments of ECE R22.05. Oblique impact testing is also included in testing off-road and circuit racing motorcycle helmets from 2026 onwards, with peak rotational acceleration and the brain injury criterion (BrIC) used in its rotational assessment (Federation Internationale de Motocyclisme, 2024). Rotational attenuation technologies, such as MIPS, are now present in motorcycle helmets as a commercially available approach to reducing rotational energy transfer to the head. Notably, however, oblique impact and rotational assessment have not yet been extended to middle and lower facial impact regions in any current standard, meaning that even where rotational loading is assessed, it is assessed only for the non-facial helmet regions that are already the most comprehensively tested. This study clearly identifies a gap in ECE 22.06 test regions across both impact location and loading condition. Hence, our results support the development of facial impact test methods to enable the assessment of helmet protection under these most prevalent impacts.

Although there was no statistically significant difference in the proportion of pathologies between the facial and non-facial impacts, we observed a trend toward a higher incidence in the facial impact cohort with numerically higher rates in the facial impact cohort. Overall, injury patterns and severity were similar for facial and non-facial impacts, both including severe outcomes. Notable injury patterns included skull fractures (particularly basilar skull fractures), subarachnoid haemorrhage, focal brain injuries and subdural haematomas. These injury patterns are consistent with previous work. One study reported that impacts to the front region frequently led to skull and basilar skull fractures, noting that these impacts were often located outside of the coverage area of helmets from the 1980s when the study was conducted (Dowdell et al., 1988). Another study of 47 fatal motorcycle collisions from 1983-1994 found that 43% of impacts were to the facial region, leading to basilar skull fractures in 66% of cases (Whyte et al., 2016). The high prevalence of basilar skull fracture was expected as BSF is associated with high fatality rates (Fujiwara et al., 2021). In contrast to these studies, our data is representative of current motorcycle collisions with a range of outcomes from no injury to fatality. Of those with known primary impact locations, we found that approximately 85% (6/7) of motorcyclists sustaining subdural haematoma experienced primarily facial impacts, along with approximately three quarters (11/15) of motorcyclists sustaining subarachnoid haemorrhage. Approximately two thirds of motorcyclists who sustained both skull fracture (11/16) and basilar skull fracture (6/9) suffered primary facial impacts. These results indicate most prevalent pathologies that should be protected against by improving helmets, and this can be driven by adopting suitable injury criteria in helmet test methods. Current helmet standards for chin bar impacts pass or fail a helmet only based on peak resultant linear acceleration and HIC (Head Injury Criterion), hence they may not address all relevant pathologies identified in this study.

Among the three structurally distinct facial helmet regions, i.e. chinbar (lower structural element of the helmet), visor (the transparent protective screen) and upper face (the helmet shell overlying the forehead and orbital area), the visor was the facial sub-region with the most primary impacts (36.5% of primary facial impacts, Table 2) but associated with the lowest average injury burden (1.35 injury categories per rider, falling to 0.87 for MAIS3+ injuries). This suggests that visor impacts, while common, may be less mechanically severe on average, possibly because the visor is a relatively large, deformable structure that distributes impact energy and survival bias may occur. Additionally, a 2023 study found that the visor redirects mid-face impact force to the forehead and lower face via the helmet’s upper rim and chinbar, resulting in less head rotation (Meng et al., 2023). This mechanism may explain the lower average injury burden. In contrast, chinbar and upper facial impacts, were associated with substantially higher injury burdens (2.67 and 3.27 injury categories per rider respectively, 1.73 and 2.47 for MAIS3+ as shown in Appendix Table 7) and had a relatively higher prevalence in the head injured cohort (Figure 9). The chinbar and upper facial region were noted by COST327 as frequently damaged locations (both around 15% of cases for all damage, compared to 16-18% in our study) (Chinn et al., 2001). In the more severe cases, it is possible that visor damage is underrepresented if the visor came off and is missed during on-scene data collection. Notably, primary facial impacts that were not specified by region had the highest injury burden, and were associated with 5.90 average injuries per rider. It is possible investigators were not able to examine the helmets as closely when the rider required greater medical attention, often resulting in text-based descriptions with no photographs which did not allow us to specify the facial damage region. Although we have a limited number of cases with identified facial impact regions, our results suggest that the upper facial region and chinbar are likely to be associated with higher numbers of head injuries, and as such they should have increased priority for future helmet design and assessment.

The rate of head injuries increased from 33% in all motorcyclists to 83% among the fatal cases, supporting previous findings that increased head injury prevalence is associated with an increase in overall severity. When comparing the RAIDS cohort of 60 fatal motorcycle collisions with the historic study of 47 fatal motorcycle collisions in Australia between 1983-1994, we found similar prevalences of subdural haemorrhage (SDH, 23%, vs our 20%, 12/60) and skull vault or facial fracture (40% vs our 37%, 22/60). However, BSF was significantly lower in our fatal cohort (our 25%, 15/60, vs 66%; χ²=18.0, p<0.0001). Taking fatal cases with confirmed facial impact, Whyte et al. (2016) found that 20% of riders sustained SDH compared to 43% of our cohort (not statistically significant, Yates corrected χ²=1.12, p=0.290), similar rates of skull vault fracture (35% vs our 43%), but higher BSF prevalence of 65% vs our 36% (not statistically significant, χ²=1.12, p=0.290). COST327 (1996–1998) reported 46 basilar skull fractures among 409 head injuries across 226 motorcyclists. As rider-level BSF prevalence was not reported, we could not directly compare our results with this more representative population. The lower rate of BSF in our RAIDS fatal injury cohort may be related to lower rates of open face helmets (8% vs 17%, χ²=1.86, p=0.172), developments in helmet and vehicle technology between the 1990s and 2010s or geographical differences in collisions and their severity between Australia and the UK. Twenty five years on from the data analysed by Whyte et al. (2016), we continue to see serious pathologies, including in riders wearing full-faced helmets that remained on for the full duration of the collision. This demonstrates the continued need for innovation in helmet design and testing.

The rate of head injury among motorcyclists with confirmed head impacts has fallen from approximately two thirds (COST327, 1996-1998) to one third in our cohort, consistent with improvements in helmet design, testing standards and vehicle front-end geometry over the same period (Chinn et al., 2001). This improvement may in part be explained by differences in collision severity in the cohorts; COST327 is a multicentre study with different sampling and inclusion criteria. However, as helmets were worn in both cohorts, it is likely that motorcyclist safety has improved since the 1990s, through helmet design, testing and other influential interventions such as speed limit and vehicle front end design (Perticone et al., 2024; Vigilante Jr, 2005; Xiao et al., 2020). As severe outcomes still remain in helmeted riders whose helmets remain on, this emphasises the continued need for improvement, especially to minimally tested regions of the helmet, such as the facial region. Having a clear understanding of both impact location and the types of injuries sustained has profound implications for designing representative helmet testing procedures and identifying how injury risk curves should be generated and implemented, as is demonstrated in previous literature (Baker et al., 2023; Baker et al., 2024b). Here, we extend what is available in literature beyond fatal-only cohorts from 25+ years ago (1980-90s data) by providing additional evidence from 2013-2025 data for all outcomes.

Despite the high quality and detail of RAIDS data, there are several challenges which resulted in the following choices being made. The travel speed values are determined by expert investigators based on all available information, and an assessment of confidence is included. To maximise the data sample size, the included speeds have all confidence assessment levels (low, medium and high), where low confidence is more likely to be estimated from the speed limit value itself. For the impact locations, after analysing both primary and all helmet damage, we chose to focus only on the subset of primary impacts in an effort to exclude non-major impacts that have a low likelihood of being causal of head injuries and inform helmet design and assessment. To mitigate the risk of bias in the assessment of primary impact locations, two researchers separately assessed the full cohort to create the confirmed head impact sub-cohort. This sub-cohort may represent better documentation and therefore be more severe (non-representative within RAIDS). Validation performed assessed this risk as low because the prevalence of head injuries was the same (33%) across both the full and confirmed head impact cohorts. It was possible for us to include 11 riders whose helmets came off (8.8% of the confirmed impact cohort), as investigators were able to confirm that the helmet remained on for the primary impact (and in some cases also the secondary impact). Generally, we found that RAIDS listed the collision with the impact vehicle as the cause for the most severe head injuries, however the confidence for this was not always reported to be “high” and ground impact can additionally cause injury.

Injury coding in RAIDS used AIS 2005 Update 2008 or AIS 2015 depending on the coding period, with one version applied consistently per case. While AIS updates can introduce new injury codes reflecting advances in diagnostic capability, a comparison of AIS versions by Loftis et al. (2018) found that the majority of head and brain injury codes relevant to this study, including skull fracture, intracranial haemorrhage and diffuse axonal injury, are consistent across these versions, and that version differences are unlikely to materially affect the injury distributions presented here (Loftis et al., 2018). Nevertheless, readers should be aware that a small number of cases may have been coded under different versions, and that AIS updates may have introduced additional specificity in some injury categories over the study period.

RAIDS is a severe subset of British collisions, hence there is a greater prevalence of more severe outcomes (Table 3) compared to a fully representative sample of motorcycle collisions across Great Britain or compared to other in-depth datasets adopting different sampling approaches. In addition, despite 353 motorcycle riders captured by RAIDS, it was only possible to confirm primary head impacts in approximately a third of the cases due to unavailability of helmet damage descriptions or helmet photographs. This limited our ability to fully utilise the RAIDS data to investigate the relationship between helmet impact location and head injury beyond descriptive analysis. In future, in line with the 2026 Road Safety Strategy for the UK, better collision-healthcare data connectivity and improved awareness among forensic collision investigators, both within the police and in hospitals, about the importance of helmet documentation could vastly improve the future availability and quality of impact location data. These efforts should underpin prospectively designed datasets will be required to more definitively quantify any excess risk associated with facial impacts.

Our analysis focused on head and facial injuries, meaning that cervical spine injuries and their relationship to helmet mass and impact kinematics were not assessed in this study. Helmet mass (which is not directly recorded within RAIDS) has been proposed as a contributory factor in ring-type basilar skull fractures via traction mechanisms, and in cervical spine loading during impact (Gibson, 2007; Konrad et al., 1996). These represent important areas for future investigation, particularly as helmet designs evolve to incorporate additional protective technologies that may affect helmet mass.

This epidemiological study does not include direct assessment of impact direction or resulting head and brain kinematics, which is a recognised limitation of retrospective in-depth collision data without computational or physical impact reconstruction. Future studies, particularly through accident reconstruction, should seek to capture impact angle and directionality to enable more complete biomechanical characterisation of facial impacts, and to inform the most appropriate facial impact testing methods to be incorporated into future standards revisions and testing protocols, informed by the underlying epidemiology presented in this study.

In summary, this study provides the first contemporary insight of 21^st century motorcycle head impact and injury outcomes at a larger scale than was previously available. Despite the good protection of helmets and a reduction in head injury rates compared to earlier reported cohorts, severe head injuries remain present. More importantly, nearly half of all primary head impacts were to the facial region, resulting in a range of injury patterns and severities similar to non-facial impact, despite facial impacts currently being minimally assessed by standards. New test methods should be developed to evaluate helmets under impacts to the facial region and assess their protection against skull fracture (vault and basilar), focal injury (which includes brain contusion, laceration and haemorrhagic contusion within the brain tissue) and bleeding pathologies (subarachnoid and subdural haemorrhages).

Claire Baker: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Zhuoxiaoyue Wang: Data curation, Formal analysis, Investigation, Validation, Writing – review & editing. Lucas Low: Investigation, Validation, Writing – review & editing. Emily Ashworth: Investigation, Validation, Writing – review & editing. Mark H. Wilson: Methodology, Writing – review & editing. David J. Sharp: Funding acquisition, Resources, Supervision, Writing – review & editing. Nils Lubbe: Conceptualization, Funding acquisition, Resources, Writing – review & editing. Shiyang Meng: Conceptualization, Funding acquisition, Project administration, Resources, Writing – review & editing. Mazdak Ghajari: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing.

We would like to thank the UK Department for Transport for providing free of charge access to this database.

Shiyang Meng and Nils Lubbe work at Autoliv (www.autoliv.com). Autoliv designs, develops and manufactures world-leading passive safety systems for the automotive industry, as well as mobility safety solutions. Results from this study may impact how Autoliv chooses to develop its products.

During the preparation of this study the author(s) used ChatGPT 5.1 and Claude 4.6 to edit and improve readability of this study. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.

No dissemination of this manuscript has occurred. Preliminary results from this work were presented as a short communication at the International Research Council on the Biomechanics of Injury (IRCOBI) conference in Vilnius, 2025.

RAIDS collection and use require very stringent ethical approvals and data security processes. These include approved applications for both a Confidentiality Advisory Group and the Research Ethics Committee and required the completion of a Data Security Protection Toolkit for ethical approval. To collect anonymized injury data, RAIDS has an agreement under Section 251. Section 251 of the NHS Act 2006 and the Regulations enable the common law of duty of confidentiality to be temporarily lifted so that confidential patient information can be transferred to an applicant without the discloser being in breach of the common law duty of confidentiality. Therefore, RAIDS does not seek permission from those who are injured. If a person would like their information removed, they are able to request this in writing. Approval to use the database for specific projects is granted by the Department for Transport. The approval includes all tasks that have been conducted for this study and all outputs, including this submitted manuscript, are checked thoroughly for anonymity and to ensure all protocols have been correctly followed prior to dissemination. Additionally, institutional approval from Imperial’s Research Governance and Integrity Team was granted for this study.

We would like to acknowledge funding from Autoliv and TBI-REPORTER which enabled this research to be conducted. Mazdak Ghajari acknowledges the support of the Royal Academy of Engineering under the Senior Research Fellowship scheme.

RAIDS data is available for free with access requests approved by the UK Department for Transport (RAIDS@dft.gov.uk).

Analysis code will be made available on request, and is expected to be shared via the TBI-REPORTER platform under RAIDS research tools.

Handling editor: Aliaksei Laureshyn, Lund University, Sweden.

Reviewers: Ciaran Simms, Trinity College Dublin, Ireland; Ulf Björnstig, Umeå University, Sweden.

Submitted: 10 March 2026; Accepted: 17 June 2026; Published: 12 August 2026.