Professional horse racing is a high-risk and dangerous sport with a high incidence of falls and injuries. While falls in horse racing are considered somewhat inevitable and carry an inherent occupational risk, little is known about the actual mechanisms of jockey injuries. Establishing injury aetiology and mechanism is a fundamental step in informing the design and implementation of future injury prevention strategies. Despite the availability of horse racing video footage, the use of video analysis to examine injury mechanisms is an underused practice. Using an expert consensus-based approach, an industry expert steering committee was assembled to develop a framework for video analysis research in horse racing. The aim of the framework is to encourage and facilitate the use of video analysis in the sport and to ensure consistency and quality of future application. To achieve consensus, a systematic review and modified Delphi method study design was used. Responses of the steering committee to two open-ended questions regarding the risk factors of falls and injury were collated and combined with findings from a literature search strategy. Appropriate descriptors and definitions were then formulated that defined and described key features of a jockey fall in horse racing and grouped into six discrete phases of an inciting event. Each member of the steering committee then examined the framework of proposed descriptors and definitions and rated their level of agreement on the 5-point Likert scale. A consensus was achieved on a total of 73 horse racing-specific descriptors and 268 associated definitions. The framework outlined in this study provides a valuable starting point for further research and practice within this area, while the recommendations and implications documented aim to facilitate the practical application of video analysis in horse racing.
- Sporting injuries
- Horse racing
- Risk factor
This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Horse racing is a competitive, high-risk sport1–3 with a high incidence of falls and injuries reported.4–10 The most recent analysis of fall and injury rates in European horse racing was conducted by O’Connor et al9 who outlined that in Irish racing, a professional jump jockey will suffer 1 fall in every 20 rides, with 20% of falls resulting in injury. In comparison, professional flat jockeys fall less frequently with 1 in every 250 rides resulting in a fall but with 35% of falls resulting in injury.
So far, injuries in horse racing have been mainly investigated from an epidemiological perspective, and while this has served to quantify the extent of injury burden faced by the sport, further research is required to understand the aetiology and mechanisms of jockey injuries. Commonly used injury prevention models11 12 have highlighted a need for a multidisciplinary approach to aetiological research in order to fully understand the interaction between risk factors, exposure and injury.
Studies conducted in other sports, including rugby13 14 and football,15 16 have demonstrated a reduction in injury rates following the implementation of injury prevention strategies. Fundamental to the development and design of such strategies is a thorough understanding of the inciting event. The use of video analysis is now considered commonplace in many sporting disciplines, particularly those involving collisions and a high risk of injury.17–20 Video analysis affords the systematic assessment of complex and dynamic sporting scenarios and is particularly useful for understanding the competitive situation, athlete behaviours and movement patterns during an inciting event.21–23 Such detail is often omitted from injury surveillance records. In the context of horse racing, understanding the characteristics of both injurious and non-injurious falls can directly inform the development of injury prevention strategies such as jockey education, falls technique training and safety wear development.
Despite the availability of video footage of all professional horse races, only two studies have sought to examine the kinematics involved during impact events and jockey falls.24 25 Perhaps such paucity of studies might be due to a lack of guidance on the use of video analysis within the horse racing setting. The absence of a sport-specific framework with clear descriptors and definitions may also hinder the application of methodologically robust video analysis and further impede the development of any injury prevention strategies that are informed by video analysis. Clearly defined descriptors and definitions improve the reliability of video analysis in a sporting context by reducing bias and subjectivity.17 21 Furthermore, the need for consistency and standardisation of video analysis has been previously acknowledged within other sporting disciplines and has been achieved through the development of a sport-specific analysis framework.21 26 Therefore, the aim of this study is to develop and achieve consensus on a framework of horse racing-specific descriptors and definitions and outline the implications, recommendations and challenges of video analysis in horse racing.
To develop the framework of descriptors and definitions and reach a consensus, a two-step process was applied as previously described by Hendricks et al21 with further consideration of the Conducting and REporting DElphi Studies guidelines.27 For the first step, a systematic search of the literature was performed. Specific search terms were used to identify peer-reviewed articles in three electronic databases: PubMed, Scopus and Web of Science. The search terms were ‘horse racing’ in the title, keywords or abstract linked in any way to the following terms: ‘video analysis’, ‘jockey injury risk factors’, ‘jockey fall risk factors’, ‘jockey catastrophic injury’, ‘jockey injury mechanism’, ‘jockey injury incidence’, ‘jockey fall’, ‘jockey video analysis’ or ‘jockey injury’ anywhere in the text with a total of nine searches performed for each database. For example, in Scopus, the full search strategy for the term ‘jockey injury’ was: (TITLE-ABS-KEY (horse AND racing) AND ALL (jockey injury) PUBYEAR<2021 LANGUAGE (English) SRCTYPE (j). If the term ‘jockey’ was omitted from the search criteria, the database yielded a heavy dominance in horse-related articles. The results of all nine database searches were merged and duplicates removed. The time frame for the literature search included any article published up to 1 November 2020. The inclusion criteria were as follows: the article needed to be published in a peer-reviewed journal in English and needed to discuss the risk factors related to jockey injuries and jockey and horse falls in horse racing. The inclusion criteria were applied at the title, abstract and full-text levels. Any article not meeting the inclusion criteria was excluded from further review. The results from all three databases were merged, and duplicates were removed, yielding a total of 87 articles that documented the potential risk factors of falls and jockey injuries. Figure 1 summarises the systematic literature search.
In step 2, the purposive recruitment of 17 steering committee members was conducted from within the British horse racing industry’s stakeholders, whose focus is on the care and well-being of professional jockeys and the industry workforce. This included representation from the jockeys’ union (Professional Jockeys Association), British horse racing’s regulatory body (British Horseracing Authority) and the jockey welfare (Injured Jockeys Fund) and training agencies (British Racing School), along with experienced medical professionals (Consultant Spinal Surgeon, Horse Racing Medical Officers) with extensive expertise in the diagnosis and management of jockey injuries. The steering committee also benefited from the representation of the British Equestrian Trade Association, which set the safety standard for body protectors used by equestrians. Finally, eight current professional jockeys (five men, three women) of varying competitive experience from both flat and jump subdisciplines also joined the steering committee.
The authorship group consisted of academics in biomechanics (DC and PM) and epidemiology (KS) and a practising physiotherapist (DL) experienced in the management and rehabilitation of jockey injuries.
Once the steering committee had been established, the development and consensus of descriptors and definitions were sought using a modified Delphi consensus method.27–29 The iterative process of the Delphi consensus method facilitated the development of the analysis framework while overcoming the geographical30 and logistical restrictions experienced due to the COVID-19 pandemic. Furthermore, this method of engagement also provided all steering committee members with the equal opportunity to contribute regardless of social status, seniority or interpersonal skills.27 31
A total of four Delphi rounds were conducted. For the first Delphi round, steering committee members completed an online questionnaire consisting of two open-ended questions: Q1. What do you think are the risk factors of falls in horse racing? and Q2. What do you think are the risk factors of injuries in horse racing? The authorship group then combined the risk factors proposed by the steering committee with the risk factors identified through the search strategy and drafted appropriate descriptors and definitions of such risk factors, and key features of a horse or jockey fall in horse racing.
The resultant framework of descriptors and definitions was structured into subsections to consider six discrete phases of an inciting event (ie, fall/unseat). These included the following:
Situational descriptors—environmental conditions prior to/under which the inciting event occurred. Including location, surface type, obstacles (where relevant), competitive scenario, jockey, horse and opponent behaviour.
Gross fall descriptors—obvious characteristics of a fall, including the type of inciting event, for example, fall or unseating.
Flight phase descriptors—specific biomechanical characteristics of the flight/fall phase prior to jockey sustaining impact.
Contact/impact occurrence descriptors—sequence and characteristics of impacts sustained during fall/inciting event.
Axial skeleton descriptors—specific characteristics involving the axial skeleton during the fall/inciting event.
Secondary impact/recovery descriptors—jockey behaviour and characteristics of any subsequent impact sustained and the recovery immediately following a fall/inciting event.
For subsequent Delphi rounds (two–four), the steering committee members convened via teleconference and examined the framework of proposed descriptors and definitions, rating their level of agreement for each on a 5-point Likert scale (1: strongly disagree; 2: disagree; 3: neither agree nor disagree; 4: agree; 5: strongly agree).32 An online platform (https://www.onlinesurveys.ac.uk/) was used to capture the agreement ratings and any anonymous comments or suggestions. The mean level of agreement (±95% CI) was calculated by summing the ratings and dividing this by the total number of responses. The consensus threshold was determined a priori to be 4/5 (80%). Descriptors and definitions failing to meet the consensus threshold and any anonymous comments were discussed, and any modifications or additions proposed were considered in each subsequent Delphi round. Once a consensus had been achieved for each of the descriptors and definitions, the framework composition was complete. The level of agreement for each of the six framework subsections is reported in the Results section.
To maximise objectivity and reliability and to minimise any ambiguity that might result in coding inaccuracies,22 26 33 three coders experienced in the analysis of human and equine movement and behaviour convened and reviewed the clarity and utilisation of the operational definitions of each analysis framework descriptor. Video footage of 20 horse racing falls (10 jump/10 flat) chosen at random from the British Horseracing Authority video archive was analysed using the video analysis framework. Coding decisions were discussed in an open forum with minimal changes deemed necessary. A single descriptor was removed from the framework due to the varying visibility of furlong markers and race start positions; therefore, coders were unable to consistently identify the distance covered prior to a fall/inciting event from the video footage alone. Furthermore, a strong preference towards coding the horse and jockey’s position in a race by grouping rather than expressed numerically was held by all analysts. While identifying position in a race is possible numerically, this requires greater time and manipulation of video footage, particularly when runners are dispersed in a race.
Finally, as not all steering committee members were available during each round, all members were offered the opportunity to review the analysis framework in its entirety and provide their definitive approval.
The search strategy yielded a total of 87 articles on risk factors related to jockey injuries and jockey and horse falls in horse racing, of which only one article provided a consensus on racing-specific definitions.34 Two further articles from other sporting disciplines that characterised impact events or athlete behaviour in response to an impact event were incorporated into the analysis framework.35 36 Following a four-round modified Delphi method, a consensus was achieved on a total of 73 horse racing-specific descriptors and 268 associated definitions that delineate the risk factors identified in the literature search strategy and Delphi round one (tables 1–6). The mean level of agreement was 4.5 (3.8–5), 4.7 (4.2–5), 4.7 (4.3–4.9), 4.7 (4.3–5), 4.8 (4.3–5) and 4.8 (4.6–5) for the situational, gross fall, flight phase, contact/impact occurrence, axial skeleton and secondary impact/recovery descriptors, respectively.
The decision to include descriptors that capture falls either prior to or immediately after the race generated much debate within the steering committee, with these failing to meet the consensus threshold. It was deemed that pre-race and post-race inciting events were not relevant for inclusion by 4 of the 15 steering committee members in attendance during round two. However, a limitation cited within a previous epidemiological study is due to the reporting of fall characteristics only for the duration of a race. This accounts for a proportion of the time the jockey is mounted on the horse.5 Pre-race activities accounted for 47% of jockey falls, while post-race activity accounted for 11% of jockey falls in flat racing.37 The mechanisms of such injuries are not fully understood, with the suggestion of jockey fatigue being associated with post-race falls,5 confirmation of this suspicion may indicate the need for improvements in jockey fitness as a fall mitigation strategy. A potential barrier to performing video analysis of the horse and jockey interaction outside of the race scenario may be due to the lack of camera coverage or preserved video footage capturing this duration.
The aim of this study was to develop and achieve consensus on a framework of sports-specific descriptors and definitions to facilitate the systematic use of video analysis in horse racing. The current study identified a total of 73 horse racing-specific descriptors and 268 associated definitions, which were organised into subsections to consider six discrete phases of an inciting event.
Injury prevention through jockey education and training
Descriptors that capture the action of the horse be this behavioural, for example, refusal or involutory, for example, spontaneous injury, were included within the analysis framework. Hitchens et al38 identified that jockeys were 171 times more likely to be injured if their horse sustained a catastrophic injury which resulted in a fall. A relationship between whip use and horse falls has also been identified39; horses that were being whipped and progressing through the race were at greater risk of falling compared with horses that had no whip encouragement. An increase in whip use may be due to equine fatigue or poor performance due to discomfort associated with an impending injury.39–41 Professional jockeys are therefore required to recognise signs of equine fatigue or distress and, in doing so, ensure the elective removal from a race, further mitigating the risk of a fall41 and subsequent injury. Video analysis capturing such events may prove a useful tool in jockey education, and descriptors to capture these risk factors are included within the framework.
Examining the behaviours of jockeys and characteristics associated with injurious and non-injurious falls will aid the identification of high-risk scenarios to be avoided, for example, maintaining contact with the reins or stirrup irons during a fall. Conversely, protective actions may also be identified, such as reducing the bodily outline upon landing, which may decrease the risk of sustaining any subsequent impact from kicks while on the ground. Quantifying these behaviours and actions will inform the content and delivery of fall training programmes. Such training exists but is based purely on anecdotal observations.
The infrastructure to facilitate the real-time analysis of video footage already exists as stewards and judges observe for any rule infringements such as improper whip use and accidental interference.42 This real-time analysis could also be extended to aid the detection and management of injuries sustained through falls. The use of video analysis has been successfully implemented in professional rugby unions to assist in the identification of sport-related concussions.18 35 Despite the under-reporting of concussive events in horse racing injury surveillance records,43 concussion is the most frequent head injury endured by jockeys (jump 47.5; flat 10.8 per 1000 race meetings).44 The introduction of mandatory medical assessment has sought to deter under-reporting.43 However, the addition of video analysis for the identification of signs of concussion could strengthen existing assessment protocols further. Furthermore, the inclusion of the descriptors and definitions outlined by Davis et al35 in the international consensus definitions of video signs in professional sport may facilitate the identification of visible sports-related concussions in horse racing.
Integration of video analysis with additional data sources
In the absence of microtechnology worn by the jockey, such as global positioning system and accelerometry, the speed of an inciting event and the impact intensity sustained can be subjectively estimated using the framework descriptors outlined. Such approaches are used effectively in the video analysis of rugby union and league tackles.19 21 36 However, the accuracy and reliability of these subjective measures within horse racing are yet to be determined. Future integration of video analysis with objective data sources that quantify variables such as velocity and impact force will improve our understanding of exposure and the aetiology of horse racing injuries.
Where possible, descriptors and definitions outlined in the European consensus on epidemiological studies of injuries in the thoroughbred horse racing industry34 have been incorporated into the framework. This is intended to strengthen the future convergence of video analysis and injury surveillance data and facilitate consistency in the reporting of injuries. Certain detail regarding intrinsic (eg, injury history, bone health) and extrinsic (eg, going) risk factors which may predispose an individual, potentially making them susceptible to injury,45–47 is not attainable from video analysis alone.21 Equally, injury surveillance data alone do not provide enough detail regarding the inciting event to design and develop injury prevention strategies, especially if jockey behaviour is the target of an intervention.21 46 Therefore, video analysis should be integrated with injury surveillance data to facilitate a multifactorial approach to injury prevention in horse racing.
Ultimately, the quality of available video footage may have a direct impact on the extent of analysis possible. Horse racing video footage is typically captured by cameras mounted on moving vehicles that pursue the field of runners as they progress through the race. The varying racecourse topographies can present a challenge in accessing a suitable camera angle and may result in vital moments being obscured from view. Furthermore, uncalibrated video footage will limit the possibility for a more quantitative biomechanical analysis of injury mechanisms and is, therefore, largely dependent on the categorisation and subjective assessment of observed actions. Model-based image-matching techniques have been used to extract human motion from uncalibrated video footage. However, its accuracy and validity are dependent on the availability of multiple views and camera angles.48 The descriptors and definitions within this framework set to identify and describe the gross movement patterns involved in jockey falls, while an accurate analysis of the external and internal biomechanical loads experienced during a fall is only possible during controlled experimental conditions (eg, laboratory studies of staged falls) due to the safety implications and the practicalities of such investigation in horse racing.
The acquisition of racecourse maps may provide orientation and awareness of environmental characteristics (eg, position of obstacles, severity of bend) of an inciting event, thus providing context to the video footage being analysed. Furthermore, racecourse maps may be electronically integrated with specialist video analysis software which will allow for the location of an inciting event to be plotted directly onto the map allowing for improved data visualisation.
Although not essential, specialist video analysis software may facilitate and expedite the coding process, particularly when the number of variables being considered is vast. Certain software programs feature utilities to capture, tag, compare and annotate video footage. Figure 2 shows an example of software coding window (Nacsport, Spain) which allows the video to be analysed and the appropriate descriptor and definition to be simultaneously selected by clicking on the relevant button, thus, improving coding efficiency and allowing for subsequent quantitative analysis of the recorded actions. However, the use of specialist software programs carries an expense and is dependent on technical knowledge and capability. As a minimum requirement, the software used to navigate and view video footage should allow for frame-by-frame analysis and slow-motion playback speeds.
In video analysis, reliability is a measure of consistency between ratings made by either the same rater (intrarater reliability) or between two different raters (inter-rater reliability).49 The method used to examine reliability will be dependent on the type of variable (continuous vs categorical) being considered, for example, number of whip strikes versus location of fall.33 The use of video analysis to observe postural alignment and quantify joint angles is challenging, with the reliability differing between descriptors.26 The operational definitions outlined within the horse racing video analysis framework look to improve coding accuracy and consistency.
Reliability testing during a video analysis project should be considered a process.21 Repeating reliability measures at the beginning and at later phases of data collection will identify any analysis issues that need to be addressed, therefore facilitating acceptable intrarater and inter-rater reliability, and improving the quality of any project output.21 22 26 33
The availability of quality video footage within amateur racing is currently somewhat lacking. Therefore, only footage featuring professional horse racing was used in the development of the video analysis framework. Consequentially, the application of the framework for the analysis of amateur racing will require further validation. Similarly, the terminology featured within the framework is specific to that of British horse racing. The application of the framework by international racing jurisdictions will need to be reviewed to accommodate the potential differences in country-specific horse racing terminology and rules.
A further limitation to be acknowledged is that not all members of the steering committee could participate in each of the four Delphi rounds. A lack of individual availability prevented full attendance; it was, however, deemed imperative that adequate representation of the professional jockeys was present in order to proceed. To further mitigate the impact of steering committee unavailability, all members were offered the opportunity to consider the final framework in its entirety and provide their approval.
The aim of this study was to develop and achieve a consensus on a framework of sports-specific descriptors and definitions to facilitate the systematic use of video analysis in horse racing. The framework outlined provides a valuable starting point for further research and practice. Which variables to use will be dependent on the objectives of future study. While falls in horse racing are considered somewhat inevitable and carry an inherent occupational risk, understanding the complex dynamics at play is imperative for jockey and equine welfare.
Patient consent for publication
This study was granted ethical approval by the University of Bath Research Ethics Approval Committee for Health (REACH) (EP 19/20 066). Participants gave informed consent to participate in the study before taking part.
The authors would like to thank all members of the steering committee for their continued support and contributions to this study. The authors would also like to thank Lauren Dallison for participating in the operational review of the video analysis framework.
Collaborators Horse Racing Video Analysis Consensus Steering Group: Dr Jerry Hill—Chief Medical Advisor, British Horseracing Authority, London, England; Mr Richard Perham—Senior Jockey Coach, British Racing School, Newmarket, England; Mrs Claire Williams—Executive Director, British Equestrian Trade Association, Wetherby, England; Ms Lisa Hancock—Chief Executive, The Injured Jockeys Fund, Newmarket, England; Dr Anna Louise McKinnon—Head of Clinical Services, The Injured Jockeys Fund, Newmarket, England; Mr Paul Struthers—Former Chief Executive, The Professional Jockeys Association, Newbury, England; Mr Jason Harvey—Consultant Spinal Surgeon, Fortius Clinic, London, England; Professor Michael Gilchrist—Medical Engineering, University College Dublin, Dublin, Ireland; Ms Kerry Kuznik—Medical Assistant and Jockey Athlete Research Coordinator, British Horseracing Authority, London, England; Mr Kevin Jones—Professional Jump Jockey, England; Miss Page Fuller—Professional Jump Jockey, England; Mr Jamie Moore—Professional Jump Jockey, England; Miss Bryony Frost—Professional Jump Jockey, England; Ms Hollie Doyle—Professional Flat Jockey, England; Mr Jim Crowley—Professional Flat Jockey, England; Mr David Egan—Professional Flat Jockey, England; Mr Tom Marquand—Professional Flat Jockey, England.
Contributors The research was conceptualised by DL and DC. DL conducted the systematic literature search. JH assisted with the identification and recruitment of industry experts for the horse racing video analysis steering committee.All authors attended the steering committee meetings and contributed to the development of the video analysis framework. DL drafted the manuscript and all authors contributed to and approved the final draft.
Funding The author(s) disclosed receipt of the following financial support for research, authorship, and/or publication of this article: this study was conducted as part of DL's PhD Studentship funded by the University of Bath and the Racing Foundation.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.